There was a stretch of years when the Inter-Services Intelligence Directorate ran the single largest covert war on the planet, and almost nobody outside a few rooms in Rawalpindi and Langley understood how completely one Pakistani institution had come to dominate it. Crates of Egyptian and Chinese rifles moved through the port of Karachi, up the road to depots near the Afghan frontier, and into the hands of fighters whose names the directorate kept in its own files and shared with no foreign service. American money, Saudi money, and a steady stream of weapons all passed through a single clearing house, and that clearing house was a spy agency answerable to the Pakistan Army rather than to any elected government. For roughly a decade, Pakistan’s premier intelligence service held more operational leverage over the future of a neighboring country than any other actor in the region, including the superpowers funding it.

That peak is the right place to begin any honest account of the agency, because everything the directorate has done since has been measured, consciously or not, against the memory of those years. The institution that allegedly cannot today protect a designated militant living openly in Bahawalpur is the same institution that once decided which Afghan commander received anti-aircraft missiles and which did not. The distance between those two states of being is the real story. It is a story of institutional decline, and the decline matters because it explains why the killings now happening on Pakistani soil are possible at all.

This account traces the spy service across five distinct phases, and the phases do not describe a rising arc. They describe the opposite. The agency that emerged from the 1980s convinced of its own indispensability spent the following four decades slowly discovering that the instruments it had built, the militant organizations it had armed and sheltered and pointed at India and Afghanistan, were becoming first unmanageable, then a liability, and finally a vulnerability that a rival service could exploit at will. Pakistan’s external intelligence apparatus built a sanctuary for violence, and that sanctuary is now the thing being dismantled, target by target, in a campaign the directorate can neither prevent nor publicly acknowledge.

To understand how a service falls that far, it helps to understand how it rose, and how strange its rise actually was. The agency was not always powerful. It was, for its first quarter century, a fairly ordinary military bureau. What changed it was a war next door, a flood of foreign cash, and a general who understood that whoever controlled the pipeline controlled the war.

The World That Built the Agency

Pakistan’s intelligence directorate was created in 1948, in the immediate aftermath of the first war over Kashmir, and its origins say a great deal about what it would later become. The new state had gone to war within months of its birth, and the conflict had exposed a humiliating gap: the army’s various branches were not sharing what they knew. A British officer serving in the Pakistan Army, Major General Robert Cawthome, is generally credited with proposing a dedicated body to coordinate intelligence across the services. The name itself, Inter-Services Intelligence, described a coordinating function rather than an operational ambition. For its founders it was meant to be plumbing, a way of moving information between the army, navy, and air force so that the failures of 1948 would not be repeated.

For most of the next twenty-five years the directorate behaved like the modest bureau it had been designed to be. It collected information on India, monitored the eastern wing of the country in what was then East Pakistan, and kept files on domestic politicians, a habit that would metastasize badly in later decades. The agency had no significant covert action arm, no proxy forces, and no claim on the levers of foreign policy. The men who ran it were army officers on rotation, and the posting was not yet the career-defining prize it would become. When the country broke apart in 1971 and the eastern wing became Bangladesh, the directorate shared in the general humiliation of Pakistan’s security establishment. It had not seen the disaster coming, and it had been unable to counter the Indian intelligence effort that helped midwife the new nation, an effort traced in detail in the complete history of India’s external service.

Two earlier shocks had already exposed how thin the bureau’s reach actually was. During the 1965 war with India, Pakistani planners had launched an operation to infiltrate fighters into Indian-administered Kashmir on the assumption that the local population would rise in support. The assumption proved badly wrong, the infiltrators were quickly isolated, and the broader war ended in a stalemate that Pakistan’s military narrative struggled to recast as victory. The episode is worth remembering because it contained, in miniature, the entire flawed theory the agency would later build into doctrine: the belief that infiltrated fighters plus local grievance would reliably produce a manageable uprising. The theory failed in 1965, and it would fail again in larger and bloodier forms, but the institution never fully absorbed the lesson. Then came the catastrophe of 1971, when the eastern wing was lost despite everything Pakistan’s security apparatus could do. A service that had been designed to coordinate information had twice watched the country stumble into disaster, and the men who would later remake it carried a powerful sense that the existing way of doing things had failed.

It is also worth being precise about how unremarkable the bureau was in these years, because the contrast with what came later is the whole point. Through the 1950s and 1960s the directorate had no meaningful budget for covert action, no stable of foreign proxies, and no independent voice in the making of policy. Its officers tracked Indian troop movements, watched Pakistan’s own politicians a little too closely, and produced assessments that the senior command read or ignored as it pleased. Nothing about the institution at this stage suggested that it would one day function as the hidden hand behind a transnational jihadist movement. The capacity for that role was not latent and waiting to be discovered. It had to be built, deliberately, by particular men responding to a particular war, and the building began in earnest only at the end of the 1970s.

The transformation began under Zulfikar Ali Bhutto and then accelerated dramatically under the man who overthrew and hanged him. Bhutto, as prime minister in the early 1970s, expanded the directorate’s domestic political role, using it to monitor and harass opponents, and that politicization left a permanent stain on the institution. But it was General Zia-ul-Haq, who seized power in 1977, who turned a coordinating bureau into something far larger. Zia was a religiously conservative officer who saw in political Islam both a tool of statecraft and a genuine conviction, and under his rule the army began to think of jihad not as a metaphor but as an instrument. The agency would become the hand that wielded that instrument.

Zia’s broader project gave the directorate the ideological raw material it would need. The general’s program of state-led Islamization reshaped law, education, and public life, and one of its most consequential elements was a rapid expansion of religious seminaries, many of them teaching a hardline interpretation that valorized armed struggle. These institutions multiplied along the Afghan frontier and across Punjab, often funded by a mixture of state encouragement and Gulf money, and they became, over time, a renewable supply of recruits. A spy service planning to wage proxy war needs three things above all: fighters, an ideology that motivates them, and an infrastructure that produces both on a continuous basis. Zia’s Pakistan supplied all three, and the directorate was positioned to draw on them. Within the agency a dedicated component, often described in later accounts as the Afghan bureau or Afghan cell, was stood up to run the coming war, and it grew with extraordinary speed from a small desk into the operational heart of the institution.

The man who presided over the directorate during the decisive years was General Akhtar Abdur Rahman, who led the service for most of the 1980s and who is fairly described as the chief architect of its wartime machine. Akhtar built the relationships, designed the compartmentalized structure that kept the Afghan program shielded from outside view, and insisted on the rule that made the agency indispensable, the rule that all support would pass through Pakistani hands. He was close to Zia, trusted by him, and given a latitude that few intelligence chiefs anywhere have enjoyed. Under Akhtar the directorate stopped being a bureau that served the army’s needs and became a power in its own right, and the officers who rose under him absorbed his conviction that the management of holy war was now the institution’s defining vocation. When Akhtar died alongside Zia in the unexplained 1988 plane crash that killed much of Pakistan’s senior leadership, he left behind an agency utterly transformed from the modest body he had joined.

What made the transformation possible was geography and timing. In December 1979 the Soviet Union sent its army into Afghanistan to prop up a faltering communist government in Kabul, and overnight Pakistan became the indispensable frontline state in a global confrontation. The United States needed a way to bleed the Soviets without committing its own forces. Saudi Arabia wanted to fund a defense of the faith. Pakistan sat on the only practical supply route to the Afghan resistance, and Zia’s government understood exactly what that meant. Islamabad would accept the role of conduit, but on one condition that shaped everything afterward: no foreign service would deal directly with the Afghan fighters. Every dollar, every rifle, every training decision would route through the directorate. The Americans, eager for the war and wary of the optics of running it themselves, agreed.

That single arrangement is the hinge of the entire story. By insisting on being the sole intermediary, Pakistan’s spy service did not merely gain access to a river of resources. It gained the power to decide which Afghans received those resources and which did not, and therefore the power to shape the Afghan resistance into the form Islamabad preferred. The agency favored the most ideologically committed Islamist commanders, men like Gulbuddin Hekmatyar, partly out of conviction and partly out of cold calculation: pious fighters loyal to a transnational cause would, the planners believed, be more reliable instruments and less likely to nurture Afghan or Pashtun nationalism of the kind that might one day press claims against Pakistan itself. The secular and nationalist Afghan factions were starved of support. The decision would echo for forty years.

The scale of what flowed through the pipeline is difficult to overstate. Over the course of the 1980s, the covert program known on the American side as Operation Cyclone channeled billions of dollars in assistance, matched roughly dollar for dollar by Saudi Arabia, into the Afghan war. The directorate built training camps, ran courses in sabotage and ambush and the use of increasingly sophisticated weaponry, and coordinated the movement of tens of thousands of fighters. It also became the gatekeeper for the foreign volunteers, the Arab and other non-Afghan militants who arrived to join the holy war, and in doing so it helped knit together networks of transnational jihadists who would later turn their attention to targets far from Kabul. The agency did not invent these men, but it provided the infrastructure, the safe passage, and the operational space in which their networks matured.

The physical mechanics of the pipeline deserve a closer look, because they explain how a single institution accumulated so much leverage. Weapons arrived by sea and air, were warehoused at depots inside Pakistan, the most notorious being the Ojhri camp near Rawalpindi, and were then distributed to the Afghan factions according to decisions made by the agency’s officers. Foreign intelligence partners supplied the money and much of the hardware, but they were deliberately kept away from the Afghan commanders themselves. American case officers, by the design Islamabad had insisted upon, generally did not meet the fighters, did not run them, and did not control which faction received what. That structural fact gave the directorate a double advantage. It could shape the Afghan resistance toward the Islamist commanders it preferred, and it could also skim, divert, and redirect resources with limited outside oversight, a freedom that bred both corruption and strategic autonomy. The 1988 explosion that destroyed the Ojhri depot, sending munitions raining across Rawalpindi and Islamabad, offered the wider public a brief and alarming glimpse of just how much lethal material the agency was managing.

One weapon system became emblematic of the agency’s wartime peak. When the United States agreed in the mid-1980s to supply shoulder-fired anti-aircraft missiles to the Afghan resistance, the decision about which commanders received these prized weapons ran, like everything else, through the directorate. The missiles degraded Soviet air superiority and are widely credited with raising the cost of the occupation, and their distribution underscored the central reality of the war: the most consequential battlefield choices were being made not in Washington or Moscow but by Pakistani intelligence officers allocating someone else’s arsenal. Alongside the hardware came the human pipeline. Recruitment networks reaching into the Arab world brought thousands of foreign volunteers through Pakistan and into the camps, and the agency tolerated and at times facilitated their passage. Among the organizers of that foreign influx were men whose later projects would horrify the world. The directorate did not foresee what it was helping to assemble, but the seedbed of transnational jihad was watered, in significant part, by the war the agency ran.

This was the world that built the modern directorate: a decade of war in which a previously modest bureau discovered that it could function as the central nervous system of an international campaign. The officers who came of age in those years internalized a particular lesson. They had, as they understood it, defeated a superpower. The Soviet army began its withdrawal in 1988 and completed it in February 1989, and to the men in Rawalpindi the lesson was not that they had been useful auxiliaries in someone else’s war. The lesson was that covert proxy warfare, run by their service, could break empires. That conviction, equal parts accurate and dangerously inflated, became the founding myth of the institution’s golden age. It would take decades, and a long sequence of failures, before the myth was finally exhausted.

The Rise

The years between the Soviet withdrawal and the turn of the century were the directorate’s high-water mark, the period when its officers genuinely believed they had discovered a repeatable formula for projecting power on the cheap. The Afghan war had proven, to their satisfaction, that a determined intelligence service could organize and direct insurgent proxies to strategic effect. What followed was an attempt to apply that formula again, this time against the rival that the Pakistani security establishment regarded as the true existential threat. The proving ground would be Kashmir.

The directorate did not create the Kashmiri grievance, and it is important to be precise about this. By the late 1980s the disputed territory had its own deep wellsprings of discontent, including a state election in 1987 widely seen as rigged, which discredited electoral politics for a generation of young Kashmiris and pushed many toward the idea of armed resistance. An indigenous insurgency was already stirring. What the agency did was recognize an opportunity and move to capture it. Pakistani planners offered training, weapons, and sanctuary, and in doing so they steadily shifted the character of the uprising. The early Kashmiri militant outfit with the broadest local support, the Jammu Kashmir Liberation Front, favored an independent Kashmir rather than accession to Pakistan, which made it unattractive to Islamabad. So the directorate’s support flowed instead toward groups that endorsed merger with Pakistan and framed the struggle in religious terms, above all Hizbul Mujahideen, and the indigenous and relatively secular face of the insurgency was gradually eclipsed.

A grim piece of timing made the Kashmir escalation possible. The Afghan war was winding down just as the Kashmir valley began to burn, which meant the agency had, ready to hand, an entire wartime apparatus that suddenly lacked a war. Training camps, instructors, weapons stockpiles, recruitment pipelines, and a cadre of battle-hardened fighters were all available for redeployment, and the directorate redeployed them. Facilities that had prepared men to fight Soviet armor were repurposed to prepare men to cross the Line of Control. Hardened veterans of the Afghan campaign, some Pakistani and some foreign, were funneled toward the new front, bringing with them a familiarity with explosives and ambush tactics that the early Kashmiri militants had lacked. Within a few years the character of the violence in the valley had shifted from a largely homegrown uprising toward a campaign with a heavy infusion of external organization, training, and personnel. The agency was not improvising. It was applying, with deliberate intent, the machine the Afghan war had built, and the speed of the pivot showed how thoroughly proxy warfare had become the institution’s reflexive answer to a strategic problem.

Then came the most consequential escalation. As the 1990s wore on, the Pakistani approach shifted from supporting Kashmiri fighters toward inserting Pakistani ones. Outfits raised, staffed, and ideologically shaped in Pakistani Punjab became the cutting edge of the violence in the valley. Lashkar-e-Taiba grew out of a missionary and militant movement headquartered at a sprawling campus near Lahore and developed into perhaps the most disciplined and lethal of these instruments, an organization whose origins, structure, and reach are examined in the complete profile of the group. The directorate’s relationship to Lashkar was unusually close, close enough that the group came to be regarded by many analysts as the most reliable of all the agency’s proxies, the one least likely to slip its leash. Other outfits passed through cycles of merger, banning, and rebranding, among them the organizations that would eventually consolidate into Jaish-e-Mohammed, the group whose founding and arc are detailed in the dedicated guide to its terror empire.

One episode from the closing months of the 1990s shows how the proxy ecosystem worked, and how little daylight separated the official Pakistani position from the operational reality. Through the decade a militant outfit operating under names such as Harkat-ul-Ansar and Harkat-ul-Mujahideen had been active in Kashmir with the toleration and support of the security establishment, and several of its figures had been captured by India. In December 1999 an Indian Airlines flight was hijacked and ultimately flown to Kandahar, then under Taliban control, and the crisis ended with India releasing three imprisoned militants in exchange for the passengers. The released men were not marginal figures. One of them, a cleric named Masood Azhar, returned to Pakistan and, within months, founded Jaish-e-Mohammed, an organization that would go on to mount some of the deadliest attacks of the following two decades. The hijacking itself, the Kandahar standoff, and the chain of consequences that followed the release are reconstructed in the complete account of that crisis. For an honest assessment of the directorate, the lesson of the episode is not about any single hijacking. It is that the agency presided over an environment in which a freed militant could establish a major new outfit, in the open, on Pakistani soil, and immediately resume operations. The proxy machine did not merely tolerate such men. It made room for them, supplied them, and counted them as assets.

The directorate’s method in this period had a recognizable shape. It cultivated proxies, but it tried to retain enough control over their leadership, funding, and logistics that they could be turned on and off as the strategic situation required. The agency wanted deniability above all. The fighters crossing the Line of Control were to be described, officially, as indigenous freedom fighters or as volunteers beyond the state’s control, and the spy service worked hard to maintain the fiction. Pakistan’s broader sheltering apparatus, the residences and camps and front organizations that allowed wanted men to live and work in the open, grew up alongside the proxy program and is mapped in the analysis of the country’s safe haven network. For a time the formula appeared to work. The insurgency in Kashmir tied down enormous numbers of Indian security forces, imposed a continuous cost on the rival state, and did so while Islamabad maintained, however implausibly, that it was merely a sympathetic bystander.

It is worth pausing on why the directorate believed this was sustainable, because the belief reveals the institutional mindset at its most confident. The planners had convinced themselves that proxy warfare occupied a uniquely advantageous space. It was cheaper than conventional war. It was, they thought, controllable, because the proxies depended on Pakistani sanctuary and supply. And it was deniable, which meant it could be conducted below the threshold that might trigger a full conventional response from a larger neighbor. The agency had, in its own estimation, found a way to wage permanent low-intensity war against a stronger adversary at acceptable cost and acceptable risk. The doctrine had a name in the strategic literature, the pursuit of strategic depth and the bleeding of the adversary by a thousand cuts, and for most of the 1990s its architects considered it a success.

The same period saw the directorate extend its reach in Afghanistan, where the formula seemed to deliver an even more spectacular return. The civil war that followed the Soviet withdrawal had left Afghanistan fractured among competing warlords, and the directorate’s old client Hekmatyar had failed to consolidate power. So Pakistani planners shifted their backing to a new movement that emerged from the religious seminaries of the Pashtun belt, a movement of students that promised order, piety, and, crucially, a friendly government in Kabul. By 1996 that movement, the Taliban, had captured the Afghan capital, and Pakistan was one of only three countries on earth to recognize the resulting government. For the directorate this looked like vindication on a grand scale. It had, it believed, installed a client regime in a neighboring country, securing the strategic depth its doctrine demanded.

Yet even the Afghan triumph carried warnings the agency chose not to read. One of those warnings had come early, in 1989, immediately after the Soviet withdrawal. Hamid Gul, who had led the directorate in the closing phase of the anti-Soviet war and who held an expansive vision of pan-Islamic strategic depth, pressed the Afghan resistance to seize the city of Jalalabad and establish a provisional government there. The assault was a costly failure. The resistance, formidable as a guerrilla force, proved unable to take and hold a defended city against the still-functioning Afghan communist army, and the defeat exposed a recurring blind spot. The directorate’s planners were repeatedly tempted to believe that proxies trained for insurgency could be scaled up into instruments of decisive conventional outcomes, and reality kept refuting the belief. A second warning lay in the company the Taliban kept. The new government in Kabul played host to Al Qaeda, whose leadership had relocated to Afghanistan, and the foreign jihadist networks that the Afghan war had incubated now had a sanctuary of their own. The directorate’s strategic depth, in other words, came bundled with a transnational terrorist organization whose ambitions had nothing to do with Pakistani interests and would soon bring catastrophe to Pakistan’s doorstep. The agency saw a client regime. It did not adequately register that it had also helped create the conditions for a global crisis.

The confidence of these years had a hard institutional edge. The covert action arm of the directorate, the wing that handled the militant proxies and the relationships with the jihadist outfits, became the most prestigious and the most autonomous part of the service. Outside scholars studying the agency, including Steve Coll in his detailed accounts of the period, have described this wing as something close to a state within the intelligence apparatus, shielded from outside scrutiny and operating with extraordinary latitude. Within Pakistan the directorate’s prestige and reach expanded accordingly. Its officers shaped not only the relationship with militant groups but, increasingly, the country’s foreign policy toward India, Afghanistan, and the wider region. The army had always been the dominant institution in Pakistan, a dominance traced in the scholarship on the military’s grip on the state, but within the army the intelligence directorate had become the indispensable instrument of regional strategy.

This was the rise. A modest coordinating bureau had become, within roughly fifteen years, the architect of Pakistan’s proxy wars on two fronts and a power center that elected governments could not control and frequently did not even fully understand. Everything about the institution’s later trajectory flows from the overconfidence of this moment, the conviction that the proxy formula was a permanent strategic asset rather than a debt that would eventually come due. The bill began to arrive almost immediately, and the first installments came from the very instruments the agency was most proud of.

The Decisions That Defined the Directorate

An institution reveals itself most clearly in its decisions, and the directorate’s history turns on a handful of choices that looked, at the moment they were made, like shrewd applications of a proven doctrine, and that look in retrospect like the precise mechanisms of its decline. Each was an attempt to extract more advantage from the proxy formula. Each instead deepened the agency’s dependence on instruments it could not ultimately govern.

The first defining decision was the choice, made and remade through the 1990s, to escalate from indigenous Kashmiri proxies to Pakistani ones, and then to allow the most capable of those Pakistani outfits to grow into large, semi-autonomous institutions with their own finances, recruitment pipelines, and ideological agendas. The logic was operational: Punjabi fighters drawn from organizations the directorate had helped build were better trained and more reliable than the looser Kashmiri formations. But the consequence was structural. By nurturing groups like Lashkar-e-Taiba and the outfits that became Jaish-e-Mohammed into mature organizations, the agency created clients that no longer depended on it for their basic existence. These groups developed independent fundraising through charity fronts, independent recruitment through their own seminaries and social networks, and independent ideological authority that did not require Pakistani validation. A proxy that can fund and staff and justify itself is no longer fully a proxy. It is a partner at best and a rival at worst, and the directorate had manufactured several of them. The intimate institutional relationship between the spy service and these outfits, the way the agency built and armed and shielded them, is documented in the analysis of the intelligence and terror nexus.

The second defining decision came at the end of the 1990s, when the directorate and the army leadership chose to test the proxy doctrine against the conventional balance directly. In 1999, Pakistani forces, including paramilitary and irregular elements, infiltrated across the Line of Control into the heights above Kargil, an operation conceived in significant part within the military intelligence and directorate orbit and kept secret even from much of Pakistan’s own civilian government. The planners appear to have believed that the incursion could present India with a fait accompli below the threshold of full war, a kind of conventionalized version of the proxy logic. The calculation failed. India responded with a substantial military campaign to retake the heights, the international community sided overwhelmingly with the position that the Line of Control must be respected, and Pakistan was forced into a humiliating withdrawal. The Kargil episode demonstrated a hard limit of the doctrine: the deniable, below-threshold approach did not scale up to seizing and holding territory, and pushing it that far produced strategic isolation rather than advantage.

Most fateful of all was a third decision, one that was not a single act but a posture adopted after September 2001, the posture that scholars and former officials have come to call the double game. When the United States launched its campaign against Al Qaeda and the Taliban regime in Afghanistan, Pakistan’s military ruler at the time, General Pervez Musharraf, made a public choice to align with Washington. Pakistan would become, officially, a partner in the campaign against terrorism, and it would receive enormous quantities of aid and military support in that role. But the directorate did not abandon its Afghan clients. Instead, the agency attempted to run both sides of the conflict at once. It cooperated with the Americans against Al Qaeda in some measure, handing over certain operatives, while simultaneously preserving the Afghan Taliban leadership and the allied factions as a strategic reserve, a hedge against the day the Americans would inevitably leave and Afghanistan’s future would again be contested.

The double game is the central decision of the directorate’s middle period, and its costs accumulated slowly and then catastrophically. In the short term it appeared clever. Pakistan collected the benefits of alliance while keeping its proxy assets intact. But the posture required the agency to deceive its most important international partner continuously, over a span of years, and the deception was eventually exposed in the most damaging way imaginable. The provision of sanctuary to the Afghan insurgency’s leadership, the toleration of the Haqqani network as it mounted attacks on foreign forces in Afghanistan from Pakistani soil, the long pattern of warning and protecting rather than capturing, all of this gradually convinced large parts of the American security establishment that Pakistan’s intelligence service was not an unreliable ally but an active adversary wearing the costume of one. Trust, once it began to drain, did not return.

Two attacks bracketed the double game and showed the world what the proxy apparatus could produce. In December 2001, gunmen assaulted the Indian Parliament complex in New Delhi, and the assault was attributed to militants from the very Punjabi outfits the directorate had cultivated. India responded by mobilizing its army along the length of the international border in a confrontation that lasted the better part of a year, holding two nuclear-armed states in an eyeball-to-eyeball standoff while the world watched in alarm. The crisis eased without full war, but it established a pattern that would recur: a spectacular proxy attack, an Indian military response short of war, and a dangerous interval in which miscalculation could have been catastrophic. The directorate had wanted deniable, controllable, below-threshold violence. What it had actually built was a mechanism for repeatedly dragging the subcontinent to the edge of a conflict no one claimed to want.

The second attack was worse. In November 2008, a small team of Lashkar-e-Taiba gunmen arrived by sea and laid siege to Mumbai for some sixty hours, killing well over one hundred and sixty people at hotels, a railway station, a hospital, and a Jewish center. The operation was not the work of an autonomous fringe. Investigation in multiple countries, including the testimony and guilty plea of David Headley, an operative who had conducted detailed reconnaissance of the targets, pointed to extensive planning, training, and direction, and raised persistent and credible questions about the involvement of serving or former officers of the directorate’s covert apparatus. Pakistan’s official response followed the now-familiar template of denial, partial acknowledgment, detentions that did not lead to durable convictions, and prosecutions that stalled for years. For the agency, the strategic cost of Mumbai was immense. The attack made it impossible for serious observers anywhere to accept the fiction that Lashkar was a charity with an unfortunate militant fringe, and it hardened the international consensus that Pakistan’s premier spy service was directing instruments of mass-casualty terrorism. The proxy that had been prized as the most reliable of all had just demonstrated, in front of a global audience, exactly how dangerous the directorate’s whole project had become.

A single event crystallized this judgment in May 2011, when American forces located and killed Osama bin Laden not in some inaccessible tribal redoubt but in a large compound in Abbottabad, a garrison town, within a short distance of Pakistan’s premier military academy. The directorate’s defenders argued incompetence: the agency had simply failed to notice. Its critics argued complicity: an institution that could track a designated militant across a continent could surely notice the most wanted man on earth living for years beside its own officer corps. Whichever explanation one accepted, the episode was devastating for the agency’s standing. Incompetence at that scale was nearly as damning as complicity, and much of the world quietly concluded that the truth lay somewhere in the unforgiving middle.

The year 2011 became, in retrospect, the moment the American partnership effectively collapsed, and a sequence of episodes beyond the Abbottabad raid drove the rupture. Early that year a contracted American operative named Raymond Davis shot and killed two men on a Lahore street, and the resulting standoff over his status and release exposed, in public and humiliating detail, how little trust survived between the two intelligence establishments. Later in the year a cross-border incident at Salala, in which American air power killed Pakistani soldiers, pushed the relationship to open hostility, and Pakistan retaliated by closing the overland supply routes that sustained foreign forces in Afghanistan. Each episode fed a narrative, already strong in Washington, that the directorate was less a flawed partner than a hostile service that happened to be on the payroll. For the agency, the consequences ran deeper than wounded pride. The American relationship had been, for three decades, a primary source of funding, equipment, training, and political cover. As that relationship curdled, the directorate found itself sustaining its proxy commitments with diminishing external support and mounting external suspicion, carrying the costs of the double game while the benefits that had once offset them drained away.

There is a fourth decision that belongs in this account, and it is less a discrete choice than a sustained failure to act. Through the 2000s and into the 2010s, as the militant ecosystem the directorate had cultivated grew larger and more ideologically radical, parts of that ecosystem turned inward against the Pakistani state itself. The Pakistani Taliban movement, formed in the tribal areas, began a campaign of mass-casualty violence inside Pakistan, striking mosques, markets, schools, and security installations. The directorate now faced the consequence its own doctrine had always risked. It had spent decades teaching that jihadist violence was a legitimate instrument of policy, and a portion of the men who absorbed that teaching had concluded that the Pakistani state, with its alliance with Washington and its compromises, was itself a legitimate target. The agency tried to draw a distinction its own history made incoherent: there were good militants, the ones still pointed at India and at Afghanistan, and bad militants, the ones now pointed at Pakistan. It moved hard against the second category while continuing to shelter the first. The distinction was strategically convenient and morally hollow, and it left the directorate fighting a war against one part of a movement it was still nurturing in another part.

December 2014 forced the contradiction into the open in the most agonizing way. Gunmen from the Pakistani Taliban stormed an army-run school in Peshawar and killed roughly one hundred and fifty people, most of them children. The massacre shocked the country into a moment of apparent resolve, and the state responded with a sweeping counter-terrorism framework, the National Action Plan, which promised, among much else, to end all distinctions between militant groups and to dismantle the entire infrastructure of organized violence. On paper it was a renunciation of the proxy doctrine itself. In practice the renunciation was selective. The campaign against the anti-state militants who had carried out the slaughter was real and sustained, and over the following years it substantially degraded the Pakistani Taliban’s capacity inside the country. But the outward-facing proxies, the organizations still pointed at India and still useful to the directorate’s regional strategy, were not dismantled with anything like the same vigor. Banned groups reappeared under fresh names. Prominent figures cycled through brief, reversible detentions. The seminaries and charity fronts kept functioning. The Peshawar atrocity had demonstrated, in the most unbearable terms, the price of treating jihadist violence as a controllable instrument, and the directorate’s response was to fight the half of the movement that threatened Pakistan while preserving the half it still hoped to aim outward. The good-militant and bad-militant distinction survived its own refutation, because the institution could not bring itself to abandon the doctrine even when its own schoolchildren were paying for it.

Taken together, these decisions describe an institution steadily losing the control that had been the entire premise of its doctrine. The agency had set out to wield proxies as precise instruments of policy. By the end of this period it was managing a sprawling, partly autonomous, partly hostile militant landscape that it could neither fully direct nor safely dismantle, while its most important external relationships curdled into suspicion. The proxy formula had not been a permanent asset. It had been a loan, and the directorate was now servicing the debt with interest.

The Institution Behind the Operations

Behind the operations and the decisions sits the harder question, the one a genuine portrait has to attempt: what kind of institution makes these choices, decade after decade, and keeps making them even as the evidence of failure accumulates? The directorate is not a person and has no single mind, but institutions do have character, formed by how they recruit, what they reward, what they fear, and the stories they tell themselves about their own past. The character of Pakistan’s intelligence service explains its trajectory at least as much as any individual decision does.

The first and most important fact about the institution is that it is an arm of the army, not of the elected state. The directorate’s senior leadership is drawn overwhelmingly from the military, its director general is a serving general, and its officers expect to return to military careers. This means the agency’s incentives are the army’s incentives, and the army’s self-conception has long placed it as the true guardian of Pakistan, standing above and if necessary against the country’s messy civilian politics. The scholarship on Pakistan’s civil-military imbalance describes an institution that does not regard itself as a servant of elected governments, and the intelligence directorate inherits that posture in concentrated form. Decisions about proxy warfare, about which militant groups to shelter, about how to manage relations with India and Afghanistan and the United States, have frequently been made within the agency and the army leadership with minimal civilian input, sometimes with active civilian exclusion. The Kargil incursion is the clearest example, planned and launched with much of the elected government reportedly kept in the dark. An institution insulated from political accountability can persist in a failing strategy far longer than one that must answer to voters, because the feedback that would force correction never reaches it.

Leadership mechanics reinforce this military character at every turn. The directorate is headed by a serving lieutenant general, appointed in practice by the army chief, and the appointment is one of the most powerful in the country. Directors general rotate through the post on tours of a few years and then return to the army’s command structure, sometimes rising to the very top. The arrangement means that no head of the agency can afford to define the institution’s interests in opposition to the army’s, because the army is both the source of the appointment and the destination after it. It also means that whatever doctrinal habits one director general entrenches are inherited by the next, who arrives steeped in the same institutional culture and dependent on the same patronage networks. There have been moments when an incoming chief or director general gestured toward reform, toward a recalibration of the relationship with the proxies, but the gestures rarely survived contact with the institution’s entrenched incentives. A handful of directors general have become public figures in their own right, their tenures associated with particular crises or controversies, and a few have wielded influence rivaling that of the army chief himself. But the office, however powerful its occupant, has never been a platform from which the proxy doctrine could be safely dismantled, because the office is embedded in the very structure that the doctrine serves.

The second defining trait is the founding myth already described, the conviction born in the 1980s that the agency had defeated a superpower through covert proxy warfare. Founding myths are powerful precisely because they encode a lesson that feels proven. The lesson the directorate drew was that proxies work, that deniable violence is a sustainable instrument, and that the service’s distinctive competence is the management of jihad. Every subsequent generation of officers entered an institution that told this story about itself. When a strategy is bound up with institutional identity in this way, abandoning it is not merely a policy adjustment. It is a kind of self-renunciation, an admission that the institution’s proudest achievement was the seed of its later troubles. Bureaucracies are extraordinarily reluctant to make admissions of that kind, and the directorate’s persistence with the proxy doctrine long after its costs were obvious is, in part, the persistence of an institution unwilling to disown its own legend.

A third trait is a deep and genuine threat perception regarding India. It would be a mistake to treat the directorate’s behavior as cynicism alone. The Pakistani security establishment’s conviction that India poses an existential danger, that it seeks to undo Pakistan, that it cannot be matched conventionally and must therefore be countered asymmetrically, is sincerely held and institutionally reinforced. The agency’s officers have come to see the long contest with India’s external service, the rivalry examined in the analysis of the two agencies set against each other, as the central drama of their professional lives. Proxy warfare, in this worldview, is not adventurism. It is the rational response of a weaker state to a stronger neighbor, the only available means of imposing cost and maintaining strategic relevance. This sincerity matters because it explains durability. A purely opportunistic institution might have abandoned a failing tactic. An institution that believes the tactic is a matter of national survival will cling to it even as it bleeds.

The fourth trait is the structural autonomy of the covert action wing, the part of the directorate that handles the militant relationships. This wing has long operated with an unusual degree of independence, shielded from scrutiny even within the Pakistani system, and its long-serving officers and contacts have developed relationships with militant leaders that span decades. Personal ties of that depth are not easily severed by a policy memo. The men who built and maintained the proxy networks have careers, reputations, and worldviews invested in those networks. Even when the directorate’s senior leadership has concluded, at various moments, that particular relationships have become liabilities, the institutional machinery for actually dismantling them has been weak, slow, and resisted from within. This is part of why the agency’s promises to act against particular groups, made repeatedly to international partners, have so consistently produced cosmetic results: detentions that do not stick, bans that produce renamed organizations, prosecutions that stall.

A fifth trait, less often connected to the proxy story but inseparable from it, is the directorate’s deep entanglement in domestic politics. From Bhutto’s era onward the agency maintained an internal political function, a capacity for monitoring, pressuring, and managing the country’s politicians, parties, judges, and journalists. Critics inside Pakistan have long alleged that this capacity has been used to engineer electoral outcomes, to assemble and dissolve political coalitions, to discipline an unruly press, and to keep civilian governments weak enough that they could never challenge the security establishment’s control of foreign and defense policy. Whatever the precise truth of any individual allegation, the structural point is clear and it matters for the proxy story. An intelligence service that spends a large share of its energy managing its own country’s politics is a service whose loyalty runs to the army and to itself rather than to elected authority. That inward-facing political power is what allows the agency to keep the proxy doctrine insulated from democratic debate. The two functions are joined: the domestic political grip protects the external covert project from the scrutiny that might otherwise force it to change. An institution that can shape who governs Pakistan need not worry that those who govern will rein it in.

These traits together produce the central pathology of the institution, which is an inability to learn at the speed its situation demands. The directorate is insulated from political feedback, bound to a founding myth, gripped by a sincere threat perception, and structurally entangled with the very networks that have become its problem. Each trait, on its own, would slow institutional adaptation. Together they nearly prevent it. This is the deepest answer to the question the brief of any honest analysis must pose, the question of whether the agency is declining or merely adapting. The evidence points toward decline, and the reason is precisely this learning deficit. Adaptation requires the capacity to recognize that a core strategy has failed and to change course. The directorate’s institutional character is almost perfectly designed to prevent that recognition. It has adjusted tactics, certainly. It has not changed doctrine. And a service that cannot change doctrine when its doctrine is killing its strategic position is not adapting. It is declining while telling itself a more comfortable story.

There is a counterargument that deserves a fair hearing, because the question is genuinely contested. Some analysts argue that what looks like decline is in fact a deliberate, calculated management of chaos, that the directorate retains its proxies precisely because an ambiguous, partly deniable, partly uncontrolled militant landscape still serves Pakistani interests by keeping India off balance and preserving leverage in Afghanistan. On this reading the agency is not a failing institution but a coldly rational one operating in a difficult environment, accepting messiness as the price of a still-functional strategy. The argument is not absurd, and the evidence does not refute it cleanly. But it strains against a hard fact: an institution genuinely in control of its environment does not find its sheltered clients being killed, one after another, in its own cities, by a rival service it cannot identify, intercept, or deter. Calculated management of chaos is a coherent description of a powerful agency. It is not a coherent description of the situation the directorate actually faces now, which is the subject of the next section.

Where the Agency Stands Now

The present condition of Pakistan’s intelligence directorate can be stated plainly: it is an institution that built a sanctuary and has lost the ability to defend it. The proxy networks still exist. The safe houses, the seminaries that double as recruitment and shelter infrastructure, the front organizations, the designated militants living under various forms of protection, all of this apparatus remains in place. What has changed, and changed decisively, is that the apparatus is no longer secure. The men inside it are being found and killed, and the directorate has been unable to stop it.

This pattern began to take recognizable shape around 2021 and 2022 and accelerated sharply thereafter. Wanted militants, men with long histories in the outfits the agency had cultivated, began dying in Pakistan in circumstances that followed a consistent template. They were shot at close range, often by attackers on motorcycles, often near mosques or homes or businesses, and the attackers vanished. Some were found killed in ways made to resemble personal disputes or sectarian violence. Others were simply gunned down in the street. The Pakistani state’s response cycled through a now-familiar sequence: initial silence or dismissal, then attribution to internal militant feuds, then, eventually, formal accusations against India’s external intelligence service. The arc of that counter-narrative, and the gaps in the evidence offered to support it, are examined in the analysis of how the killings are being investigated and decoded. What matters for an assessment of the directorate is not the attribution question but the simple operational fact underneath it. Designated, sheltered, supposedly protected men were dying, and the agency whose protection they relied on could not keep them alive.

The geography of the pattern made the failure especially stark. The killings were not confined to remote districts where the state’s writ runs thin. They reached into Lahore and its surroundings, into Karachi, into Rawalpindi itself, the garrison city that hosts the army’s general headquarters and the directorate’s own senior offices, and into Sialkot and other towns of the Punjabi heartland, as well as the territory across the Line of Control. These are not ungoverned spaces. They are the most heavily policed and intelligence-saturated places in the country, and the fact that wanted men could be located and killed there carried a message the directorate could not miss: the rival’s reach extended into the agency’s own backyard. The toll accumulated steadily, then faster, until the cumulative figure ran into the dozens, and the seniority of those killed climbed over time. Early targets tended to be operational and mid-tier figures, the kind of men whose deaths could plausibly be dismissed as the result of personal feuds. Later, the campaign reached organizational figures of real weight, men whose elimination removed institutional knowledge, fundraising relationships, and command authority that could not be quickly replaced. A pattern that began as a series of ambiguous incidents had matured into a sustained and discriminating campaign, and the directorate’s inability to interrupt it at any point along that progression was the clearest possible measure of how far its protective capacity had eroded.

This is the precise inversion of the institution’s golden age, and the inversion is worth stating in full because it is the heart of the matter. In the 1980s the directorate ran the largest covert proxy war on the planet and decided the fate of a neighboring country. In the present phase it cannot guarantee the safety of a single wanted man inside one of its own garrison cities. The agency that once exported deniable violence with confidence is now on the receiving end of it, and it finds itself in exactly the position it spent forty years imposing on others. It is the target of a campaign it can neither prevent nor publicly acknowledge, because acknowledging the campaign would mean acknowledging that the men being killed were under Pakistani protection in the first place.

The directorate’s options in this situation are genuinely poor, and surveying them is the clearest way to measure how far the institution has fallen. It cannot openly defend the targets, because openly defending them would confirm the sheltering relationship the state officially denies. It cannot easily relocate or hide them all, because the proxy networks are large, geographically dispersed, and socially embedded, and because the militant leaders themselves have long been accustomed to operating in the open and resist being driven underground. It cannot retaliate symmetrically with confidence, because doing so risks escalation with a larger conventional power and because the rival service has given it few clear targets to retaliate against. And it cannot simply dismantle the networks to deny the rival a target list, both because of the internal institutional resistance described earlier and because dismantling the proxies would mean surrendering the strategic instrument the agency still regards, against the evidence, as essential. The directorate is, in the precise sense, cornered. Every direction it might move is blocked by some combination of its own past commitments and its present weakness.

The conventional military picture compounds the problem rather than relieving it. After the 2025 confrontation between the two countries and the ceasefire that followed it, the conventional channel between India and Pakistan entered a period of relative restraint. But the covert campaign against the sheltered militants did not slow with it. If anything it intensified, which tells an important story about the relationship between the two tracks. The directorate’s traditional doctrine assumed that the proxy and conventional tracks could be managed together, escalated and de-escalated as a single coordinated instrument. The current situation demonstrates that the rival has decoupled them: the conventional track can be quiet while the covert track runs hot. This decoupling is itself a measure of the directorate’s lost control. The agency can no longer set the tempo of the contest. It can only react to a tempo set by someone else.

That loss of tempo has a deeper meaning for an institution built around the management of escalation. The entire proxy doctrine rested on a planning assumption: that the security establishment could read the state of the contest, calibrate the pressure, and turn the violence up or down to keep it within bounds it found tolerable. Deniability, controllability, and below-threshold calibration were the doctrine’s three promises, and all three depended on the directorate retaining the initiative. The current situation breaks the assumption at its foundation. When a rival can run a lethal covert campaign on its own schedule, inside Pakistani cities, regardless of whether the conventional relationship is tense or calm, the directorate is no longer calibrating anything. It is being acted upon. Its planners can no longer promise the army leadership that the proxy contest is being kept within manageable limits, because the limits are now being set elsewhere. An institution that has lost the ability to control the tempo of the very kind of warfare it pioneered has lost the thing that made it strategically distinctive. What remains is an agency still holding the apparatus of a doctrine whose central premise no longer holds.

It is worth being careful and honest about the limits of what can be known here. The directorate is a secretive institution, the rival service is at least as secretive, and the killings themselves are denied by one government and only obliquely discussed by the other. Much of the assessment in this account rests on pattern, on the published work of scholars and journalists, on the memoirs of retired officers with their own agendas, and on the visible behavior of states rather than on documents. A responsible analysis acknowledges that the picture is assembled from circumstantial material and that specific operations cannot be confirmed. But the central judgment does not depend on any single contested fact. It depends on a pattern that is not seriously disputed by anyone: that wanted militants long associated with Pakistan’s proxy apparatus have been dying inside Pakistan in significant numbers, and that the institution responsible for their protection has not stopped it. That pattern, whatever its precise authorship, is sufficient to establish the conclusion. The directorate’s defensive capability, in the one domain where it most needs to perform, has failed.

The agency today is therefore best understood not as a powerful institution managing a difficult environment but as a diminished one absorbing a sustained defeat. It retains real capabilities. It remains formidable inside Pakistan as an instrument of domestic political management, surveillance, and control, and it retains influence over the country’s politics that no elected institution can match. But the specific competence on which its golden-age prestige rested, the management of covert proxy power as a tool of regional strategy, has been hollowed out. The proxies have become liabilities. The sanctuary has become a hunting ground. And the institution that built both can only watch.

The Legacy and the Network It Cannot Protect

Every institution leaves something behind, and the directorate’s legacy is unusually tangible. It is not a doctrine in a manual or a reputation in a chancery. It is a physical and human infrastructure, spread across Pakistani cities and the territory it administers, and the tragedy of the agency’s trajectory is that this infrastructure has outlived the strategic purpose it was built to serve. The legacy is the network, and the network is now the vulnerability.

Consider what the directorate actually constructed over four decades. It built or sponsored a system of training camps. It cultivated a constellation of seminaries that functioned simultaneously as ideological factories, recruitment pipelines, and shelter. It nurtured charity fronts that raised money and laundered the reputations of banned organizations. It arranged the residences, the protected movement, the official non-interference that allowed designated militants to live ordinary visible lives in cities like Lahore, Bahawalpur, Rawalpindi, and Karachi. It helped knit together the leadership cadres of multiple major militant organizations and maintained the relationships that kept those cadres available as instruments. This apparatus was, in its time, a genuine strategic asset, the physical embodiment of the proxy doctrine. The country’s broader sheltering system, the way an entire civilian and institutional landscape was bent to accommodate violence, is the subject of the dedicated mapping of the safe haven network.

The cruel logic of the present phase is that every feature that once made this network valuable now makes it dangerous to the men inside it. The network was valuable because it was open: militant leaders could recruit, fundraise, preach, and organize without hiding, which made them effective instruments. That same openness makes them locatable. The network was valuable because it was embedded: the seminaries and front organizations and protected residences tied the proxies into Pakistani society and gave them durability. That same embeddedness makes them predictable, rooted to known places. The network was valuable because it was extensive, a deep bench of organizations and leaders. That same extent gives a hostile service a long target list. The directorate built a sanctuary optimized for the projection of deniable violence outward, and a sanctuary optimized for projection outward turns out to be poorly designed for the protection of its inhabitants against violence projected inward.

There is a further, quieter dimension to this exposure, and it concerns people rather than buildings. A training camp can be relocated. A charity front can be renamed and reconstituted under a fresh registration. A residence can be exchanged for another residence in a different neighborhood. What cannot be reconstituted quickly is the human capital at the top of these organizations, the founders and operational planners and ideological figureheads who accumulated their authority over thirty or forty years of war. These men are not interchangeable functionaries. Each represents a deep store of relationships, credibility among recruits, and tactical knowledge that took a generation to build, and each is therefore the single most valuable and least replaceable asset in the entire proxy structure. A campaign that concentrates on precisely these figures is attacking the apparatus at the one point where losses compound rather than absorb. Pakistan’s spy service can watch a midlevel operative be replaced within months. It cannot replace a founding commander at all, and it knows that the patient removal of such men, one at a time, hollows the network from the top in a way that no amount of physical reconstruction can offset. The legacy infrastructure is vulnerable, but the legacy leadership is irreplaceable, and a rival that understands the difference will always aim at the second.

This is the doctrinal lesson at the center of the agency’s story, and it generalizes beyond Pakistan. A state that builds a proxy apparatus is building a hostage to fortune. For as long as the state retains overwhelming control of its own territory, intelligence dominance, and the initiative, the apparatus is an asset. The moment a capable adversary develops the reach to operate inside that territory, the apparatus inverts. It becomes a fixed, mapped, attributable collection of targets that the sheltering state cannot defend without confessing to the sheltering. The directorate’s experience is the clearest modern demonstration of this inversion, and it is why the agency’s history reads as a single long decline rather than a cycle. The institution did not merely lose a round of a contest. It built something whose value depended on a permanent advantage it could not in fact keep.

How the directorate’s trajectory compares with that of its principal rival sharpens the point. India’s external service spent its own history moving in the opposite direction, from a defensive intelligence posture toward a more assertive operational one, an evolution traced in the institutional history of the rival agency and in the account of how its operational doctrine changed. The long record of that rival’s activity inside Pakistan, from the 1971 war through the decades since, is examined in the dedicated history of its operations across the border. The two services have followed mirrored arcs. One built an outward-facing proxy machine and then watched its strategic environment turn against it. The other built, more slowly and more recently, the reach to operate inside its rival’s space, and arrived at that capability just as the rival’s accumulated network sat exposed and waiting. The directorate’s decline and the rival’s ascent are not two separate stories. They are one story told from two ends, and the militant networks the agency cannot protect are where the two ends meet.

The regional implications of this reversal extend well beyond the two principal antagonists. For decades the calculation in capitals across South Asia and beyond rested on an assumption that Pakistan’s proxy apparatus, however dangerous, was at least a known quantity, an instrument with an address and a chain of command and therefore a thing that could be negotiated with, pressured, or deterred through its sponsor. The slow erosion of the directorate’s control corrodes that assumption. An apparatus whose sponsor can no longer reliably switch it on or off, protect it, or speak for it is more unpredictable, not less, and unpredictability in a nuclear neighborhood is nobody’s preferred outcome. There is a hard irony in this. The agency’s critics long argued that the proxy networks should be dismantled, and they were right that the networks were a danger. But a network dismantled by its sponsor in an orderly way and a network hollowed out by a hostile service in an attritional campaign are not the same thing, and the second is a far more volatile process than the first. The directorate’s loss of control does not make the region safer in any simple sense. It replaces a managed danger with an unmanaged one, and the unmanaged version carries risks of escalation and miscalculation that the managed version, for all its evils, did not. This is the uncomfortable shape of the present: the proxy doctrine has failed, but its failure is being administered by an adversary rather than corrected by its author, and that distinction will shape South Asian security for years to come.

What survives, then, is a paradox. The directorate’s legacy network still exists, still recruits, still preaches, still poses a danger to the region and beyond. The proxy organizations have not been destroyed, and treating them as spent would be a serious error; they retain the capacity for mass-casualty violence, as their histories make clear. But the network no longer delivers what it was built to deliver. It was meant to give Pakistan a controllable, deniable instrument of regional power. It now gives Pakistan a liability it cannot dismantle and cannot defend, an apparatus that generates international isolation, internal blowback, and a steady toll of dead clients. The directorate spent its golden age believing it had found a permanent shortcut to strategic power. Its legacy is the discovery, written across four decades and an expanding list of graves, that there was no shortcut, only a debt, and that the debt is still being collected.

Frequently Asked Questions

Q: What is the complete history of ISI covert operations?

The history of Pakistan’s intelligence directorate divides naturally into five phases. It began in 1948 as a modest military coordination bureau with no significant covert action role. It was transformed in the 1980s into the central conduit of the anti-Soviet Afghan war, the period of its greatest power. Through the 1990s it applied the proxy formula to Kashmir and to installing a friendly regime in Afghanistan. After 2001 it adopted the double game, cooperating with Washington while sheltering Afghan insurgents, a posture that drained its international credibility. And from roughly 2021 onward it entered a defensive phase in which the militant clients it had sheltered for decades began dying inside Pakistan in a campaign it could not stop. The through line is not growth but a long arc from modest origins through an inflated peak to a diminished present.

Q: When was ISI at its most powerful?

Pakistan’s intelligence directorate reached its peak during the 1980s and into the early 1990s, the years of the anti-Soviet war in Afghanistan and the immediate aftermath. During that period it controlled the entire pipeline of weapons and money flowing to the Afghan resistance, decided which commanders were armed and which were starved, and ran what was at the time the largest covert war anywhere on earth. The Soviet withdrawal, completed in February 1989, was read inside the institution as proof that its proxy doctrine could break a superpower. That conviction, accurate in part and dangerously inflated in part, defined the agency’s self-image for decades afterward and shaped every choice it made as its actual position deteriorated.

Q: How did the Afghan jihad shape ISI’s capability?

The Afghan war shaped the directorate more profoundly than any other single experience. It transformed a small coordinating bureau into a large operational service with deep expertise in training, arming, and directing irregular fighters. It established the principle, insisted on by Islamabad and accepted by Washington, that no foreign service would deal directly with the Afghan resistance, which handed the agency enormous leverage. It built the physical infrastructure of camps and logistics that would later be repurposed for Kashmir. And it created a founding myth, the belief that covert proxy warfare run by the directorate could defeat empires, which became the institutional conviction the agency could never afterward bring itself to abandon even as the evidence against it accumulated. The war also reshaped the agency’s internal culture in ways that outlasted the conflict itself. A generation of officers built their careers, their reputations, and their personal networks inside the Afghan program, and those officers carried the assumptions of that program upward as they were promoted. By the time the most senior leadership of the directorate consisted of men formed by the jihad years, the proxy doctrine was no longer one option among several. It had become the institutional common sense, the lens through which every subsequent problem was instinctively read. That is why the Afghan war matters so much more than its dates suggest: it did not merely give the agency a set of capabilities, it gave the agency an identity, and an institution rarely abandons an identity simply because the strategy attached to it stops working.

Q: Is ISI declining or adapting to new circumstances?

An honest answer has to concede that the question is genuinely contested, but the weight of evidence points toward decline. The case for adaptation holds that the agency deliberately tolerates an ambiguous militant landscape because chaos still serves Pakistani interests. The case for decline holds that an institution genuinely in control would not find its sheltered clients being killed in its own cities by a rival it cannot identify or deter. The deeper issue is the directorate’s institutional inability to learn. It is insulated from political accountability, bound to a founding myth, and structurally entangled with its own proxies, and these traits together prevent the doctrinal change that real adaptation would require. Tactical adjustment is not the same as adaptation, and doctrine has not changed.

Q: How does ISI’s trajectory compare to RAW’s?

The two services have followed mirrored arcs, which is why their histories are best read together. Pakistan’s directorate moved from modest beginnings to an inflated peak as a proxy-running machine and then into decline as that machine became a liability. India’s external service moved more slowly in the opposite direction, from a defensive intelligence posture toward a more assertive operational capability, and reached that capability just as its rival’s accumulated network sat exposed. One agency built an outward-facing apparatus and watched its environment turn hostile. The other built inward reach and arrived at it at the perfect moment. The comparison is developed fully in the dedicated analysis of the two agencies set against each other.

Q: Did ISI manage the Kashmir proxy war independently of the elected government?

To a very substantial degree, yes. The directorate is an arm of the army rather than of the elected state, its leadership is military, and decisions about proxy warfare have frequently been made within the agency and the army command with minimal civilian involvement. The clearest illustration is the 1999 Kargil incursion, which by multiple accounts was planned and launched with much of Pakistan’s own civilian government kept in the dark. This insulation from political accountability is one of the central reasons the agency was able to persist with a failing proxy strategy for so long. An institution that does not have to answer to voters never receives the feedback that would force a correction. It is worth being precise about what this autonomy does and does not mean. It does not mean the directorate operates as a rogue body cut off from the rest of the Pakistani state. It means something more particular, that on questions of proxy warfare and regional strategy the relevant decisions are taken within a closed military circle, and elected officials are informed, managed, or bypassed rather than consulted. Civilian governments in Islamabad have at various points tried to assert control over the agency, and those attempts have generally ended with the civilians retreating. The practical result is a strategy that never had to survive a real democratic argument, never had to be defended to a skeptical parliament or a cost-conscious electorate, and so never accumulated the political opposition that in most countries eventually retires a failing policy.

Q: Has ISI lost control of the militant groups it helped create?

It has lost the kind of control its doctrine originally assumed. In the proxy formula’s design, militant outfits were meant to be precise instruments, switched on and off as strategy required, dependent on the directorate for funding, recruitment, and survival. But the agency allowed its most capable clients to grow into large organizations with independent finances through charity fronts, independent recruitment through their own seminaries, and independent ideological authority. A group that can fund and staff and justify itself is no longer a controllable proxy. Some of these networks even turned their violence against the Pakistani state itself. The directorate retains influence over the surviving outfits, but the clean on-off control the doctrine promised is gone. The mechanism of this drift is worth stating plainly, because it was not an accident so much as a predictable consequence of success. Each tool the directorate gave its clients to make them more effective also made them more independent. Charity fronts were created to give the groups deniable funding, but a group with its own revenue does not need to ask permission. Seminaries were encouraged to supply a steady flow of recruits, but a group that controls its own recruitment pipeline controls its own future manpower. Ideological media and preaching networks were tolerated because they sustained morale and justified the cause, but a group that owns its own ideological authority can define its own enemies. The agency built capable proxies, and capability and obedience turned out to be in tension. The more useful a client became, the less reliably it could be switched off.

Q: What was the ISI double game after 9/11?

The double game refers to the posture the directorate adopted after September 2001. Publicly, Pakistan under General Musharraf aligned with the United States in the campaign against terrorism and received substantial aid in that role. Privately, the agency preserved its Afghan clients, sheltering elements of the Taliban leadership and tolerating the Haqqani network as a strategic reserve for the day American forces departed. The agency thus ran both sides at once, collecting the benefits of alliance while keeping its proxy assets intact. The posture looked clever in the short term but required sustained deception of Pakistan’s most important partner, and its eventual exposure did lasting damage to the agency’s international credibility. The double game is sometimes described as a clever hedge, and in narrow tactical terms it was: it did preserve a set of assets that became relevant again when Western forces eventually left Afghanistan. But the strategic price was severe and is often underweighted. Sustaining the deception required the directorate to mislead the partner whose aid, intelligence sharing, and diplomatic cover it most depended on, and a relationship built on a deception that large cannot survive the deception’s discovery. When the accumulated evidence of sheltering became impossible to dismiss, the agency did not simply lose a round of bargaining. It lost the presumption of good faith, and once an intelligence service is presumed to be acting in bad faith, every subsequent assurance it offers is discounted in advance. The double game bought time and kept some assets alive, but it spent something harder to rebuild, which was the willingness of others to take the institution at its word.

Q: What did the Abbottabad raid reveal about ISI?

In May 2011 the operation that killed Osama bin Laden located him not in a remote tribal hideout but in a substantial compound in Abbottabad, a garrison town, near Pakistan’s premier military academy. The episode forced an unforgiving choice between two explanations. Either the directorate had failed to notice the most wanted man on earth living for years beside its own officer corps, which implied a staggering level of incompetence, or elements of the agency had known, which implied complicity. Neither explanation was survivable for the institution’s reputation. The raid crystallized a judgment, already forming in much of the world, that Pakistan’s intelligence service was not a flawed ally but an active adversary, and trust did not recover.

Q: Why can ISI not protect the militants being killed in Pakistan?

The directorate is cornered by its own past choices. It cannot openly defend the targeted men because doing so would confirm the sheltering relationship the Pakistani state officially denies. It cannot easily hide a proxy apparatus that is large, dispersed, socially embedded, and accustomed to operating in the open. It cannot retaliate symmetrically with confidence against a rival that has given it few clear targets and that operates under the shadow of a larger conventional power. And it cannot simply dismantle the networks, both because of internal institutional resistance and because it still regards the proxies as a strategic asset. Every available option is blocked by some combination of past commitment and present weakness.

Q: What is the covert action wing of the ISI?

Within the directorate, the wing that handles relationships with militant organizations and the management of proxy forces has long operated with unusual autonomy. Outside scholars studying the agency have described it as something close to a state within the intelligence service, shielded from outside scrutiny and operating with extraordinary latitude. Its long-serving officers developed personal relationships with militant leaders that span decades, ties that are not easily severed by any policy decision. This structural autonomy is one reason the agency’s repeated promises to act against particular groups have so consistently produced cosmetic results: the machinery for actually dismantling the proxy relationships is weak and resisted from within. There is an important consequence that follows from this internal structure. Because the relationships with militant organizations are held by a relatively narrow cadre of long-serving specialists, those relationships are personal as much as institutional, and personal ties of two or three decades’ standing do not dissolve on the strength of a policy memorandum. When the political leadership signals a desire to act against a particular group, the order has to pass through the very officers whose careers, contacts, and self-image are bound up with keeping that group intact. The result is a built-in friction that converts firm decisions into cosmetic gestures, arrests that precede quiet releases, bans that produce renamed organizations. Anyone trying to understand why the directorate’s pledges to act against its proxies have so rarely produced durable results has to look here, at the gap between what the institution’s leadership can announce and what its proxy-handling cadre will actually permit.

Q: How did the Kargil operation in 1999 damage ISI’s strategy?

The Kargil incursion was an attempt to test the proxy doctrine against the conventional balance directly. Pakistani forces, including paramilitary and irregular elements, infiltrated across the Line of Control to seize heights and present India with a fait accompli below the threshold of full war. The attempt failed completely. India mounted a substantial campaign to retake the heights, the international community sided overwhelmingly with respecting the Line of Control, and Pakistan was forced into a humiliating withdrawal. Kargil exposed a hard limit of the directorate’s doctrine: the deniable, below-threshold approach did not scale up to seizing and holding territory, and pushing it that far produced strategic isolation rather than the advantage its planners expected.

Q: Did ISI create the Kashmir insurgency?

No, and precision matters here. By the late 1980s the disputed territory had genuine indigenous grievances, including a state election in 1987 widely seen as rigged, which discredited electoral politics and pushed many young Kashmiris toward armed resistance. An indigenous uprising was already stirring. What the directorate did was recognize the opportunity and move to capture it, steering training, weapons, and sanctuary toward groups that favored merger with Pakistan and a religious framing of the struggle, while the indigenous and relatively secular factions were starved of support. Over the 1990s the agency went further, shifting from supporting Kashmiri fighters to inserting Pakistani ones. The agency did not start the fire, but it poured fuel on it and changed its character. That change of character is the part most worth dwelling on, because it had consequences that ran against Pakistan’s own stated aims. An indigenous Kashmiri movement, rooted in local grievances and led by Kashmiris, would have been politically harder for India to dismiss as simple cross-border terrorism. By steering the struggle toward groups that favored merger with Pakistan, that imported non-Kashmiri fighters, and that framed the conflict in explicitly religious terms, the directorate made the insurgency easier to portray as an external proxy war rather than a homegrown revolt. In trying to capture the uprising, the agency helped strip it of the very indigenous legitimacy that would have made it most useful as a political weapon. It is one of the clearest cases in the agency’s record of a proxy intervention that advanced the institution’s sense of control while undercutting the larger objective the control was meant to serve.

Q: How autonomous is ISI from civilian control in Pakistan?

The directorate operates with a degree of autonomy from elected institutions that has few parallels among major intelligence services. Its leadership is military, its director general is a serving general, and its officers expect to return to military careers, which means its incentives are the army’s incentives. The Pakistani army has long regarded itself as the true guardian of the state, standing above civilian politics, and the intelligence directorate inherits that posture in concentrated form. Major decisions about proxy warfare, militant relationships, and regional strategy have frequently been made with minimal civilian input. This insulation is central to understanding why the agency could persist for so long in a strategy that was visibly failing.

Q: What is ISI’s relationship with Lashkar-e-Taiba and Jaish-e-Mohammed?

The directorate’s relationship with these organizations is the clearest example of how its proxy strategy created liabilities. Lashkar-e-Taiba, which grew from a missionary and militant movement headquartered near Lahore, became perhaps the most disciplined of the agency’s instruments and was long regarded as the most reliable, the proxy least likely to slip its leash. The outfits that consolidated into Jaish-e-Mohammed followed a parallel path of cultivation. But by nurturing these groups into mature organizations with independent funding, recruitment, and ideological authority, the agency created clients that no longer depended on it for survival. The intimate institutional ties between the spy service and these outfits are documented in the dedicated analysis of the intelligence and terror nexus.

Q: Did the proxy strategy ever turn against Pakistan itself?

Yes, and this is one of the heaviest costs of the directorate’s doctrine. Through the 2000s and 2010s the militant ecosystem the agency had cultivated grew larger and more radical, and a significant portion of it turned inward. The Pakistani Taliban movement, formed in the tribal areas, launched a campaign of mass-casualty violence against mosques, markets, schools, and security installations inside Pakistan. The directorate had spent decades teaching that jihadist violence was a legitimate instrument of policy, and some of the men who absorbed that teaching concluded that the Pakistani state itself was a legitimate target. The agency tried to distinguish good militants from bad ones, but the distinction its own history made incoherent. The blowback also imposed a cost that is easy to overlook, which is the cost of divided attention and divided resources. An intelligence service fighting a domestic insurgency cannot give the same focus to projecting power outward, and the directorate found itself spending the 2010s simultaneously running proxies abroad and battling proxies’ ideological cousins at home. The 2014 massacre of schoolchildren in Peshawar made the contradiction impossible to ignore, producing a national plan against militancy that, on paper, committed the state to acting against all armed groups without distinction. In practice the old separation between useful proxies and hostile ones persisted, because abandoning it would have meant abandoning the doctrine itself. The proxy strategy had reached the point where it generated the very threat it then had to be defended against, and the institution still could not bring itself to conclude that the strategy was the problem.

Q: Can ISI stop the targeted killings inside Pakistan?

The evidence so far indicates that it cannot, at least not with the tools and posture it currently has. The pattern of wanted militants being killed inside Pakistan took recognizable shape around 2021 and 2022 and intensified afterward, including through the period after the 2025 confrontation when the conventional channel between the two countries was relatively quiet. That the covert campaign continued while the conventional track was calm suggests the rival has decoupled the two, running one hot while the other stays cold. The directorate can no longer set the tempo of the contest; it can only react to a tempo set by someone else, which is itself a measure of how far its operational control has eroded. What would actually stopping the campaign require, and why is each path blocked? Hardening the targets would mean moving militant leaders into hiding, which destroys their value as recruiters and organizers and amounts to defeating the proxy doctrine in order to save it. Deterring the rival would require credible retaliation, which the directorate struggles to deliver against an adversary that presents few fixed targets and operates beneath the umbrella of a larger conventional power. Negotiating an end would require admitting the sheltering relationship that gives the campaign its targets, which the Pakistani state cannot do without enormous diplomatic cost. Dismantling the networks outright would remove the targets but also surrender the doctrine, and the institution is not willing. Every exit is sealed by some earlier commitment, which is why the killings continue.

Q: What is the lasting legacy of ISI covert operations?

The directorate’s legacy is a physical and human infrastructure: training facilities, seminaries that double as recruitment and shelter, charity fronts, protected residences, and the leadership cadres of multiple militant organizations. In its time this apparatus was a genuine strategic asset, the embodiment of the proxy doctrine. The paradox of the present is that every feature that once made the network valuable, its openness, its social embeddedness, its sheer extent, now makes it a liability. A sanctuary built to project deniable violence outward turns out to be poorly designed to protect its own inhabitants from violence projected inward. The agency believed it had found a permanent shortcut to strategic power, and its legacy is the discovery that there was only a debt.

Q: Why does ISI’s history matter for understanding the shadow war?

Because the directorate built the thing the shadow war is dismantling. The campaign of targeted killings inside Pakistan is not striking random individuals; it is striking the human nodes of an apparatus the agency spent four decades constructing. Understanding why that apparatus is vulnerable requires understanding how and why it was built, what strategic assumptions underpinned it, and how those assumptions failed. The agency’s decline from the institution that ran the largest covert war on earth to one that cannot protect a single client in its own garrison cities is the institutional precondition for everything happening now. The networks were always a debt rather than an asset, and the shadow war is the form the collection has taken.