Files
PaperClipAI/server/src/services/environment-task-runtime.ts
T
Nicky LeachandPaperclip 6d72243d8a fix(runtime): recover task creation after a lost acquire reply (#18692)
## Thinking Path

> - Paperclip manages agent execution and the resources that execution
owns.
> - A retained environment task can need a replacement before the next
turn.
> - Core records the creation attempt before it asks the provider to
acquire it.
> - A lost reply can leave that journal without the resource ID needed
for cleanup.
> - This change lets an opted-in provider recover the identity from that
recorded attempt.
> - Core still requires the existing exact-task termination proof before
it releases ownership.

## Linked Issues or Issue Description

Refs #18405 and #18658. This is the separate lost-reply follow-up from
their recovery audit.

**Draft status:** Both prior review findings are fixed in
`a66520c8d4e06e0eec6ccb5446190ace0ca3003f`. Fresh standard Greptile is
5/5 with no unresolved threads. Current-head CI is green (53 successful
checks/statuses including Snyk, three expected skips). Apex completed
5/5 on this exact head at 2026-11-03T00:54:59Z (request 6592407130),
with no unresolved threads. The review cycle is complete; the PR remains
draft for human merge review.

**What happened?**

If a replacement task's acquire reply is never received, the creation
journal can reach manual reconciliation without a resource ID. Startup
cleanup cannot complete, and ordinary Stop cannot release the run owner.

**Expected behavior**

When the selected provider supports creation inspection, Core should
retain enough binding information to inspect the recorded attempt and
continue verified cleanup. An inspected identity must not count as
termination proof.

**Steps to reproduce**

1. Use a retained task whose provider supports creation inspection.
2. Lose the replacement acquire reply before Core records its resource
ID.
3. Restart cleanup and inspect the existing creation journal.
4. Confirm that the exact task is stopped and its termination receipt is
saved before cancellation releases the owner.

**Paperclip version or commit**

Current head `a66520c8d4e06e0eec6ccb5446190ace0ca3003f`, based on master
`3a7bb4dac62dbd7a34a55deb665c55df310bdf64`. Tested with isolated
PostgreSQL fixtures.

**Deployment mode**

Local source tests with controlled provider responses. No deployment or
live model request.

## What Changed

- Save the selected inspection capability and prior opaque binding in
the existing creation journal. Acquire through that same worker so a
worker replacement cannot change the recorded contract.
- Make failed opted-in creations due through the existing cleanup
notification and saved retry deadline.
- Add optional SDK observation field `creationLease`. Use it only to
recover the requested creation's identity and binding, then follow the
existing Stop, status, completion and receipt checks.
- Keep old entries and providers without the recorded opt-in under their
existing reconciliation policy.
- Add actual PostgreSQL recovery, cancellation and worker-replacement
regressions. Document the provider contract.

## Verification

- Current correction: 53 distinct targeted cases pass (20 real-worker
IPC, 12 task-recovery PostgreSQL and 21 generic-sandbox cases), with
server typechecking. The IPC selection was rerun after the final
late-observation ordering refinement; the other tested sources were
unchanged.
- An assembled live-child IPC and PostgreSQL case proves cleanup and
cancellation release the owner despite the original worker never
replying and remaining alive. Accepted host work, unrelated acquisitions
and late observations remain accounted for. Removing the recovery
handoff and restoring the old drain snapshot produce distinct assertion
failures.
- The unsupported-host control skips all 12 PostgreSQL cases without
setup; removing the guard fails the controlled setup. These tests do not
constitute live VM/provider qualification.
- Initial head `5db6882` passed an exact reference-tree comparison
against master plus the reviewed follow-up patch, 97 targeted cases and
server typechecking. Those checks predate the final inspection
correction.
- Previous head `59f87fa` passed 32 targeted cases (11 task-recovery and
21 generic-sandbox cases) and server typechecking. Both `pending` and
`needs_reconciliation` observations without `creationLease` preserve
ownership and get a future retry deadline. Restoring the previous guard
makes both new cases fail.
- The reproduced failure was a wrong generic cleanup shape that could
resolve its journal while leaving the owner held—not a demonstrated
false ownership release. The correction reuses the existing
missing-identity guard before the generic decoder and leaves generic
sandbox behavior unchanged.
- Earlier affected selection: 72 cases passed on the original follow-up
base. This overlaps the final selection and is not an additional count
of unique final-head tests.
- Two isolated negative controls restore the missing task inspection and
redirected-worker behaviors. Both fail at the expected recovery and
dispatch assertions without changing production source files.
- The PostgreSQL cases verify one cleanup lease, both late-result
orders, recovery after the old journal deadline, ordinary legacy
behavior, retained ownership on missing or mismatched identity, and
cancellation acknowledgement after exact termination proof.
- The tests use actual database and run-control services with controlled
worker callbacks. They do not prove live IPC or production recovery
timing.
- A compatible provider is needed for the new recovery path. No normal
successful-startup database or network call is added. No ordinary TTFT
improvement is claimed; the repair removes an indefinite ownership hold
for supported lost replies. No full local CI suite was run; GitHub runs
the full regression gates.

## Risks

The provider must recover the identity for the recorded creation
operation and preserve its selected binding. `creationLease` does not
authorize a new turn or prove that a process stopped. Older journal
entries do not infer a new identity after an upgrade. Unavailable
original machine bindings or credentials retain the existing uncertainty
behavior; this change does not repair machine or credential loss. No
schema migration, deployment or ROM qualification is included.

## Model Used

OpenAI Codex, GPT-6 family. Exact model ID and context window are not
exposed in this session. Tool-assisted implementation, targeted tests
and independent subagent review.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-11-02 17:45:07 -08:00

514 lines
41 KiB
TypeScript

import { attachSandboxCreation, environmentService } from "./environments.js";
import { createOwnedSandbox, queueSandboxCreationCleanup, recordSandboxCreationResource } from "./sandbox-creation-operations.js";
import { writeEnvironmentCleanup } from "./environment-cleanup-notifications.js";
import { releaseProviderRecoveryTransfers, stageProviderFreshWorkspace } from "./provider-fresh-workspace.js";
import { readWorkspaceProviderBinding } from "@paperclipai/shared";
import { settlePluginTaskCredentialHandoff } from "./plugin-task-credential-writeback.js";
import { and, eq, inArray, or, sql } from "drizzle-orm";
import { environmentLeases, executionWorkspaces, environments, heartbeatRuns, issues, projects, type Db } from "@paperclipai/db";
import { freshWorkspaceSchema, environmentTaskOperationSchema, parseEnvironmentTaskResult,
type HostToWorkerMethodName, type HostToWorkerMethods, type PluginEnvironmentTaskOperation } from "@paperclipai/plugin-sdk";
import { pluginRegistryService } from "./plugin-registry.js";
import { hasRemoteTerminationReceipt, remoteExecutionHasStopped, remoteTerminationReceipt } from "./remote-execution-termination.js";
import type { PluginWorkerManager } from "./plugin-worker-manager.js";
import { NativeExecutionContext } from "./native-runtime/native-execution-context.js";
import { pluginAgentRuntimeIdentity } from "./plugin-agent-runtime-state.js";
export async function pluginEnvironmentSupportsTasks(db: Db, workers: PluginWorkerManager, lease: { companyId: string; metadata: Record<string, unknown> | null }) {
const metadata = lease.metadata ?? {};
if (metadata.driver !== "plugin" || typeof metadata.pluginId !== "string" || typeof metadata.driverKey !== "string") return false;
const plugin = await pluginRegistryService(db).getById(metadata.pluginId);
const driver = plugin?.manifestJson.environmentDrivers?.find(value => value.driverKey === metadata.driverKey);
const methods = workers.getWorker(metadata.pluginId)?.supportedMethods ?? [];
return plugin?.status === "ready" && driver?.supportsTasks === true &&
plugin.manifestJson.capabilities.includes("environment.drivers.register") &&
methods.includes("environmentTask");
}
/** Server-only dispatch for plugin-provided Runner execution.
* The caller supplies an authorized company and a persisted lease,
* never a plugin identity, resource endpoint, or agent/project authority.
* A timeout is ambiguous: retry the same operation against the same lease.
*/
export async function executeEnvironmentTask(db: Db, workers: PluginWorkerManager, input: {
companyId: string;
leaseId: string;
operation: PluginEnvironmentTaskOperation;
executionContext?: unknown;
}) {
const operation = environmentTaskOperationSchema.parse(input.operation);
const [row] = await db.select({ lease: environmentLeases, environment: environments, run: heartbeatRuns })
.from(environmentLeases)
.leftJoin(environments, eq(environments.id, environmentLeases.environmentId))
.leftJoin(heartbeatRuns, and(eq(heartbeatRuns.id, environmentLeases.heartbeatRunId), eq(heartbeatRuns.companyId, environmentLeases.companyId)))
.where(and(eq(environmentLeases.id, input.leaseId), eq(environmentLeases.companyId, input.companyId))).limit(1);
if (!row || !row.lease.providerLeaseId) throw new Error("Plugin-provided Runner execution lease unavailable");
const { lease, environment, run } = row;
const context = input.executionContext;
if (context !== undefined) {
if (!(context instanceof NativeExecutionContext) || !run) {
throw new Error("native_execution_context_unavailable");
}
context.assertScope({ companyId: input.companyId, agentId: run.agentId, runId: run.id });
}
const taskId = row.lease.providerLeaseId;
const metadata = lease.metadata ?? {};
const probe = metadata.runnerTaskProbe as { runId?: unknown; sessionId?: unknown; expiresAtMs?: unknown } | undefined;
const activeProbe = lease.heartbeatRunId === null && lease.issueId === null && probe
&& typeof probe.runId === "string" && typeof probe.sessionId === "string"
&& typeof probe.expiresAtMs === "number" && probe.expiresAtMs > Date.now();
const checkpointIdentity = operation.kind === "read_state" || operation.kind === "read_file" ? operation.identity
: operation.kind === "archive_state" ? operation.priorIdentity : null;
const runner = checkpointIdentity ? { runnerId: checkpointIdentity.runnerInstanceId, leaseId: checkpointIdentity.environmentLeaseId,
runId: checkpointIdentity.runId, sessionId: checkpointIdentity.normalizedSessionId } : operation.kind === "submit" ? operation.runner : operation.kind === "prepare" ? {
runnerId: operation.requirements.identity.runnerInstanceId,
leaseId: operation.requirements.identity.environmentLeaseId,
runId: operation.requirements.identity.runId, sessionId: operation.requirements.identity.normalizedSessionId,
} : null;
const launches = operation.kind === "submit" || operation.kind === "prepare";
const scopesProjects = launches || operation.kind === "check_project_auth" || operation.kind === "read_file";
const resumeFrom = launches || checkpointIdentity ? (operation as { resumeFrom?: string }).resumeFrom : undefined;
const savedIdentity = metadata.runnerTaskIdentity as Record<string, unknown> | undefined;
const prior = operation.kind === "prepare" ? operation.priorIdentity : undefined;
const priorSaved = metadata.runnerTaskPriorIdentity as Record<string, unknown> | undefined;
const matches = (identity: typeof prior, saved: Record<string, unknown> | undefined) => !!identity && !!saved &&
identity.runnerInstanceId === saved.runnerId && identity.environmentLeaseId === saved.leaseId &&
identity.normalizedSessionId === saved.sessionId && identity.runId === saved.runId &&
identity.turnId === saved.turnId && identity.itemId === saved.itemId;
const ownCheckpoint = checkpointIdentity && !resumeFrom &&
(matches(checkpointIdentity,savedIdentity) || operation.kind !== "read_file" && matches(checkpointIdentity,priorSaved));
if (checkpointIdentity && !resumeFrom && !ownCheckpoint) throw new Error("Plugin Runner state identity mismatch");
const retainedCheckpointRead = operation.kind === "read_state" && ownCheckpoint &&
lease.leasePolicy === "reuse_by_environment" && ["retained", "expired"].includes(lease.status) &&
run?.status === "succeeded" && metadata.runnerTaskDispatched === true &&
(metadata.runnerTaskLifecyclePolicy as {mode?:unknown}|undefined)?.mode === "warm";
const retainedConnection = operation.kind === "connection" && environment !== null &&
lease.leasePolicy === "reuse_by_environment" && lease.status === "retained" &&
run?.status === "succeeded" && savedIdentity?.runId === run.id &&
metadata.runnerTaskDispatched === true && metadata.runnerTaskExecutionStopped !== true &&
metadata.runnerTaskRetiring !== true && (!lease.expiresAt || lease.expiresAt.getTime() > Date.now()) &&
(metadata.runnerTaskLifecyclePolicy as { mode?: unknown } | undefined)?.mode === "warm";
if ((operation.kind === "read_file" || operation.kind === "describe" || operation.kind === "prepare" || operation.kind === "submit" || (operation.kind === "connection" && !retainedConnection) || (operation.kind === "read_state" && metadata.runnerTaskExecutionStopped !== true && !retainedCheckpointRead) || operation.kind === "archive_state" || operation.kind === "check_project_auth" || operation.kind === "copilot_metadata") &&
(metadata.runnerTaskRetiring === true || !environment || (!activeProbe && run?.status !== "running") || lease.status !== "active" || (lease.expiresAt && lease.expiresAt.getTime() <= Date.now()))) throw new Error("Plugin-provided Runner execution lease is not active");
if (operation.kind === "copilot_metadata" && !activeProbe) throw new Error("Copilot metadata requires a private setup lease");
const warmAttachment = !!prior && metadata.runnerTaskReused === true &&
(matches(prior, savedIdentity) || matches(prior, priorSaved) && operation.kind === "prepare" && matches(operation.requirements.identity, savedIdentity)) && runner?.runnerId === prior.runnerInstanceId &&
runner?.leaseId === prior.environmentLeaseId && runner?.sessionId === prior.normalizedSessionId;
if (prior && !warmAttachment) throw new Error("Plugin warm task prior identity mismatch");
if (runner && ((!checkpointIdentity || !resumeFrom) && !(ownCheckpoint && operation.kind !== "read_file") && runner.runId !== (activeProbe ? probe.runId : run?.id) || (!resumeFrom && runner.leaseId !== lease.id && !ownCheckpoint && !warmAttachment)
|| (activeProbe && (runner.sessionId !== probe.sessionId || (launches && operation.projectIds.length > 0))))) {
throw new Error("Plugin-provided Runner execution identity mismatch");
}
if (runner && resumeFrom) {
const inspectCheckpoint = operation.kind === "read_state";
const sources = await db.select({ id: environmentLeases.id, runId: environmentLeases.heartbeatRunId, metadata: environmentLeases.metadata }).from(environmentLeases).where(and(
eq(environmentLeases.companyId, input.companyId), eq(environmentLeases.provider, lease.provider!),
eq(environmentLeases.providerLeaseId, resumeFrom),
inspectCheckpoint ? undefined : sql`${environmentLeases.metadata}->>'runnerTaskSessionRetained' = 'true'`,
checkpointIdentity ? eq(environmentLeases.heartbeatRunId, checkpointIdentity.runId) : undefined,
)).limit(2);
const priorRunId = sources.length === 1 ? sources[0]!.runId : null;
const sourceIdentity = sources.length === 1 ? sources[0]!.metadata?.runnerTaskIdentity as Record<string, unknown> | undefined : undefined;
if (activeProbe || !priorRunId || (checkpointIdentity && (checkpointIdentity.runId !== priorRunId
|| checkpointIdentity.turnId !== sourceIdentity?.turnId || checkpointIdentity.itemId !== sourceIdentity?.itemId)) || resumeFrom !== await resolvePluginTaskResume(db, {
companyId: input.companyId, leaseId: lease.id,
priorIdentity: { runnerInstanceId: runner.runnerId, environmentLeaseId: runner.leaseId,
runId: priorRunId, normalizedSessionId: runner.sessionId },
inspectCheckpoint,
})) throw new Error("Plugin Runner resume identity mismatch");
// Older providers misclassified suspended checkpoints. A fresh provider
// attestation may repair availability, but never supplies stop/owner authority.
if (inspectCheckpoint && sources[0]!.metadata?.runnerTaskSessionRetained !== true) {
const status = await executeEnvironmentTask(db, workers, { companyId: input.companyId,
leaseId: sources[0]!.id, operation: { kind: "status" } });
if (status.kind !== "status" || status.executionStopped !== true || status.sessionStateAvailable !== true ||
resumeFrom !== await resolvePluginTaskResume(db, { companyId: input.companyId, leaseId: lease.id,
priorIdentity: { runnerInstanceId: runner.runnerId, environmentLeaseId: runner.leaseId,
runId: priorRunId, normalizedSessionId: runner.sessionId } })) {
throw new Error("Plugin Runner retained session is not authorized");
}
}
}
if (metadata.driver !== "plugin" || typeof metadata.pluginId !== "string" || typeof metadata.driverKey !== "string") {
throw new Error("Plugin-provided Runner execution lease has no pinned plugin driver");
}
const plugin = await pluginRegistryService(db).getById(metadata.pluginId);
const driver = plugin?.manifestJson.environmentDrivers?.find(value => value.driverKey === metadata.driverKey);
if (!plugin || plugin.status !== "ready" || !driver?.supportsTasks ||
!plugin.manifestJson.capabilities.includes("environment.drivers.register") ||
!workers.getWorker(plugin.id)?.supportedMethods.includes("environmentTask")) {
throw new Error("Plugin-provided Runner execution provider unavailable");
}
const callWorker = async <M extends HostToWorkerMethodName>(method: M, params: HostToWorkerMethods[M][0], timeout?: number): Promise<HostToWorkerMethods[M][1]> => {
if (!context) {
const cleanup = metadata.runnerTaskStartupCleanup as { contextId?: unknown; requiresPreparationStop?: unknown } | undefined;
if (["stop", "status", "complete", "credentials"].includes(operation.kind) && typeof cleanup?.contextId === "string" && typeof cleanup.requiresPreparationStop === "boolean") {
return workers.call(plugin.id, method, params, timeout, undefined, undefined, cleanup.contextId);
}
return timeout === undefined ? workers.call(plugin.id, method, params) : workers.call(plugin.id, method, params, timeout);
}
const settled = context.trackWorker();
const result = await workers.call(plugin.id, method, params, Math.max(1, Math.min(timeout ?? 15_000, context.deadlineAtMs - Date.now())), undefined, settled, context.id);
context.assertOpen();
return result;
};
const issue = scopesProjects && lease.issueId ? await db.select({ id: issues.id }).from(issues)
.where(and(eq(issues.id, lease.issueId), eq(issues.companyId, input.companyId))).limit(1).then(rows => rows[0]) : null;
if (scopesProjects && lease.issueId && !issue) throw new Error("Plugin-provided Runner execution issue unavailable");
const workspaceBinding = readWorkspaceProviderBinding(lease.metadata);
if (metadata.workspaceProvider !== undefined && !workspaceBinding) throw new Error("Invalid managed workspace binding");
if (workspaceBinding) {
const [workspace] = await db.select().from(executionWorkspaces).where(and(
eq(executionWorkspaces.id, workspaceBinding.executionWorkspaceId), eq(executionWorkspaces.companyId, input.companyId)));
const saved = readWorkspaceProviderBinding(workspace?.metadata);
if (!workspace || workspace.id !== lease.executionWorkspaceId || workspace.providerType !== "plugin" || workspace.providerRef !== workspaceBinding.workspaceRef ||
(scopesProjects && !["active", "idle", "in_review"].includes(workspace.status)) ||
workspaceBinding.pluginId !== plugin.id || !saved || JSON.stringify(saved) !== JSON.stringify(workspaceBinding)) {
throw new Error("Managed workspace execution binding changed");
}
}
let freshWorkspace = metadata.freshWorkspace === undefined ? undefined : freshWorkspaceSchema.parse(metadata.freshWorkspace);
if (freshWorkspace && driver.freshWorkspaceVersion !== 1 && (launches || operation.kind === "describe" || operation.kind === "check_project_auth"))
throw new Error("Plugin-provided fresh workspace preparation unavailable");
if (freshWorkspace && operation.kind === "prepare") {
if (!lease.issueId) throw new Error("Fresh workspace task ownership is missing");
freshWorkspace = await stageProviderFreshWorkspace({ db, companyId: input.companyId,
issueId: lease.issueId, ...freshWorkspace, existing: freshWorkspace });
}
const projectIds = scopesProjects ? operation.projectIds : [];
if (scopesProjects && Array.isArray(metadata.runnerTaskProjectIds) &&
JSON.stringify(projectIds) !== JSON.stringify(metadata.runnerTaskProjectIds)) throw new Error("Plugin Runner project scope changed");
if (activeProbe && projectIds.length > 0) throw new Error("Plugin Runner probe cannot use project mounts");
if (projectIds.length > 0) {
const authorizedProjects = await db.select({ id: projects.id }).from(projects)
.where(and(eq(projects.companyId, input.companyId), inArray(projects.id, projectIds)));
if (authorizedProjects.length !== projectIds.length) throw new Error("Plugin-provided Runner execution project unavailable");
}
// Provider identity comes from the lease. Editing the environment must never
// redirect status or cleanup to a replacement plugin.
const config = (environment?.config ?? {}) as Record<string, unknown>;
const driverConfig = config.pluginKey === plugin.pluginKey && config.driverKey === metadata.driverKey
? (config.driverConfig as Record<string, unknown> | undefined) ?? {} : {};
let requiredProjectAuthProof: Record<string, unknown> | undefined;
if (operation.kind === "read_file") {
const proof = metadata.runnerTaskProjectAuthValidation as { projectIds?: string[] } | undefined;
if (metadata.runnerTaskExecutionStopped === true || !Array.isArray(proof?.projectIds) ||
JSON.stringify([...proof.projectIds].sort()) !== JSON.stringify([...operation.projectIds].sort()))
throw new Error("Plugin Runner file scope unavailable");
}
if (operation.kind === "submit") {
const proof = metadata.runnerTaskProjectAuthValidation as { configurationId?: unknown; provider?: unknown; projectIds?: string[] } | undefined;
const provider = operation.runner.harness === "codex" ? "openai" : operation.runner.harness === "claude" ? "anthropic" : operation.runner.harness;
if (!provider || !operation.preparedLaunchId || proof?.configurationId !== operation.preparedLaunchId || proof.provider !== provider
|| (!Array.isArray(proof.projectIds) ? null : JSON.stringify([...proof.projectIds].sort())) !== JSON.stringify([...operation.projectIds].sort())
|| (operation.credential && (operation.credential.type === "codex-auth-json" ? "openai" : "anthropic") !== provider)) {
throw new Error("Plugin Runner managed project authentication validation unavailable");
}
requiredProjectAuthProof = proof as Record<string, unknown>;
}
if (operation.kind === "check_project_auth" || operation.kind === "prepare") {
await db.update(environmentLeases).set({ metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) - 'runnerTaskProjectAuthValidation'` })
.where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, input.companyId)));
}
if (operation.kind === "submit" && activeProbe) {
const remainingMs = Number(probe.expiresAtMs) - Date.now();
if (remainingMs <= 0 || !operation.launch) throw new Error("Plugin-provided Runner probe lease is not active");
// Runner itself must stop if Core disappears during a setup probe.
const requestedLimits = [operation.launch.maxRuntimeMs, operation.launch.maxLifetimeMs]
.filter((value): value is number => typeof value === "number" && value > 0);
operation.launch.maxRuntimeMs = Math.min(remainingMs, ...requestedLimits);
delete operation.launch.maxLifetimeMs;
}
if (operation.kind === "submit" || operation.kind === "copilot_metadata") {
// No ticket, headers, environment values, or launch paths belong in metadata.
const marked = await db.update(environmentLeases).set({
metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || ${JSON.stringify({ runnerTaskLease: true, runnerTaskDispatched: true, runnerTaskExecutionStopped: false, runnerTaskSessionRetained: false, ...(operation.kind === "submit" ? { runnerTaskIdentity: operation.runner, runnerTaskResumeFrom: operation.resumeFrom ?? null } : {}) })}::jsonb`,
}).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, input.companyId),
eq(environmentLeases.status, "active"), sql`coalesce(${environmentLeases.metadata}->>'runnerTaskRetiring', 'false') != 'true'`,
requiredProjectAuthProof ? sql`${environmentLeases.metadata}->'runnerTaskProjectAuthValidation' = ${JSON.stringify(requiredProjectAuthProof)}::jsonb` : undefined,
)).returning({ id: environmentLeases.id });
if (marked.length !== 1) throw new Error("Plugin-provided Runner execution lease is not active");
}
if (warmAttachment && operation.kind === "prepare") {
const i = operation.requirements.identity;
const startup = context?.pluginSupported ? { contextId: context.id, deadlineAtMs: context.deadlineAtMs, runId: i.runId } : undefined;
const marked = await db.update(environmentLeases).set({ metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || ${JSON.stringify({
runnerTaskPriorIdentity: { runnerId: prior!.runnerInstanceId, leaseId: prior!.environmentLeaseId, runId: prior!.runId,
sessionId: prior!.normalizedSessionId, turnId: prior!.turnId, itemId: prior!.itemId },
runnerTaskIdentity: { runnerId: i.runnerInstanceId, leaseId: i.environmentLeaseId, runId: i.runId,
sessionId: i.normalizedSessionId, turnId: i.turnId, itemId: i.itemId },
...(startup ? { runnerTaskStartup: startup } : {}),
})}::jsonb` }).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.status, "active"),
sql`coalesce(${environmentLeases.metadata}->>'runnerTaskRetiring', 'false') != 'true'`,
sql`${environmentLeases.metadata}->'runnerTaskIdentity' = ${JSON.stringify(savedIdentity)}::jsonb`)).returning({id:environmentLeases.id});
if (marked.length !== 1) throw new Error("Plugin warm task attachment lost authority");
}
try {
if (operation.kind === "describe" && metadata.runnerTaskReused === true) {
const state = parseEnvironmentTaskResult({kind:"status"}, taskId, await callWorker("environmentTask", {
driverKey: metadata.driverKey, companyId: input.companyId, environmentId: environment?.id ?? null,
issueId: lease.issueId, config: driverConfig, taskId, runId: run?.id ?? null, agentId: run?.agentId ?? null,
projectIds, lease: {providerLeaseId: taskId, metadata}, operation:{kind:"status"},
}));
if (state.kind === "status" && state.executionStopped === true) {
await executeEnvironmentTask(db, workers, {...input, operation:{kind:"status"}});
await executeEnvironmentTask(db, workers, {...input, operation:{kind:"complete"}});
// Preserve the predecessor's stop proof for checkpoint authorization.
if (typeof metadata.runnerTaskPreviousLeaseId === "string") await db.update(environmentLeases).set({
metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || ${JSON.stringify({runnerTaskExecutionStopped:true,
runnerTaskSessionRetained:state.sessionStateAvailable === true})}::jsonb`,
}).where(and(eq(environmentLeases.id, metadata.runnerTaskPreviousLeaseId), eq(environmentLeases.companyId, input.companyId)));
const replacementDriverKey = metadata.driverKey;
const creationWorker = context && driver.creationOperationVersion === 1
? workers.getWorker(plugin.id) : undefined;
const recoverableCreation = creationWorker?.supportedMethods.includes("environmentInspectLease") === true &&
metadata.providerMetadata && typeof metadata.providerMetadata === "object" && !Array.isArray(metadata.providerMetadata);
let abandoned = false;
const fresh = await createOwnedSandbox(db, {
companyId: input.companyId, environmentId: environment!.id, heartbeatRunId: lease.heartbeatRunId,
issueId: lease.issueId, executionWorkspaceId: lease.executionWorkspaceId, provider: lease.provider!,
pluginId: plugin.id, driverKey: metadata.driverKey, config: {}, requestKey: taskId,
metadata: { driver: "plugin", pluginId: plugin.id, pluginKey: plugin.pluginKey, driverKey: metadata.driverKey,
runnerTaskLease: true, agentId: run!.agentId,
...(context ? { runnerTaskCreationContextId: context.id } : {}),
...(recoverableCreation ? { creationOperationVersion: 1, providerMetadata: metadata.providerMetadata } : {}) },
}, async identity => {
const params = { driverKey: replacementDriverKey, companyId: input.companyId, environmentId: environment!.id,
issueId: lease.issueId, config: driverConfig, runId: identity.runId, creationOperationId: identity.id,
agentId: run!.agentId, adapterType: "paperclip_runner",
...(typeof metadata.runnerTaskRequestedExpiresAt === "string" ? { requestedExpiresAt: metadata.runnerTaskRequestedExpiresAt } : {}) };
if (!context) return workers.call(plugin.id, "environmentAcquireLease", params, 30_000);
const settled = context.trackWorker();
let observedResource: Parameters<typeof queueSandboxCreationCleanup>[1] | undefined;
let dispatched = false;
try {
return await context.wait(() => {
dispatched = true;
const observe = async (result: HostToWorkerMethods["environmentAcquireLease"][1]) => {
if (!result.providerLeaseId || result.providerLeaseId === taskId) throw new Error("Plugin replacement task identity unavailable");
const resource = { id: identity.id, companyId: input.companyId, environmentId: environment!.id,
providerLeaseId: result.providerLeaseId, providerMetadata: result.metadata };
await recordSandboxCreationResource(db, resource);
observedResource = resource;
if (abandoned || !context.active || context.signal.aborted || Date.now() >= context.deadlineAtMs) {
await queueSandboxCreationCleanup(db, resource);
}
};
return recoverableCreation ? creationWorker!.call("environmentAcquireLease", params, 30_000, undefined, settled, context.id, observe)
: workers.call(plugin.id, "environmentAcquireLease", params, 30_000, undefined, settled, context.id, observe);
});
} catch (error) {
if (!dispatched) settled();
abandoned = true;
if (observedResource) await queueSandboxCreationCleanup(db, observedResource);
throw error;
}
});
if (!fresh.providerLeaseId || fresh.providerLeaseId === taskId) throw new Error("Plugin replacement task identity unavailable");
const {runnerTaskIdentity:_identity,runnerTaskPriorIdentity:_prior,runnerTaskCredentialHandoff:_handoff,
runnerTaskResumeFrom:_resume,runnerTaskPreviousLeaseId:_previous,...rest}=metadata;
const replacementMetadata={...rest,providerMetadata:fresh.metadata ?? {},runnerTaskReused:false,runnerTaskDispatched:false,
runnerTaskExecutionStopped:false,runnerTaskSessionRetained:false};
try {
context?.assertOpen();
if (!context) await recordSandboxCreationResource(db, { id: String(fresh.metadata!.creationOperationId), companyId: input.companyId,
environmentId: environment!.id, providerLeaseId: fresh.providerLeaseId });
const expiresAt=fresh.expiresAt ? new Date(fresh.expiresAt) : null;
const deadline=typeof metadata.runnerTaskRequestedExpiresAt === "string" ? Date.parse(metadata.runnerTaskRequestedExpiresAt) : null;
if ((expiresAt && (!Number.isFinite(expiresAt.getTime()) || expiresAt.getTime() <= Date.now())) ||
(deadline !== null && (!Number.isFinite(deadline) || !expiresAt || expiresAt.getTime() > deadline)))
throw new Error("Plugin replacement task expiry is not bounded");
await attachSandboxCreation(db, String(fresh.metadata!.creationOperationId), { companyId: input.companyId, environmentId: environment!.id, heartbeatRunId: lease.heartbeatRunId }, async tx => {
context?.assertOpen();
const [updated] = await tx.update(environmentLeases).set({providerLeaseId:fresh.providerLeaseId,expiresAt,
metadata:{...replacementMetadata,creationOperationId:fresh.metadata!.creationOperationId},
}).where(and(eq(environmentLeases.id,lease.id),eq(environmentLeases.providerLeaseId,taskId),
eq(environmentLeases.status,"active"), sql`coalesce(${environmentLeases.metadata}->>'runnerTaskRetiring', 'false') != 'true'`)).returning({id:environmentLeases.id});
if (!updated) throw new Error("Plugin task replacement ownership changed");
const attached = (await environmentService(tx).getLeaseById(updated.id))!;
context?.assertOpen();
return attached;
});
} catch (error) {
if (context) await queueSandboxCreationCleanup(db, { id: String(fresh.metadata!.creationOperationId), companyId: input.companyId,
environmentId: environment!.id, providerLeaseId: fresh.providerLeaseId, providerMetadata: fresh.metadata });
throw error;
}
return executeEnvironmentTask(db, workers, input);
}
}
if (operation.kind === "complete") {
await settlePluginTaskCredentialHandoff(db, { companyId: input.companyId, leaseId: lease.id }, metadata, async accountIdentity => {
const snapshot = await executeEnvironmentTask(db, workers, { ...input, operation: { kind: "credentials", accountIdentity } });
if (snapshot.kind !== "credentials") throw new Error("Plugin Runner credential snapshot unavailable");
return snapshot.credential;
});
}
const executionContext = context?.pluginSupported && ["describe", "prepare", "check_project_auth", "connection"].includes(operation.kind) ? await context.projection({ pluginId: plugin.id,
leaseId: lease.id, taskId, environmentId: environment!.id }, async () => {
try {
const [identity, configuration] = await Promise.all([
pluginAgentRuntimeIdentity(db, plugin.id, input.companyId, run!.agentId),
pluginRegistryService(db).getConfig(plugin.id, input.companyId),
]);
if (!identity) throw new Error("Missing transport binding");
return { instanceSecrets: identity.instanceSecrets, configuration: configuration?.configJson ?? {} };
} catch { throw new Error("Plugin task transport unavailable"); }
}) : undefined;
const params = {
driverKey: metadata.driverKey, companyId: input.companyId, environmentId: environment?.id ?? null,
issueId: lease.issueId, config: driverConfig,
taskId, runId: run?.id ?? (activeProbe ? String(probe.runId) : null), agentId: run?.agentId ?? null, projectIds,
lease: { providerLeaseId: lease.providerLeaseId, metadata: lease.metadata ?? undefined },
...(workspaceBinding ? { workspace: workspaceBinding } : {}),
...(freshWorkspace && scopesProjects ? { fresh: freshWorkspace } : {}), operation,
executionContextSupported: 1 as const,
...(executionContext ? { executionContext } : {}),
};
const timeout = operation.kind === "copilot_metadata" ? Math.max(1, Math.min(40_000, Number(probe?.expiresAtMs) - Date.now()))
: executionContext ? Math.max(1, context!.deadlineAtMs - Date.now()) : 15_000;
const result = await callWorker("environmentTask", params, timeout);
if (executionContext) context!.assertOpen();
const parsed = parseEnvironmentTaskResult(operation, taskId, result);
if (operation.kind === "read_file") {
const [current] = await db.select({ lease: environmentLeases, run: heartbeatRuns }).from(environmentLeases)
.leftJoin(heartbeatRuns, and(eq(heartbeatRuns.id, environmentLeases.heartbeatRunId), eq(heartbeatRuns.companyId, environmentLeases.companyId)))
.where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, input.companyId))).limit(1);
if (!current || current.lease.status !== "active" || current.run?.status !== "running" ||
current.lease.metadata?.runnerTaskRetiring === true || current.lease.metadata?.runnerTaskExecutionStopped === true ||
current.lease.expiresAt && current.lease.expiresAt.getTime() <= Date.now()) throw new Error("Plugin Runner file lease expired during transfer");
}
if (freshWorkspace && parsed.kind === "description" && parsed.freshWorkspaceVersion !== 1)
throw new Error("Plugin-provided fresh workspace preparation unavailable");
if (operation.kind === "prepare" && parsed.kind === "prepared" && parsed.projectAuthCheckAtSubmit) {
const harness = operation.requirements.provider === "acpx" ? operation.requirements.settings.acpxAgent : operation.requirements.provider;
if (!harness) throw new Error("Plugin Runner prepared provider unavailable");
const provider = harness === "codex" ? "openai" : harness === "claude" ? "anthropic" : harness;
const saved = await db.update(environmentLeases).set({ metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || ${JSON.stringify({
runnerTaskProjectAuthValidation: { configurationId: parsed.launch.configurationId, provider, projectIds, atSubmit: true },
})}::jsonb` }).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, input.companyId),
eq(environmentLeases.status, "active"), sql`coalesce(${environmentLeases.metadata}->>'runnerTaskRetiring', 'false') != 'true'`)).returning({ id: environmentLeases.id });
if (saved.length !== 1) throw new Error("Plugin Runner project validation lease expired");
}
if (operation.kind === "check_project_auth" && parsed.kind === "project_auth") {
if (parsed.configurationId !== operation.configurationId || parsed.provider !== operation.provider) throw new Error("Plugin Runner project validation identity mismatch");
if (parsed.allowed) {
const saved = await db.update(environmentLeases).set({ metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || ${JSON.stringify({
runnerTaskProjectAuthValidation: { configurationId: parsed.configurationId, provider: parsed.provider, projectIds },
})}::jsonb` }).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, input.companyId),
eq(environmentLeases.status, "active"), sql`coalesce(${environmentLeases.metadata}->>'runnerTaskRetiring', 'false') != 'true'`)).returning({ id: environmentLeases.id });
if (saved.length !== 1) throw new Error("Plugin Runner project validation lease expired");
}
}
if ((parsed.kind === "status" || parsed.kind === "copilot_metadata") && parsed.executionStopped === true) {
await db.update(environmentLeases).set({
metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || ${JSON.stringify({ runnerTaskExecutionStopped: true,
runnerTaskSessionRetained: parsed.kind === "status" && parsed.sessionStateAvailable === true,
...(parsed.kind === "status" && parsed.preparationStopped === true ? { runnerTaskPreparationStopped: true } : {}) })}::jsonb`,
}).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, input.companyId)));
}
if (operation.kind === "complete") {
if (freshWorkspace && metadata.runnerTaskExecutionStopped === true)
await releaseProviderRecoveryTransfers(input.companyId, freshWorkspace);
// A physical task can back several run leases. Confirm retirement for every prior owner.
if (lease.leasePolicy === "reuse_by_environment" && metadata.runnerTaskExecutionStopped === true) {
const owners=await db.select().from(environmentLeases).where(and(eq(environmentLeases.companyId,input.companyId),
eq(environmentLeases.provider,lease.provider!),eq(environmentLeases.providerLeaseId,taskId)));
for(const owner of owners) {
const receipt=remoteTerminationReceipt(owner,{providerLeaseId:taskId,state:"stopped"});
if(!receipt) continue;
await writeEnvironmentCleanup(db, tx => tx.update(environmentLeases).set({
...(owner.status === "retained" ? {status:"released" as const,releasedAt:new Date(),cleanupStatus:"success" as const}:{}),
metadata:sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || ${JSON.stringify({remoteExecutionTermination:receipt,runnerTaskExecutionStopped:true})}::jsonb`,
}).where(eq(environmentLeases.id,owner.id)), { companyId: owner.companyId, runId: owner.heartbeatRunId ?? undefined });
}
}
await db.update(environmentLeases).set({
metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || '{"runnerTaskDispatched":false}'::jsonb`,
}).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, input.companyId),
sql`${environmentLeases.metadata}->>'runnerTaskExecutionStopped' = 'true'`));
}
return parsed;
} catch {
// Worker errors and schema diagnostics can contain request credentials.
throw new Error("Plugin-provided Runner execution operation unavailable; reconcile the same task before retrying");
}
}
/** Failure leaves the dispatch marker and provider handle available to cleanup retries. */
export async function settleDispatchedEnvironmentTask(db: Db, workers: PluginWorkerManager, lease: { id: string; companyId: string }) {
const [current] = await db.update(environmentLeases).set({
metadata: sql`coalesce(${environmentLeases.metadata}, '{}'::jsonb) || '{"runnerTaskRetiring":true}'::jsonb`,
}).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, lease.companyId)))
.returning({ metadata: environmentLeases.metadata, heartbeatRunId: environmentLeases.heartbeatRunId });
const startup = current?.metadata?.runnerTaskStartup;
const startupCleanup = current?.metadata?.runnerTaskStartupCleanup as { requiresPreparationStop?: boolean } | undefined;
if (current?.metadata?.runnerTaskDispatched !== true && !startup && !startupCleanup) {
if (current?.metadata) await settlePluginTaskCredentialHandoff(db, { companyId: lease.companyId, leaseId: lease.id }, current.metadata,
async () => { throw new Error("Undispatched task must not request credentials"); });
return current?.metadata?.runnerTaskExecutionStopped === true;
}
const scope = { companyId: lease.companyId, leaseId: lease.id };
try {
const contextId = (startupCleanup as { contextId?: unknown } | undefined)?.contextId ?? (startup as { contextId?: unknown } | undefined)?.contextId;
const draining = typeof contextId === "string" ? workers.drainStartup?.(contextId) : undefined;
void draining?.catch(() => undefined);
await executeEnvironmentTask(db, workers, { ...scope, operation: { kind: "stop" } });
const state = await executeEnvironmentTask(db, workers, { ...scope, operation: { kind: "status" } });
if (state.kind !== "status" || state.executionStopped !== true || ((startup || startupCleanup?.requiresPreparationStop) && state.preparationStopped !== true)) throw new Error("Plugin Runner termination is unconfirmed");
await draining;
await executeEnvironmentTask(db, workers, { ...scope, operation: { kind: "complete" } });
return true;
} catch {
await writeEnvironmentCleanup(db, tx => tx.update(environmentLeases).set({ status: "pending_cleanup", cleanupStatus: "failed",
failureReason: "plugin_runner_task_cleanup_unconfirmed", releasedAt: new Date(), updatedAt: new Date(),
}).where(and(eq(environmentLeases.id, lease.id), eq(environmentLeases.companyId, lease.companyId))), { companyId: lease.companyId, runId: current.heartbeatRunId ?? undefined });
throw new Error("Plugin Runner termination is unconfirmed; the saved lease requires cleanup");
}
}
/** A saved provider task can resume only within the same authorized agent workspace. */
export async function resolvePluginTaskResume(db: Db, input: {
companyId: string; leaseId: string;
priorIdentity: { runnerInstanceId: string; environmentLeaseId: string; runId: string; normalizedSessionId: string };
/** Read-only discovery; launching still requires an attested retained checkpoint. */
inspectCheckpoint?: boolean;
}): Promise<string> {
const rows = await db.select({ lease: environmentLeases, run: heartbeatRuns }).from(environmentLeases)
.innerJoin(heartbeatRuns, and(eq(heartbeatRuns.id, environmentLeases.heartbeatRunId), eq(heartbeatRuns.companyId, input.companyId)))
.where(and(eq(environmentLeases.companyId, input.companyId),
or(eq(environmentLeases.id, input.leaseId), eq(heartbeatRuns.id, input.priorIdentity.runId))));
const next = rows.find(row => row.lease.id === input.leaseId);
// A retained PRP lease is logical; later provider tasks have different physical lease IDs.
const candidates = rows.filter(row => {
const identity = row.lease.metadata?.runnerTaskIdentity as Record<string, unknown> | undefined;
return row.lease.id !== input.leaseId && row.run.id === input.priorIdentity.runId &&
(input.inspectCheckpoint || row.lease.metadata?.runnerTaskSessionRetained === true) &&
identity?.runnerId === input.priorIdentity.runnerInstanceId && identity.leaseId === input.priorIdentity.environmentLeaseId &&
identity.sessionId === input.priorIdentity.normalizedSessionId && identity.runId === input.priorIdentity.runId &&
!rows.some(successor => successor.lease.id !== input.leaseId &&
successor.lease.metadata?.runnerTaskResumeFrom === row.lease.providerLeaseId);
});
const old = candidates.length === 1 ? candidates[0] : undefined;
const identity = old?.lease.metadata?.runnerTaskIdentity as Record<string, unknown> | undefined;
if (!old || !next || !old.lease.providerLeaseId || !identity ||
identity.runnerId !== input.priorIdentity.runnerInstanceId || identity.leaseId !== input.priorIdentity.environmentLeaseId ||
identity.sessionId !== input.priorIdentity.normalizedSessionId || identity.runId !== input.priorIdentity.runId ||
old.lease.metadata?.driver !== "plugin" || next.lease.metadata?.driver !== "plugin" ||
!next.lease.providerLeaseId || next.lease.providerLeaseId === old.lease.providerLeaseId ||
(old.run.id !== next.run.id && !["succeeded", "interrupted", "failed", "cancelled", "timed_out"].includes(old.run.status)) ||
old.run.agentId !== next.run.agentId || old.lease.issueId !== next.lease.issueId ||
old.lease.executionWorkspaceId !== next.lease.executionWorkspaceId ||
!old.lease.environmentId || old.lease.environmentId !== next.lease.environmentId ||
!next.lease.provider?.startsWith("plugin:") || old.lease.provider !== next.lease.provider ||
old.lease.metadata?.pluginId !== next.lease.metadata?.pluginId || old.lease.metadata?.driverKey !== next.lease.metadata?.driverKey ||
next.lease.status !== "active" || next.lease.releasedAt || next.run.status !== "running" ||
(next.lease.expiresAt && next.lease.expiresAt.getTime() <= Date.now()) ||
!(old.run.id === next.run.id
? rows.filter(row => row.lease.id !== next.lease.id).every(row => hasRemoteTerminationReceipt(row.lease))
: await remoteExecutionHasStopped(db, input.companyId, old.run.id))) {
throw new Error("Plugin Runner retained session is not authorized");
}
return old.lease.providerLeaseId;
}