Files
PaperClipAI/tests/runner-e2e/instruction-persistence.ts
DottaandPaperclip a6306ba606 feat(runner): consolidate Cursor production integration (#15075)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - The native Runner keeps provider sessions under company authority,
approvals, budgets and durable recovery.
> - Cursor work was spread across candidate branches. The published
branch lacked later plan, permission and cleanup fixes.
> - Production also needs public installation and matching runtime
assets for local and Daytona execution.
> - This pull request consolidates Cursor onto current mainline recovery
behavior and completes that installation path.
> - The installed v11 release passed focused local and Daytona
qualification after the generic mode and lifecycle cleanup. The later
model-selection correction and current mainline merge produce v14
artifacts that need matching release qualification.
> - Cursor admission is enabled in source; publish only an artifact
combination with matching qualification. Native AskQuestion and complete
per-run dollar accounting remain excluded.

## Linked Issues or Issue Description

Refs: #14435, #14631, #14669, #14699, #14724.

This completes the Cursor implementation by @cryppadotta from combined
source `22c78242a4e0c2369fecf0c2dc4e7600fbad6706`. It preserves newer
mainline recovery, completion and warm-directory behavior. Pi and
Copilot remain gated.

## What Changed

- Generate named Rust and TypeScript ACPX release profiles from one
manifest. Share runtime pins with packaging and server verification.
Preserve vendor runtime versions; bind the updated ACPX patch to Cursor
profile v14 and reject stale generated declarations at build/typecheck.
- Remove ACPX model allowlists, including the former Codex and Pi
restrictions and the duplicate developer test-drive gate. Send any
explicit model ID unchanged to its provider and verify the effective
selection before prompting. The bundled ACPX package forwards unlisted
IDs, rejects mismatched acknowledgements, and restores the exact
selection after session load. It does not expand Cursor model aliases.
Provider rejection, mismatch, or missing model controls fails without a
fallback. Model examples live in evaluation fixtures, outside runtime
declarations.

- Add pinned Cursor execution, contained instructions, exact model
verification and Agent/Plan/Ask modes.
- Carry an opaque generic `mode` identifier in shared native execution,
sidecar, Rust and recovery contracts. The provider adapter owns
supported modes, defaults, native translation and acknowledgement.
- Keep native RPC recognition, accepted-plan interpretation and
permission evidence behind provider adapters. Shared settlement and
recovery verify normalized facts and their committed evidence.
- Replace the Cursor-only warm-attachment branch with a runner-owned
capability. Only Cursor opts into it. Move profile compatibility and
optional usage parsing into provider metadata and adapters.
- Write generic plan-wait receipts. Read exact historical Cursor
receipts through a separate compatibility decoder. Reject mixed formats
and preserve existing authority checks.
- Carry native plans, semantic questions, todos, child activity,
permission identities and partial usage diagnostics through the Runner.
- Preserve durable response delivery, cancellation, warm ownership and
process retirement.
- Finish accepted planning runs successfully. Keep their tasks open for
explicit direction. Acceptance does not start implementation.
- Ship `paperclipai runtime setup cursor` and its provisioner through
the public package. npm installation does not download Cursor. Setup
uses the OS account's closure-keyed cache so system-wide npm packages
can remain read-only. Run it as the Paperclip service account.
- Include Cursor in normal provider packs and Daytona images for macOS
ARM64/x64 and Linux x64.
- Reject stale release packs by source revision and current ACPX/Cursor
pins before assembly writes files. Verify current Cursor
version/profile/closure again at runtime.
- Ship all three daemon targets and the expected Linux image-pack
identity. A macOS controller uses its packaged Linux daemon for Daytona.
Image mismatches fail before provider launch.
- Use the vendored Runner boundary for installed readiness probes.
Verify the actual installed Cursor probe.
- Verify compiled public Daytona plugins and their release versions in
installed smokes.
- Record exact artifacts, the acceptance matrix, retained failures,
supported capabilities and rollback behavior in the [readiness
report](https://github.com/paperclipai/paperclip/blob/codex/cursor-production-readiness/doc/plans/2026-10-03-cursor-production-readiness.md).

## Verification

- Current head `f7ec5cc1f0e30c62a829c016ff2013a2a9d79037` merges
mainline `faa8e452c73bae5e044dd6379179a00106abb131`. It keeps Cursor
plan and cancellation guards alongside mainline historical-question
filtering. The evaluation catalog includes both Cursor and expanded
adapter accounting cases (683 total). Recursive typecheck, full build,
696 lifecycle/recovery tests, 45 fixture tests and fixture typecheck
passed. Current-head CI passed: 56 successful checks, one neutral and
four skipped. [Complete
CI](https://github.com/paperclipai/paperclip/actions/runs/37557996535).
The fresh Base Greptile review is 5/5 on this exact head, with 304 files
reviewed, zero new comments and zero unresolved threads. The user
authorized overriding the CODEOWNER review gate after checks passed; no
failing checks are overridden. Prior results below retain their own head
identities.
- Corrective head `3d2b168366258036f6b6a6fccb382c49138cc601` fixes the
post-merge Apex finding. Automatic-review and new-evidence
reconciliation preserve pending child results and recheck delivery under
the status lock before completing. Account repair now excludes unrelated
secret consumers and requires the failed agent's identity. Regression
coverage includes the commit race, delivery statuses,
current-run/current-intent exclusions, repeated reconciliation, both
database reconciliation paths, and credential consumer boundaries. All
184 affected tests, server typecheck and server build passed.
Current-head Base Greptile review is 5/5, with 304 files reviewed, zero
new comments and zero unresolved threads. Current-head CI passed: 56
successful checks, one neutral and four skipped. [Complete
CI](https://github.com/paperclipai/paperclip/actions/runs/37535994724).
This Base review is distinct from the earlier Apex review.
- Merge head `5957c257a` reconciles mainline `b508a05c4`. It preserves
both accepted-plan waits and pending-child-completion checks, current
provider selectors, task-creation response identities, and mainline ACPX
missing-file handling. The combined patch is bound to Cursor profile
v14; historical records keep their original identities.
- Merge head `5957c257a` passed recursive typecheck, full build, 43
installed ACPX/package contracts, 107 provider UI and plan/recovery
tests, 593 database-backed lifecycle tests, 49 profile/native contract
tests, 45 Product E2E fixture tests, fixture typecheck, token gates,
three provider-free browser task-creation cases, and Runner
conformance/replay checks. Its complete CI passed (55 successful checks,
one neutral and four skipped), while Apex returned 2/5 with a
child-delivery finding addressed below.
- The local full-suite attempt again failed the unchanged Git streaming
test (360-second timeout) and was stopped. The concurrent local Rust
attempt failed four unchanged Codex process/deadline tests; all four
passed serially without code changes in 7.29 seconds after removing the
competing test load. These failed commands are retained and are not
reported as full-suite passes; the fresh Linux CI runs are tracked
separately.
- The previous head `907bdb2a2778c7ffeb4a662a91460c9d1ddfc9c5` earned
Apex 5/5 with zero comments after fixing all three findings: per-user
install cache, stale release-pack rejection, and public Linux smoke
account/home handling. Its real built installer passed from read-only
public packages on macOS ARM64 and Linux x64. All 137 release-registry
checks and 64 ACPX package contracts passed. That review does not cover
this mainline reconciliation.
- Prior `beadd3654` passed the full CI matrix; its one unchanged chat
test failure and successful single retry remain in the [CI
history](https://github.com/paperclipai/paperclip/actions/runs/37521449327).
Historical results below remain attributed to their original builds.

- Fixture follow-up `dd59d7e82b103a88b7cbd7d2c38b612c0fbbff7a` removes
provider-specific model choices from generic offline ACPX tests. The
fake sidecar preserves the model and session identity selected at open
through suspension. Affected verification passed: 106 Rust tests and 73
TypeScript tests. This commit changes test code only; the
production-code checks below retain their recorded identities. Its CI
and Greptile review later passed; those results belong to that
historical head.
- Model-selection cleanup `9a070808b48960a41fdfd369ae0636b95af82459`:
252 focused Runner tests passed (six platform skips), covering all six
ACPX agents, native model acknowledgement, rejected selections,
installation integrity and recovery identity. The merged branch passed
recursive typecheck, full build, token gates, server admission (19
tests), and the Product E2E catalog (45 tests). The acceptance catalog
passed all four tests. The full Rust suite passed: 643 tests, 2 ignored.
It verifies sidecar acknowledgement of unlisted models and rejection of
model mismatches. The final commits only update Rust tests; production
sources match the verified build at
`65ec3279ac50185e3cda109b5cfd9b4f56105de0`. No new paid provider calls
were made.
- The merge preserves both Cursor and the new mainline public-MCP
fixture cases. Auto-merge remains disabled; the latest follow-up status
is recorded above. The local `pnpm test:run` attempt hit the unchanged
Git streaming test's 300-second timeout and was interrupted before
merging mainline. The broad Runner attempt found obsolete single-model
assertions plus three macOS fixture-path failures caused by a
`/private/tmp` override. The assertions are corrected; affected
TypeScript checks passed with the standard macOS temporary directory,
and the complete Rust suite passed. Neither interrupted command is a
full-suite pass.
- Earlier declaration-cleanup head `6f4a5e9e2` passed recursive
typecheck, build, Rust and focused tests. Its CI later exposed a test
expecting duplicated Grok digest literals. The current source fixes that
assertion to compare launcher bytes with the shared manifest. Historical
successes and failed attempts are retained; no new live provider
qualification is claimed.
- Previous head `e75fde6098b0ddd8cec765bfb6ecaeecb88a26a6` passed
complete CI (56 successful checks, one neutral, four skipped) and
Greptile 5/5. [Historical complete
CI](https://github.com/paperclipai/paperclip/actions/runs/37489112305).
Those results are not claimed for the cleanup head.
- Frozen live application: `d7b696f9b8f79095233e9e3d56d23e6a6018dd48`.
Public package version: `0.0.0-cursor-verify.3d0c9b7761c6`. The
declaration cleanup preserves release pins and does not relabel that
tested artifact as a build of the new source. Mainline through
`e34abee670` was reconciled while preserving accepted-plan waits,
provider-capacity handling, and both Cursor and public-MCP fixtures.
- Clean normal installation, explicit Cursor setup and daemon resolution
passed on macOS ARM64, macOS x64 under Rosetta, and Linux x64. npm
lifecycle hooks ran without silently downloading Cursor.
- Historical v11 live matrix: **18/18 passed with cleanup** (nine local,
nine Daytona) after the generic mode and lifecycle cleanup. The campaign
has 23 attempts; all five failures and their diagnoses remain recorded.
Exact case identities, hashes and limits are in the readiness report.
All provider calls are real, use the explicit Luna model and
company-bound credentials, and run without qualification or
runtime-asset overrides.
- The immutable Daytona image is
`ghcr.io/paperclipai/paperclip-daytona-runner@sha256:d6259b6bba094702c13fc2283bd85550849c1c53145b656fb2746778f9fa1747`.
The public Daytona plugin is installed independently and its version is
checked.
- Recursive typecheck, full build, token gates and Runner
contract/conformance/replay checks passed on the frozen application. Its
complete Linux CI suite passed. The duplicate local full-suite command
was incomplete after timing failures; affected repeats passed, but that
command is not reported as a clean pass.
- Qualification fixtures passed typecheck, 1,675 Vitest tests (one
skip), 128 Node checks, three provider-free browser tests, and 150
focused lifecycle tests after the final diagnostic correction. The
affected legacy Cursor command file also passed all five tests after
removing its shorter 10-second override; it now inherits the suite’s
standard 15-second timeout. Greptile is 5/5 on `e75fde609` with no
unresolved review threads. CI results above are recorded separately from
historical build results.

## Risks

- Cursor v14 includes the updated ACPX dependency patch and release
identity. The v11 live matrix and image below remain historical
evidence. They do not certify new v14 package/image artifacts.

- ACPX accepts models beyond the qualification fixtures. Availability
and entitlement depend on the provider. Successful configuration is not
a claim of live qualification for every model.
- Shared mode is an opaque identifier. Provider adapters own its
meaning. Incompatible historical sessions remain fenced; exact committed
plan waits and task history remain inspectable.
- Native AskQuestion is excluded. Paperclip semantic questions are
supported. Authoritative per-run dollar accounting is unavailable;
partial counters remain diagnostics and unknown cost is not zero.
- Image input, detailed native diffs, deeper child transcripts and
native plan-file export remain follow-ups.
- macOS x64 has clean-install and daemon-startup proof under Rosetta,
not a separate live campaign on Intel hardware.
- Release only the tested package/image combination. Merging this PR
does not publish npm packages or deploy that image. Later builds need
their own release verification. Rollback disables new Cursor admission
while preserving records and recovery inspection.
- A model can fail an exact instruction: one cancelled-plan attempt
returned the wrong summary marker despite correct cancellation. The
unchanged repeat passed; both results remain in the report.

> ROADMAP.md was checked. This completes existing native Runner/Cursor
work; it does not add an independent core feature proposal.

## Model Used

OpenAI Codex, GPT-6. The exact serving variant and context window are
not exposed in this session. The agent used reasoning, repository
inspection, code execution, protocol tests and browser-backed Product
E2E tools. Cursor acceptance uses the explicit
`gpt-5.6-luna[context=272k,reasoning=medium,fast=false]` model. That is
the evaluated provider model.

## 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 — affected suites passed;
full CI and the retained local failed attempts are recorded separately
above.
- [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 — 56 successful checks, one
neutral and four skipped on `f7ec5cc1f0e30c62a829c016ff2013a2a9d79037`
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups —
fresh Base review passed on `f7ec5cc1f0e30c62a829c016ff2013a2a9d79037`;
zero new comments and no unresolved threads. The earlier Apex finding
remains fixed.
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-10-06 20:48:15 -05:00

233 lines
20 KiB
TypeScript

import { createHash, randomBytes } from "node:crypto";
import { expect, type Page } from "@playwright/test";
import { pollUntil, type RunnerApi } from "./api.js";
import { captureFirstTaskAttachments } from "./first-task-attachments.js";
import { collectRunEvents } from "./run-observations.js";
import { createTaskThroughUi } from "./user-actions.js";
import type { LiveFixtureValues } from "./live-fixtures.js";
import type { MatrixExecution, RunnerTaskFixture } from "./types.js";
type Row = Record<string, any>;
export const instructionNonceLine = (nonce: string) => `Instruction persistence nonce: ${nonce}\n`;
export const instructionPersistenceTask: RunnerTaskFixture = {
id: "private-copy-persists", label: "Agent directory survives a fresh task",
groups: [], workMode: "standard", flow: "instruction_persistence",
expectedRunCount: 6, attemptTimeoutMs: { local: 20 * 60_000, daytona: 20 * 60_000 },
expectedTerminalState: { issue: "done", run: "succeeded" },
buildTitle: nonce => `Persist private instructions ${nonce}`,
buildVisibleMarker: () => "INSTRUCTIONS-VERIFIED",
buildPrompt: nonce => [
"Edit your own registered writable agent instruction entry with ordinary filesystem tools. The runtime guidance gives its exact private path.",
"Use Node.js built-in fs for these byte-preserving edits. Apply each append exactly once: inspect the existing suffix before retrying any command, because a warning does not imply that its writes failed.",
`Preserve its existing bytes and append exactly this UTF-8 suffix, represented as a JSON string: ${JSON.stringify(`\n${instructionNonceLine(nonce)}`)}`,
"Decode the JSON string once and append those bytes. Do not trim or normalize the existing file and do not add another blank line or separator.",
`In AGENT_HOME, create notes/retained.txt containing exactly ${JSON.stringify(`Personal file nonce: ${nonce}\n`)}. Create notes/bytes.bin with exactly the bytes [0,255,17,128,9]. Read notes/from-editor.txt and append exactly a newline followed by Edited by agent. and a final newline.`,
"Do not use update_agent_instructions, restore_agent_instructions, or an instructions API to save it. Do not edit repository AGENTS.md or the read-only loaded bundle.",
"After verifying the edits, upload a small text/plain attachment named agent-file-check.txt containing only 'Private file edits verified'. Use this attachment as your task completion evidence; the personal files themselves stay in AGENT_HOME.",
"Reply only Instruction copy edited without printing filesystem paths, then complete this task after the file edit. Paperclip will collect it after the provider stops; do not claim it has already persisted. Do not create further tasks.",
].join("\n"),
buildMatchers: () => [], // Independent current file and attachment oracle below.
};
export function gradeInstructionPersistence(input: { before: Row; after: Row; firstRunId: string; expectedContent: string; proof: Row | undefined; expectedProof: string; saveEvent?: Row; fileProof?: boolean }) {
return [
{ id: "directory-save-receipt", passed: input.saveEvent?.runId === input.firstRunId && input.saveEvent?.state === "saved" },
{ id: "nested-and-binary-files", passed: input.fileProof === true },
{ id: "exact-canonical-bytes", passed: input.after.content === input.expectedContent && input.after.contentHash === createHash("sha256").update(input.expectedContent).digest("hex") },
{ id: "fresh-task-downloaded-proof", passed: input.proof?.contentVerified === true && input.proof.body === input.expectedProof },
].map(check => ({ ...check, detail: check.passed ? `${check.id} verified independently` : `${check.id} missing or incorrect` }));
}
export async function runInstructionPersistenceFlow(input: {
page: Page; api: RunnerApi; fixtures: LiveFixtureValues; execution: MatrixExecution; nonce: string;
secrets: readonly string[]; deadlineAt: number;
restart(): Promise<void>;
observe(issue: Row, runs: Row[]): void;
capture(id: string, label: string, file: string): Promise<void>;
evidence(name: string, data: unknown): Promise<void>;
}) {
const { page, api, fixtures, execution, nonce } = input;
// Fixture names contain the campaign nonce. Use an unrelated value that the
// fresh task can obtain only from the saved entry (or forbidden task history).
const persistedNonce = randomBytes(16).toString("hex");
const filePath = `/api/agents/${fixtures.agent.id}/instructions-bundle/file?path=AGENTS.md`;
const before = await api.get<Row>(filePath);
if (typeof before.content !== "string" || !before.contentHash) throw new Error("Managed instructions must expose a current file hash");
const expectedContent = `${before.content}\n${instructionNonceLine(persistedNonce)}`;
const instructionsUrl = `/${fixtures.company.issuePrefix}/agents/${fixtures.agent.id}/instructions`;
const editorText = `Editor nonce: ${randomBytes(16).toString("hex")}`;
await page.goto(instructionsUrl);
await page.getByRole("button", { name: "Add agent file", exact: true }).click();
await page.getByPlaceholder("TOOLS.md").fill("notes/from-editor.txt");
await page.getByRole("button", { name: "Create", exact: true }).click();
await page.getByRole("group", { name: "Instruction file view" }).getByRole("button", { name: "edit", exact: true }).click();
await page.getByRole("textbox", { name: "Instruction file editor" }).fill(editorText);
await page.getByRole("button", { name: "Save changes", exact: true }).click();
await expect(page.getByRole("button", { name: "Save changes", exact: true })).toBeDisabled();
await expect(page.getByRole("button", { name: "History", exact: true })).toHaveCount(0);
let issue: Row = {};
let runs: Row[] = [];
async function create(title: string, prompt: string) {
const createdTask = await createTaskThroughUi({ page, issuePrefix: fixtures.company.issuePrefix!, agentName: fixtures.agent.name, title, prompt, workMode: "standard", projectName: fixtures.project?.name });
const found = await pollUntil({ label: `instruction task ${title}`, deadlineAt: input.deadlineAt,
load: async () => (await api.get<Row[]>(`/api/companies/${fixtures.company.id}/issues?limit=100`)).find(row => row.id === createdTask.issueId), accept: row => Boolean(row) });
if (!found) throw new Error("Browser-created instruction task missing");
issue = found;
input.observe(issue, runs);
await page.goto(`/${fixtures.company.issuePrefix}/issues/${issue.identifier ?? issue.id}`);
}
async function settle(count: number) {
await pollUntil({ label: `instruction run ${count} completed`, deadlineAt: input.deadlineAt,
load: async () => {
issue = await api.get<Row>(`/api/issues/${issue.id}`);
const listed = await api.get<Row[]>(`/api/companies/${fixtures.company.id}/heartbeat-runs?limit=100`);
runs = await Promise.all(listed.map(row => api.get<Row>(`/api/heartbeat-runs/${row.id}`)));
runs.sort((a, b) => String(a.createdAt).localeCompare(String(b.createdAt)));
input.observe(issue, runs);
return { issue, runs };
},
accept: state => state.issue.status === "done" && state.runs.length === count && state.runs.every(row => row.status === "succeeded"),
reject: state => state.runs.some(row => ["failed", "cancelled", "timed_out"].includes(row.status)) ? "Instruction task provider run failed" : state.runs.length > count ? "Instruction task dispatched an extra run" : state.issue.status === "blocked" && state.runs.length === count && state.runs.every(row => row.status === "succeeded") ? `Instruction task reported a terminal blocker: ${JSON.stringify(state.issue.unblockDescriptor ?? {})}` : undefined,
});
expect(runs.every(row => row.runtimeMode === execution.profile.expectedRuntimeMode)).toBe(true);
await page.reload();
await expect(page.getByTestId("issue-detail-header").getByRole("button", { name: "Change status (current: Done)", exact: true })).toBeVisible();
}
await create(execution.task.buildTitle(nonce), execution.task.buildPrompt(persistedNonce));
await settle(1);
const firstRunId = runs[0]!.id;
const after = await pollUntil({ label: "stopped agent directory save", deadlineAt: Math.min(input.deadlineAt, Date.now() + 30_000),
load: () => api.get<Row>(filePath), accept: row => row.content === expectedContent });
const events = await collectRunEvents<Row>((afterSeq, limit) => api.get(`/api/heartbeat-runs/${firstRunId}/events?afterSeq=${afterSeq}&limit=${limit}`));
const saveEvent = events.find(row => row.eventType === "instruction_save" && row.payload?.state === "saved");
expect(saveEvent).toBeTruthy();
const readPersonal = (name: string) => api.get<Row>(`/api/agents/${fixtures.agent.id}/instructions-bundle/file?path=${encodeURIComponent(name)}`);
const note = await readPersonal("notes/retained.txt");
const fromEditor = await readPersonal("notes/from-editor.txt");
const binaryResponse = await api.request.get(`/api/agents/${fixtures.agent.id}/instructions-bundle/file?path=notes%2Fbytes.bin&download=true`);
expect(binaryResponse.ok()).toBe(true);
const binary = await binaryResponse.body();
const fileProof = note.content === `Personal file nonce: ${persistedNonce}\n` && fromEditor.content === `${editorText}\nEdited by agent.\n` && binary.equals(Buffer.from([0,255,17,128,9]));
expect(fileProof).toBe(true);
const history = await api.get<Row>(`/api/agents/${fixtures.agent.id}/instructions-bundle/history?path=AGENTS.md`);
expect(history.revisions).toHaveLength(0);
await input.evidence("instruction-first-save.json", { before, after, run: runs[0], events });
await input.capture("instruction-edited", "Private instructions saved after provider stop", "instruction-edited.png");
// A new server and a new issue cannot pass by retaining model conversation.
await input.restart();
expect((await api.get<Row>(filePath)).content).toBe(expectedContent);
await create("Read persisted instructions", [
"Read your own loaded agent instruction entry (or its current registered private copy) using ordinary filesystem tools.",
"Find the line beginning 'Instruction persistence nonce: '. Copy that entire line plus one final newline into instruction-proof.txt. Then append the exact bytes of notes/retained.txt from AGENT_HOME. Verify notes/bytes.bin contains the bytes [0,255,17,128,9]. Do not infer the value from this task title or other task history. Do not change your instructions.",
"Upload instruction-proof.txt as a text/plain task attachment named instruction-proof.txt using the normal artifact workflow. A local file alone is insufficient.",
`Reply with exactly ${execution.task.buildVisibleMarker(nonce)} and complete the task.`,
].join("\n"));
await settle(2);
const attachments = await captureFirstTaskAttachments(api, [{ ...issue, id: String(issue.id) }], input.secrets);
const proof = attachments.find(row => row.originalFilename === "instruction-proof.txt" || row.name === "instruction-proof.txt");
const final = await api.get<Row>(filePath);
expect(final.revision.id).toBe(after.revision.id);
const checks = gradeInstructionPersistence({ before, after, firstRunId, expectedContent, proof, expectedProof: `${instructionNonceLine(persistedNonce)}Personal file nonce: ${persistedNonce}\n`, saveEvent: { runId: firstRunId, state: saveEvent?.payload?.state }, fileProof });
await expect(page.getByTestId("task-chat-agent-bubble").filter({ hasText: execution.task.buildVisibleMarker(nonce) }).last()).toBeVisible();
await input.capture("final-state", "Fresh task downloaded the persisted instruction nonce", "final-state.png");
expect(checks.filter(check => !check.passed), "Independent instruction persistence checks").toEqual([]);
checks.push({ id: "editor-round-trip-no-history", passed: true, detail: "A browser-created supporting file was edited by the agent; all current bytes persisted without revision rows" });
const restored = final;
// The provider publishes an ordinary attachment before a bounded command wait.
// A board edit during that run is superseded only for the same changed file.
const conflictSuffix = `\nLast completed synchronization: ${nonce}\n`;
const expectedCandidate = `${restored.content}${conflictSuffix}`;
await create("Sync a concurrent instruction edit", [
"Use Node.js built-in fs. Apply the append exactly once, checking existing bytes before any retry.",
`Append exactly this UTF-8 suffix to your current registered writable instruction entry, represented as a JSON string: ${JSON.stringify(conflictSuffix)}`,
"Decode the JSON string once. Preserve all existing bytes. Do not use an instruction revision tool or instructions API.",
"After the file edit, upload a text/plain task attachment named instruction-candidate-ready.txt with the text ready. Use the ordinary artifact workflow.",
"Then execute the ordinary shell command sleep 45 and wait for it to finish. This gives the board time to edit the canonical instructions concurrently. Do not complete the task before that command finishes.",
"After the wait completes, reply Candidate edit ready and complete the task. Do not change the instructions again or claim that they saved.",
].join("\n"));
await pollUntil({ label: "provider staged concurrent instruction edit", deadlineAt: input.deadlineAt,
load: () => api.get<Row[]>(`/api/issues/${issue.id}/attachments`),
accept: rows => rows.some(row => row.originalFilename === "instruction-candidate-ready.txt" || row.name === "instruction-candidate-ready.txt") });
const active = await api.get<Row[]>(`/api/issues/${issue.id}/runs`);
expect(active.some(row => row.status === "running")).toBe(true);
const boardMarker = "Concurrent board instruction edit.";
await page.goto(instructionsUrl);
await page.getByText("AGENTS.md", { exact: true }).first().click();
await page.getByRole("group", { name: "Instruction file view" }).getByRole("button", { name: "edit", exact: true }).click();
const entryEditor = page.getByRole("textbox", { name: "editable markdown" });
await entryEditor.click();
await entryEditor.press("ControlOrMeta+End");
await entryEditor.press("Enter");
await entryEditor.pressSequentially(boardMarker);
await page.getByRole("button", { name: "Save changes", exact: true }).click();
await expect(page.getByRole("button", { name: "Save changes", exact: true })).toBeDisabled();
const board = await api.get<Row>(filePath);
expect(board.content).toContain(boardMarker);
expect(board.contentHash).not.toBe(restored.contentHash);
const unrelatedContent = `Concurrent independent file: ${nonce}`;
const unrelated = await api.request.put(`/api/agents/${fixtures.agent.id}/instructions-bundle/file`, {
data: { path: "notes/concurrent-editor.txt", content: unrelatedContent, baseHash: null },
});
expect(unrelated.ok()).toBe(true);
await page.goto(`/${fixtures.company.issuePrefix}/issues/${issue.identifier ?? issue.id}`);
await settle(3);
const syncRunId = runs[2]!.id;
const resolved = await pollUntil({ label: "last completed sync wins", deadlineAt: Math.min(input.deadlineAt, Date.now() + 30_000),
load: () => api.get<Row>(filePath), accept: row => row.content === expectedCandidate });
const candidates = await api.get<Row[]>(`/api/agents/${fixtures.agent.id}/instructions-bundle/candidates`);
expect(candidates.some(row => row.runId === syncRunId)).toBe(false);
expect((await readPersonal("notes/concurrent-editor.txt")).content).toBe(unrelatedContent);
const syncEvents = await collectRunEvents<Row>((afterSeq, limit) => api.get(`/api/heartbeat-runs/${syncRunId}/events?afterSeq=${afterSeq}&limit=${limit}`));
expect(syncEvents.some(row => row.eventType === "instruction_save" && row.payload?.state === "saved")).toBe(true);
await page.goto(instructionsUrl);
await expect(page.getByRole("button", { name: "Review preserved files", exact: true })).toHaveCount(0);
checks.push({ id: "per-file-last-sync-wins", passed: true, detail: "The later agent sync replaced the concurrent browser edit to its changed entry, preserved an unrelated new file, and created no conflict candidate" });
await page.goto(`/${fixtures.company.issuePrefix}/issues/${issue.identifier ?? issue.id}`);
await input.capture("last-sync-wins", "Concurrent changes synchronized per file without a conflict-review step", "last-sync-wins.png");
const quotaTask = (action: string, receipt: string) => [
"This is a controlled persistent-storage quota check. Use ordinary Node.js filesystem tools in your registered AGENT_HOME. Do not edit AGENTS.md.",
action,
`Upload a small text/plain task attachment named ${receipt}.txt containing the observed file size or cleanup result. This attachment is the primary task deliverable.`,
"Complete this task normally after uploading the receipt. A persistent-file storage warning is expected and must not prevent completion. Do not perform additional cleanup or change other personal files.",
].join("\n");
await create("Reach the agent file storage limit", quotaTask(
"Create quota-cache.bin using fs.openSync with flag w, fs.ftruncateSync(fd, 268435456), and fs.closeSync. This is a sparse fixture file, not a download. Verify its size using fs.statSync without reading the large contents.", "quota-full"));
await settle(4);
const fullRun = runs[3]!;
expect(fullRun.resultJson?.instructionSave).toMatchObject({ state: "saved", storageWarning: expect.stringContaining("Agent storage is full") });
await page.goto(`/${fixtures.company.issuePrefix}/agents/${fixtures.agent.id}/runs/${fullRun.id}`);
await expect(page.getByRole("note").filter({ hasText: "Agent storage warning" })).toContainText("Runs can continue");
// Public campaign screenshots are limited to sanitized task routes. Verify
// the warning in the real run UI, then capture its completed task outcome.
await page.goto(`/${fixtures.company.issuePrefix}/issues/${issue.identifier ?? issue.id}`);
await input.capture("storage-warning", "Task succeeded while agent storage reached its limit", "storage-warning.png");
await create("Keep running while agent storage is full", quotaTask(
"Verify quota-cache.bin already exists and its size is exactly 268435456. Grow only this file to 268435457 bytes with fs.truncateSync, then verify the new size. Leave it above the limit for this run's sync check.", "quota-exceeded"));
await settle(5);
const exceededRun = runs[4]!;
expect(exceededRun.resultJson?.instructionSave).toMatchObject({ state: "unavailable", errorCode: "AGENT_FILES_LIMIT_EXCEEDED", storageWarning: expect.stringContaining("Runs can continue") });
const fullEvents = await collectRunEvents<Row>((afterSeq, limit) => api.get(`/api/heartbeat-runs/${exceededRun.id}/events?afterSeq=${afterSeq}&limit=${limit}`));
expect(fullEvents.some(row => row.eventType === "instruction_save" && row.level === "warn" && row.payload?.state === "prepared" && row.payload?.storageWarning)).toBe(true);
await create("Clean up agent storage during a normal task", quotaTask(
"Verify restored quota-cache.bin has size 268435456: the rejected oversized edit must not have replaced its saved bytes. Delete quota-cache.bin with fs.unlinkSync, then write notes/after-quota.txt containing exactly 'Runs still work after quota cleanup'.", "quota-cleaned"));
await settle(6);
const cleanedRun = runs[5]!;
expect(cleanedRun.resultJson?.instructionSave).toMatchObject({ state: "saved", storageWarning: null });
expect((await readPersonal("notes/after-quota.txt")).content).toBe("Runs still work after quota cleanup");
const bundle = await api.get<Row>(`/api/agents/${fixtures.agent.id}/instructions-bundle`);
expect(bundle.files.some((file: Row) => file.path === "quota-cache.bin")).toBe(false);
await page.goto(`/${fixtures.company.issuePrefix}/agents/${fixtures.agent.id}/runs/${cleanedRun.id}`);
await expect(page.getByRole("note").filter({ hasText: "Agent storage warning" })).toHaveCount(0);
checks.push({ id: "storage-full-does-not-block-runs", passed: true, detail: "A run saved a file at quota, a subsequent run succeeded despite an oversized save rejection, and the next run removed the full file and cleared its warning; all three tasks completed" });
await input.evidence("api-state.json", { issue, runs, checks, canonicalInstructions: resolved, attachments });
await input.evidence("instruction-persistence.json", { checks, before, after, final, restored, board, candidates, resolved, syncEvents, runs, attachments, fullEvents });
await page.goto(`/${fixtures.company.issuePrefix}/issues/${issue.identifier ?? issue.id}`);
await input.capture("storage-recovered", "Agent completed a normal task and cleared storage warning after cleanup", "storage-recovered.png");
return { issue, runs, checks };
}