mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-24 01:41:42 +02:00
## Thinking Path > - Paperclip manages AI agents that perform work. > - Paperclip Runner connects durable task runs to local provider processes. > - The full-stack paid matrix exposed failures after the runner integrity repair. > - Verified JavaScript entrypoints lost their relative module graph when Linux executed them through descriptor paths. > - Returned provider startup errors also remained pending and became indeterminate after recovery. > - Sparse Codex tool lifecycle events lost the `write_document` identity before task transcript projection. > - This pull request repairs those three boundaries and makes the structured-question fixture deterministic. > - The benefit is repeatable provider startup, exact failure replay, and correct inline Plan placement. ## Linked Issues or Issue Description Refs #12721 and #12700. **What happened?** The paid runner matrix failed ACPX and OpenCode startup before provider session creation. The runner journal then replaced the original startup error with an indeterminate recovery result. Native Codex saved a Plan but rendered it only as a fallback card. A legacy Claude waiting reply could also echo the reserved terminal marker before the answer arrived. **Expected behavior** Verified JavaScript providers must start from immutable descriptor-backed artifacts. Returned startup failures must persist as terminal failed command results. Native tool lifecycle updates must preserve the `write_document` boundary. Pre-answer fixture output must not contain the reserved terminal marker. **Steps to reproduce** 1. Run the local provider cells in the Runner Full-Stack E2E workflow. 2. Observe ACPX and OpenCode fail during `session.open` before provider execution. 3. Observe recovery report `execution_indeterminate` instead of the original startup error. 4. Run the native Codex Plan cell and observe the fallback Plan card after the tool activity row. 5. Run the legacy Claude structured-question resume cell and observe an early marker echo in waiting prose. **Paperclip version or commit** `0f9452101740835ce0b1488a204bf48acd5bafc3` **Deployment mode** Local development with the paid GitHub Actions acceptance workflow. ## What Changed - Bundle the ACPX sidecar and OpenCode proxy as self-contained Node ESM entrypoints before hashing and verified descriptor launch. - Anchor ACPX dynamic provider package resolution at a controller-derived provider-pack root and keep that root out of the provider child environment. - Persist executor-returned startup errors as redacted durable failed command results while retaining indeterminate recovery for true process death. - Coalesce sparse native tool items by stable ID so a late `write_document` name, input, and result reach the transcript boundary once. - Forbid the structured-question fixture from spelling or announcing its reserved terminal marker before the user answers. ## Verification - Rust and TypeScript regression tests cover durable failed replay, true crash ambiguity, bundle closure, package-root derivation, environment filtering, exact Codex tool lifecycle coalescing, and prompt determinism. - Local execution is intentionally limited to formatters and static diff checks. GitHub Actions will run tests, type checks, builds, and security checks. - After ordinary CI is green, scoped paid cells will validate one ACPX launch, one OpenCode launch, native Codex Plan projection, and legacy Claude structured resume before a complete matrix rerun. - Prior failing matrix: https://github.com/paperclipai/paperclip/actions/runs/33682434315 ## Risks - Bundling changes the bytes covered by provider launch hashes. Provider-pack generation already hashes the final built files. - ACPX still loads qualified provider packages dynamically. The controller supplies a normalized package root, while existing version, digest, path, and descriptor checks remain active. - Durable `failed` is terminal. Replays return the same redacted result and do not execute the provider effect twice. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used OpenAI Codex based on GPT-5 with agentic reasoning, repository inspection, code editing, Git, parallel subagents, and GitHub Actions coordination. The exact deployed snapshot and context-window size are not exposed to this task. ## 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 linked related public work or described the bug in this PR - [x] I have not referenced internal or instance-local Paperclip issues or links - [x] My branch name describes the change and contains no internal ticket id - [ ] I have run tests locally and they pass (intentionally deferred to GitHub Actions) - [x] I have added or updated tests where applicable - [x] No documentation change is required for this runtime repair - [x] I have considered and documented the risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge
305 lines
10 KiB
TypeScript
305 lines
10 KiB
TypeScript
import { cp, mkdir, readFile, readdir } from "node:fs/promises";
|
|
import path from "node:path";
|
|
import { pathToFileURL } from "node:url";
|
|
import type { RunnerE2EResult } from "./types.js";
|
|
|
|
interface WorkflowJob {
|
|
name: string;
|
|
run_attempt: number;
|
|
started_at: string;
|
|
}
|
|
|
|
interface WorkflowJobsResponse {
|
|
jobs: WorkflowJob[];
|
|
attempts: Array<{
|
|
run_attempt: number;
|
|
run_started_at: string;
|
|
}>;
|
|
}
|
|
|
|
export interface SelectRerunArtifactsInput {
|
|
artifactRoot: string;
|
|
selectedRoot: string;
|
|
jobs: WorkflowJobsResponse;
|
|
expectedExecutionIds: readonly string[];
|
|
workflowRunId: string;
|
|
workflowRunAttempt: number;
|
|
sourceSha: string;
|
|
sourceRef: string;
|
|
workflowRunUrl: string;
|
|
}
|
|
|
|
function safeIdentifier(value: string, label: string) {
|
|
if (!value || !/^[A-Za-z0-9_.-]+$/u.test(value)) {
|
|
throw new Error(
|
|
`${label} contains unsafe characters: ${JSON.stringify(value)}`,
|
|
);
|
|
}
|
|
return value;
|
|
}
|
|
|
|
function positiveInteger(value: unknown, label: string) {
|
|
if (typeof value !== "number" || !Number.isSafeInteger(value) || value < 1) {
|
|
throw new Error(`${label} must be a positive integer`);
|
|
}
|
|
return value;
|
|
}
|
|
|
|
function timestamp(value: unknown, label: string) {
|
|
if (typeof value !== "string" || !Number.isFinite(Date.parse(value))) {
|
|
throw new Error(`${label} must be an ISO timestamp`);
|
|
}
|
|
return Date.parse(value);
|
|
}
|
|
|
|
async function walk(root: string): Promise<string[]> {
|
|
const entries = await readdir(root, { withFileTypes: true });
|
|
const files: string[] = [];
|
|
for (const entry of entries) {
|
|
const full = path.join(root, entry.name);
|
|
if (entry.isDirectory()) files.push(...(await walk(full)));
|
|
else if (entry.isFile()) files.push(full);
|
|
}
|
|
return files;
|
|
}
|
|
|
|
/**
|
|
* Selects the artifact produced by the latest workflow job for each execution.
|
|
* GitHub reruns retain earlier-attempt artifacts, so validity or result
|
|
* timestamps must never be used to choose between workflow attempts.
|
|
*/
|
|
export async function selectRerunArtifacts(input: SelectRerunArtifactsInput) {
|
|
const runId = safeIdentifier(input.workflowRunId, "workflow run ID");
|
|
const currentAttempt = positiveInteger(
|
|
input.workflowRunAttempt,
|
|
"workflow run attempt",
|
|
);
|
|
const expected = input.expectedExecutionIds.map((executionId) =>
|
|
safeIdentifier(executionId, "execution ID"),
|
|
);
|
|
if (expected.length === 0 || new Set(expected).size !== expected.length) {
|
|
throw new Error("expected execution IDs must be non-empty and unique");
|
|
}
|
|
const preexistingSelections = await readdir(input.selectedRoot).catch(
|
|
(error: NodeJS.ErrnoException) => {
|
|
if (error.code === "ENOENT") return [];
|
|
throw error;
|
|
},
|
|
);
|
|
if (preexistingSelections.length > 0) {
|
|
throw new Error("selected artifact root must start empty");
|
|
}
|
|
const expectedSet = new Set(expected);
|
|
const attemptStartedAt = new Map<number, number>();
|
|
for (const attemptMetadata of input.jobs.attempts) {
|
|
const attempt = positiveInteger(
|
|
attemptMetadata.run_attempt,
|
|
"workflow metadata attempt",
|
|
);
|
|
if (attempt > currentAttempt || attemptStartedAt.has(attempt)) {
|
|
throw new Error(`invalid workflow metadata for attempt ${attempt}`);
|
|
}
|
|
attemptStartedAt.set(
|
|
attempt,
|
|
timestamp(
|
|
attemptMetadata.run_started_at,
|
|
`workflow attempt ${attempt} start`,
|
|
),
|
|
);
|
|
}
|
|
for (let attempt = 1; attempt <= currentAttempt; attempt += 1) {
|
|
if (!attemptStartedAt.has(attempt)) {
|
|
throw new Error(`workflow metadata omitted attempt ${attempt}`);
|
|
}
|
|
}
|
|
const attemptsByExecution = new Map<string, Map<number, WorkflowJob>>();
|
|
for (const candidate of input.jobs.jobs) {
|
|
if (!expectedSet.has(candidate.name)) continue;
|
|
const attempt = positiveInteger(candidate.run_attempt, "job run attempt");
|
|
if (attempt > currentAttempt) {
|
|
throw new Error(
|
|
`job ${candidate.name} claims future workflow attempt ${attempt}`,
|
|
);
|
|
}
|
|
const jobStartedAt = timestamp(
|
|
candidate.started_at,
|
|
`job ${candidate.name} start`,
|
|
);
|
|
// GitHub's filter=all response synthesizes current-attempt rows for jobs
|
|
// retained from an earlier attempt. Their started_at remains before the
|
|
// current attempt's trusted run_started_at, so they are not real reruns.
|
|
if (jobStartedAt < attemptStartedAt.get(attempt)!) continue;
|
|
const attempts = attemptsByExecution.get(candidate.name) ?? new Map();
|
|
if (attempts.has(attempt)) {
|
|
throw new Error(
|
|
`workflow attempt ${attempt} contains duplicate job ${candidate.name}`,
|
|
);
|
|
}
|
|
attempts.set(attempt, candidate);
|
|
attemptsByExecution.set(candidate.name, attempts);
|
|
}
|
|
|
|
const latestAttemptByExecution = new Map<string, number>();
|
|
for (const executionId of expected) {
|
|
const attempts = [...(attemptsByExecution.get(executionId)?.keys() ?? [])];
|
|
if (attempts.length === 0) {
|
|
throw new Error(
|
|
`workflow job history omitted expected execution ${executionId}`,
|
|
);
|
|
}
|
|
latestAttemptByExecution.set(executionId, Math.max(...attempts));
|
|
}
|
|
|
|
const recognizedArtifactNames = new Map<
|
|
string,
|
|
{ executionId: string; workflowAttempt: number }
|
|
>();
|
|
for (const executionId of expected) {
|
|
for (const workflowAttempt of attemptsByExecution
|
|
.get(executionId)!
|
|
.keys()) {
|
|
recognizedArtifactNames.set(
|
|
`runner-e2e-${runId}-${workflowAttempt}-${executionId}`,
|
|
{ executionId, workflowAttempt },
|
|
);
|
|
}
|
|
}
|
|
|
|
const artifactEntries = await readdir(input.artifactRoot, {
|
|
withFileTypes: true,
|
|
}).catch((error: NodeJS.ErrnoException) => {
|
|
if (error.code === "ENOENT") return [];
|
|
throw error;
|
|
});
|
|
const artifactDirectories = new Map<string, string>();
|
|
for (const entry of artifactEntries) {
|
|
const identity = recognizedArtifactNames.get(entry.name);
|
|
if (!identity || !entry.isDirectory()) {
|
|
throw new Error(`downloaded unexpected runner artifact ${entry.name}`);
|
|
}
|
|
artifactDirectories.set(
|
|
entry.name,
|
|
path.join(input.artifactRoot, entry.name),
|
|
);
|
|
}
|
|
|
|
const selections: Array<{
|
|
executionId: string;
|
|
workflowAttempt: number;
|
|
artifactName: string;
|
|
}> = [];
|
|
for (const executionId of expected) {
|
|
const workflowAttempt = latestAttemptByExecution.get(executionId)!;
|
|
const artifactName = `runner-e2e-${runId}-${workflowAttempt}-${executionId}`;
|
|
const artifactDirectory = artifactDirectories.get(artifactName);
|
|
// A latest job without an artifact must remain missing. Falling back to an
|
|
// older successful artifact would mask an infrastructure/upload failure.
|
|
if (!artifactDirectory) continue;
|
|
|
|
const campaignName = `gha-${runId}-${workflowAttempt}-${executionId}`;
|
|
const topLevelEntries = await readdir(artifactDirectory, {
|
|
withFileTypes: true,
|
|
});
|
|
if (
|
|
topLevelEntries.length !== 1 ||
|
|
topLevelEntries[0]?.name !== campaignName ||
|
|
!topLevelEntries[0].isDirectory()
|
|
) {
|
|
throw new Error(
|
|
`${artifactName} must contain only its exact campaign ${campaignName}`,
|
|
);
|
|
}
|
|
const campaignDirectory = path.join(artifactDirectory, campaignName);
|
|
const resultFiles = (await walk(campaignDirectory)).filter(
|
|
(file) => path.basename(file) === "result.json",
|
|
);
|
|
if (resultFiles.length === 0) {
|
|
throw new Error(`${artifactName} contains no normalized result`);
|
|
}
|
|
for (const resultFile of resultFiles) {
|
|
const result = JSON.parse(
|
|
await readFile(resultFile, "utf8"),
|
|
) as RunnerE2EResult;
|
|
if (result.executionId !== executionId) {
|
|
throw new Error(
|
|
`${artifactName} contains result for ${String(result.executionId)}`,
|
|
);
|
|
}
|
|
const source = result.source;
|
|
if (
|
|
!source ||
|
|
source.sha !== input.sourceSha ||
|
|
source.ref !== input.sourceRef ||
|
|
source.workflowRunUrl !== input.workflowRunUrl
|
|
) {
|
|
throw new Error(`${artifactName} contains result from another source`);
|
|
}
|
|
}
|
|
const destination = path.join(
|
|
input.selectedRoot,
|
|
artifactName,
|
|
campaignName,
|
|
);
|
|
await mkdir(path.dirname(destination), { recursive: true });
|
|
await cp(campaignDirectory, destination, {
|
|
recursive: true,
|
|
force: false,
|
|
errorOnExist: true,
|
|
});
|
|
selections.push({ executionId, workflowAttempt, artifactName });
|
|
}
|
|
return selections;
|
|
}
|
|
|
|
async function main() {
|
|
const required = (name: string) => {
|
|
const value = process.env[name]?.trim();
|
|
if (!value) throw new Error(`${name} is required`);
|
|
return value;
|
|
};
|
|
const expected = JSON.parse(
|
|
required("PAPERCLIP_RUNNER_E2E_EXPECTED_IDS"),
|
|
) as unknown;
|
|
if (
|
|
!Array.isArray(expected) ||
|
|
expected.some((value) => typeof value !== "string")
|
|
) {
|
|
throw new Error(
|
|
"PAPERCLIP_RUNNER_E2E_EXPECTED_IDS must be a JSON string array",
|
|
);
|
|
}
|
|
const jobs = JSON.parse(
|
|
await readFile(required("PAPERCLIP_RUNNER_E2E_JOBS_JSON"), "utf8"),
|
|
) as WorkflowJobsResponse;
|
|
if (!jobs || !Array.isArray(jobs.jobs) || !Array.isArray(jobs.attempts)) {
|
|
throw new Error("workflow jobs JSON must contain jobs and attempts arrays");
|
|
}
|
|
const runId = required("GITHUB_RUN_ID");
|
|
const serverUrl = required("GITHUB_SERVER_URL");
|
|
const repository = required("GITHUB_REPOSITORY");
|
|
const selections = await selectRerunArtifacts({
|
|
artifactRoot: required("PAPERCLIP_RUNNER_E2E_ARTIFACT_ROOT"),
|
|
selectedRoot: required("PAPERCLIP_RUNNER_E2E_SELECTED_ROOT"),
|
|
jobs,
|
|
expectedExecutionIds: expected,
|
|
workflowRunId: runId,
|
|
workflowRunAttempt: Number(required("GITHUB_RUN_ATTEMPT")),
|
|
sourceSha: required("PAPERCLIP_RUNNER_E2E_SOURCE_SHA"),
|
|
sourceRef: required("PAPERCLIP_RUNNER_E2E_SOURCE_REF"),
|
|
workflowRunUrl: `${serverUrl}/${repository}/actions/runs/${runId}`,
|
|
});
|
|
console.log(
|
|
`Selected ${selections.length}/${expected.length} latest cell artifacts`,
|
|
);
|
|
}
|
|
|
|
if (
|
|
process.argv[1] &&
|
|
import.meta.url === pathToFileURL(path.resolve(process.argv[1])).href
|
|
) {
|
|
await main().catch((error) => {
|
|
console.error(error instanceof Error ? error.message : String(error));
|
|
process.exitCode = 1;
|
|
});
|
|
}
|