CodeIssuesDiscussionsWikiPull RequestsProjectsCommitsActionsReleasesContributorsPulse● GatesSecuritySettingsDeploymentsPipelineInsightsAgents✨ Explain✨ Ask AI✨ Workspace✨ Spec✨ Tests▓ Debt Map✨ NL Search🏛 Archaeology
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 | /**
* Tests for src/lib/ai-ci-healer.ts.
*
* Two layers:
* 1. Pure helpers — buildCiHealPrompt + fixesToFindings + small private
* helpers exposed via __test. Always run.
* 2. End-to-end with a fake Claude client + a real DB row + bare repo.
* Gated on DATABASE_URL via the HAS_DB skipIf pattern used across
* the suite.
*
* The Anthropic client is faked via the public `client` option so we never
* touch the network or require ANTHROPIC_API_KEY. We also DI the
* patch-generator into `healOneRun` so each test pins its own outcome.
*/
import { describe, it, expect, beforeAll, afterAll } from "bun:test";
import { join } from "path";
import { mkdir, rm } from "fs/promises";
import { eq, and } from "drizzle-orm";
import {
analyzeFailedWorkflowRun,
buildCiHealPrompt,
fixesToFindings,
healOneRun,
runCiHealerTick,
__test,
} from "../lib/ai-ci-healer";
import { db } from "../db";
import {
auditLog,
pullRequests,
repositories,
users,
workflowJobs,
workflowRuns,
workflows,
} from "../db/schema";
import {
createOrUpdateFileOnBranch,
initBareRepo,
} from "../git/repository";
const HAS_DB = Boolean(process.env.DATABASE_URL);
const TEST_REPOS = join(
import.meta.dir,
"../../.test-repos-ai-ci-healer-" + Date.now()
);
beforeAll(async () => {
process.env.GIT_REPOS_PATH = TEST_REPOS;
process.env.DATABASE_URL = process.env.DATABASE_URL || "";
// Make sure autopilot doesn't short-circuit our tests via env.
delete process.env.AUTOPILOT_DISABLED;
await rm(TEST_REPOS, { recursive: true, force: true });
await mkdir(TEST_REPOS, { recursive: true });
});
afterAll(async () => {
await rm(TEST_REPOS, { recursive: true, force: true });
});
// ---------------------------------------------------------------------------
// Pure helpers
// ---------------------------------------------------------------------------
describe("buildCiHealPrompt", () => {
it("embeds repo, sha, yaml, and failed job logs", () => {
const prompt = buildCiHealPrompt({
repoFullName: "alice/web",
commitSha: "abcdef1234567890",
workflowYaml: "name: ci\non: [push]\njobs:\n build:\n steps: []",
failedJobs: [
{ name: "build", conclusion: "failure", logs: "TypeError: x is not a function" },
],
});
expect(prompt).toContain("alice/web");
expect(prompt).toContain("abcdef123456");
expect(prompt).toContain("name: ci");
expect(prompt).toContain("TypeError: x is not a function");
// Strict JSON schema cue
expect(prompt).toContain('"rootCause"');
expect(prompt).toContain('"suggestedFixes"');
});
it("handles the empty-jobs case gracefully", () => {
const prompt = buildCiHealPrompt({
repoFullName: "x/y",
commitSha: "1234567",
workflowYaml: "name: ci",
failedJobs: [],
});
expect(prompt).toContain("(no failed-job logs available)");
});
});
describe("fixesToFindings", () => {
it("maps SuggestedFix[] onto GateTestFinding[] preserving path + severity", () => {
const findings = fixesToFindings(
[
{ path: "src/a.ts", description: "fix imports", severity: "high" },
{ path: "src/b.ts", description: "guard null", severity: "medium" },
],
"deadbeef"
);
expect(findings.length).toBe(2);
expect(findings[0].path).toBe("src/a.ts");
expect(findings[0].severity).toBe("high");
expect(findings[0].id).toContain("ci-heal-deadbeef");
expect(findings[1].path).toBe("src/b.ts");
expect(findings[1].severity).toBe("medium");
});
it("drops fixes with no path", () => {
const findings = fixesToFindings(
[
{ path: "", description: "noop" },
{ path: "src/c.ts", description: "ok" },
],
"x"
);
expect(findings.length).toBe(1);
expect(findings[0].path).toBe("src/c.ts");
});
it("defaults severity to high when omitted", () => {
const findings = fixesToFindings(
[{ path: "src/d.ts", description: "no sev" }],
"x"
);
expect(findings[0].severity).toBe("high");
});
});
describe("__test internals", () => {
it("normaliseSeverity accepts the canonical levels case-insensitively", () => {
expect(__test.normaliseSeverity("HIGH")).toBe("high");
expect(__test.normaliseSeverity("Medium")).toBe("medium");
expect(__test.normaliseSeverity("critical")).toBe("critical");
expect(__test.normaliseSeverity("low")).toBe("low");
});
it("normaliseSeverity rejects garbage", () => {
expect(__test.normaliseSeverity("bananas")).toBeUndefined();
expect(__test.normaliseSeverity(undefined)).toBeUndefined();
expect(__test.normaliseSeverity(null)).toBeUndefined();
expect(__test.normaliseSeverity(42)).toBeUndefined();
});
it("truncate caps long strings + appends marker", () => {
expect(__test.truncate("abc", 10)).toBe("abc");
const long = "x".repeat(20);
const t = __test.truncate(long, 5);
expect(t.startsWith("xxxxx")).toBe(true);
expect(t).toContain("(truncated)");
});
});
// ---------------------------------------------------------------------------
// Skip when AUTOPILOT_DISABLED: tick must no-op cleanly
// ---------------------------------------------------------------------------
describe("runCiHealerTick — env gates", () => {
it("no-ops when AUTOPILOT_DISABLED=1", async () => {
const prev = process.env.AUTOPILOT_DISABLED;
process.env.AUTOPILOT_DISABLED = "1";
try {
const summary = await runCiHealerTick({
findCandidates: async () => {
throw new Error("should not be called");
},
});
expect(summary).toEqual({
considered: 0,
healed: 0,
gaveUp: 0,
skipped: 0,
});
} finally {
if (prev === undefined) delete process.env.AUTOPILOT_DISABLED;
else process.env.AUTOPILOT_DISABLED = prev;
}
});
it("no-ops when ANTHROPIC_API_KEY is unset", async () => {
const prev = process.env.ANTHROPIC_API_KEY;
delete process.env.ANTHROPIC_API_KEY;
try {
const summary = await runCiHealerTick({
findCandidates: async () => {
throw new Error("should not be called");
},
});
expect(summary.considered).toBe(0);
expect(summary.healed).toBe(0);
} finally {
if (prev !== undefined) process.env.ANTHROPIC_API_KEY = prev;
}
});
});
// ---------------------------------------------------------------------------
// DB-backed end-to-end. Builds a real failed run + jobs + repo + workflow
// rows, then drives `healOneRun` with a fake Claude client.
// ---------------------------------------------------------------------------
function fakeClient(responseText: string) {
return {
messages: {
create: async () => ({
content: [{ type: "text" as const, text: responseText }],
}),
},
} as any;
}
interface Fixture {
repoId: string;
repoName: string;
ownerUsername: string;
workflowId: string;
runId: string;
baseSha: string;
}
async function seedFailedRun(label: string): Promise<Fixture> {
const username = `cihealer_${label}_${Date.now()}_${Math.random()
.toString(36)
.slice(2, 6)}`;
const [u] = await db
.insert(users)
.values({
username,
email: `${username}@example.com`,
passwordHash: "x",
})
.returning({ id: users.id });
const repoName = `subject_${label}_${Date.now()}`;
const [r] = await db
.insert(repositories)
.values({
ownerId: u.id,
name: repoName,
diskPath: `/tmp/${username}/${repoName}`,
defaultBranch: "main",
})
.returning({ id: repositories.id, name: repositories.name });
await initBareRepo(username, repoName);
const seeded = await createOrUpdateFileOnBranch({
owner: username,
name: repoName,
branch: "main",
filePath: "src/index.ts",
bytes: new TextEncoder().encode(
"export function broken(){ return undefined.length; }\n"
),
message: "seed",
authorName: "Seeder",
authorEmail: "s@e.com",
});
if ("error" in seeded) throw new Error("seed failed: " + seeded.error);
const [w] = await db
.insert(workflows)
.values({
repositoryId: r.id,
name: "ci",
path: ".gluecron/workflows/ci.yml",
yaml: "name: ci\non: [push]\njobs:\n build:\n steps:\n - run: bun test",
parsed: JSON.stringify({
name: "ci",
on: ["push"],
jobs: { build: { steps: [{ run: "bun test" }] } },
}),
})
.returning({ id: workflows.id });
// Insert run as `failure` with createdAt safely in the past so the
// candidate finder considers it (HEAL_MIN_AGE_MS = 60s).
const oldCreatedAt = new Date(Date.now() - 5 * 60 * 1000);
const [run] = await db
.insert(workflowRuns)
.values({
workflowId: w.id,
repositoryId: r.id,
runNumber: 1,
event: "push",
ref: "refs/heads/main",
commitSha: seeded.commitSha,
status: "failure",
conclusion: "failure",
queuedAt: oldCreatedAt,
startedAt: oldCreatedAt,
finishedAt: oldCreatedAt,
createdAt: oldCreatedAt,
})
.returning({ id: workflowRuns.id });
await db.insert(workflowJobs).values({
runId: run.id,
name: "build",
jobOrder: 0,
runsOn: "default",
status: "failure",
conclusion: "failure",
exitCode: 1,
steps: "[]",
logs:
"==> bun test\nTypeError: Cannot read properties of undefined (reading 'length')\n at broken (src/index.ts:1:38)\n[exit 1 in 120ms]",
startedAt: oldCreatedAt,
finishedAt: oldCreatedAt,
});
return {
repoId: r.id,
repoName,
ownerUsername: username,
workflowId: w.id,
runId: run.id,
baseSha: seeded.commitSha,
};
}
describe.skipIf(!HAS_DB)("ai-ci-healer DB-backed E2E", () => {
it("opens a patch PR + writes an ai.ci.healed audit row on the happy path", async () => {
const fx = await seedFailedRun("happy");
const cannedClaude = JSON.stringify({
rootCause:
"broken() references undefined.length, which throws at runtime.",
fixable: true,
suggestedFixes: [
{
path: "src/index.ts",
description: "Return a numeric literal instead of dereferencing undefined.",
severity: "high",
},
],
});
const cannedPatch = JSON.stringify({
explanation: "Replaced the bad expression with a safe literal.",
patches: [
{
path: "src/index.ts",
new_content: "export function broken(){ return 0; }\n",
},
],
});
const client = fakeClient(cannedClaude);
// Use the real patch generator — feed it the same fake client for the
// patch call. The generator picks `client` from the opts we forward.
const { generatePatchForGateTestFinding } = await import(
"../lib/ai-patch-generator"
);
const out = await healOneRun(fx.runId, {
client,
// Wrap the real generator so we can swap in the second fake response
// for the patch step (it makes a fresh call). The healer hands the
// client through, so we wrap to replace the response between steps.
generatePatch: async (opts) => {
return generatePatchForGateTestFinding({
...opts,
client: fakeClient(cannedPatch),
branchOverride: `ai-patch/ci-heal-${Date.now()}`,
});
},
});
expect(out.outcome).toBe("healed");
expect(typeof out.prNumber).toBe("number");
expect(out.branch).toContain("ai-patch/ci-heal-");
// PR row exists
const prs = await db
.select({ number: pullRequests.number, headBranch: pullRequests.headBranch })
.from(pullRequests)
.where(eq(pullRequests.repositoryId, fx.repoId));
expect(prs.length).toBe(1);
expect(prs[0].number).toBe(out.prNumber!);
// Audit marker present
const audits = await db
.select({ action: auditLog.action })
.from(auditLog)
.where(
and(
eq(auditLog.targetType, "workflow_run"),
eq(auditLog.targetId, fx.runId)
)
);
expect(audits.some((a) => a.action === "ai.ci.healed")).toBe(true);
}, 30_000);
it("writes ai.ci.gave_up + opens no PR when Claude says unfixable", async () => {
const fx = await seedFailedRun("unfix");
const cannedClaude = JSON.stringify({
rootCause: "The npm registry returned 503 mid-install.",
fixable: false,
suggestedFixes: [],
unfixableReason: "External registry outage — retry later.",
});
const client = fakeClient(cannedClaude);
const out = await healOneRun(fx.runId, { client });
expect(out.outcome).toBe("gave_up");
// No PR row
const prs = await db
.select({ number: pullRequests.number })
.from(pullRequests)
.where(eq(pullRequests.repositoryId, fx.repoId));
expect(prs.length).toBe(0);
// Audit marker present as gave_up
const audits = await db
.select({ action: auditLog.action })
.from(auditLog)
.where(
and(
eq(auditLog.targetType, "workflow_run"),
eq(auditLog.targetId, fx.runId)
)
);
expect(audits.some((a) => a.action === "ai.ci.gave_up")).toBe(true);
expect(audits.some((a) => a.action === "ai.ci.healed")).toBe(false);
}, 30_000);
it("skips runs that already have a marker (no double-processing)", async () => {
const fx = await seedFailedRun("dedupe");
// Pre-insert a marker
await db.insert(auditLog).values({
action: "ai.ci.healed",
targetType: "workflow_run",
targetId: fx.runId,
metadata: JSON.stringify({ pre: true }),
});
// Should refuse to act — Claude must NOT be called.
let claudeCalled = 0;
const client = {
messages: {
create: async () => {
claudeCalled += 1;
return { content: [{ type: "text" as const, text: "{}" }] };
},
},
} as any;
const out = await healOneRun(fx.runId, { client });
expect(out.outcome).toBe("skipped");
expect(claudeCalled).toBe(0);
}, 30_000);
});
// ---------------------------------------------------------------------------
// analyzeFailedWorkflowRun direct-call sanity (the public surface the
// task description explicitly calls out as the entry point).
// ---------------------------------------------------------------------------
describe.skipIf(!HAS_DB)("analyzeFailedWorkflowRun", () => {
it("returns parsed analysis with patchablePaths when Claude says fixable", async () => {
const fx = await seedFailedRun("analyze");
const canned = JSON.stringify({
rootCause: "Null deref in src/index.ts.",
fixable: true,
suggestedFixes: [
{ path: "src/index.ts", description: "Add null guard.", severity: "high" },
],
});
const analysis = await analyzeFailedWorkflowRun(fx.runId, {
client: fakeClient(canned),
});
expect(analysis).not.toBeNull();
expect(analysis!.rootCause).toContain("Null deref");
expect(analysis!.patchablePaths).toEqual(["src/index.ts"]);
expect(analysis!.suggestedFixes.length).toBe(1);
}, 30_000);
it("returns null when Claude says unfixable", async () => {
const fx = await seedFailedRun("analyze-unfix");
const canned = JSON.stringify({
rootCause: "Registry outage.",
fixable: false,
suggestedFixes: [],
});
const analysis = await analyzeFailedWorkflowRun(fx.runId, {
client: fakeClient(canned),
});
expect(analysis).toBeNull();
}, 30_000);
it("returns null for non-failure runs", async () => {
const fx = await seedFailedRun("non-failure");
// Flip the run back to success — analyzer should bail.
await db
.update(workflowRuns)
.set({ status: "success", conclusion: "success" })
.where(eq(workflowRuns.id, fx.runId));
const analysis = await analyzeFailedWorkflowRun(fx.runId, {
client: fakeClient('{"rootCause":"x","fixable":true,"suggestedFixes":[]}'),
});
expect(analysis).toBeNull();
}, 30_000);
});
|