mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-13 15:55:46 +01:00
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The chat-feature evals measure whether agents complete real chat requests through the product > - The chat catalog eval harness and its repairs live on the `chat-eval/master-candidate` branch, not on master > - Today's work made 139 of 242 chat evals pass in verified single-case live runs. Each product fix landed on master as its own pull request > - The eval-side changes (setups, graders, fault seams, evidence capture) still exist only on the candidate > - This pull request lands the harness on master. Master stays the source of truth for product behavior > - The benefit is one tree: evals run against master commits, and product fixes reach the evals as soon as they merge ## Linked Issues or Issue Description Refs #19441, Refs #19599, Refs #20270, Refs #20334, Refs #20342, Refs #20344, Refs #20354, Refs #20360, Refs #20373, Refs #20382, Refs #20409 **What existing behavior does this improve?** The chat catalog eval harness and the product it evaluates live on a long-running branch outside master. Eval fixes do not reach master, so master-based eval runs use old setups and graders. **Subsystem affected** `tests/runner-e2e` chat catalog harness, plus a small set of product changes that exist only on the candidate. **Proposed change** Land the harness on master. Every product fix from this eval work already merged to master as its own pull request. Conflicts keep master's product behavior. ## What Changed - Most of the diff (808 of 851 files) is under `tests/runner-e2e/`: chat-catalog domains, scenarios, descriptors, the TestOps manifest, private evidence tooling, and their tests. - Graders now check saved state, setup literals and rendered or clicked links. Many keyword and regex checks on free-form replies are removed. Each changed grader has a bumped `oracleVersion` and calibration tests (a valid reply, a reworded valid reply, and broken variants that must fail). - The shared link helper `hasLink` no longer treats a list link (for example `/COM/issues?view=mine`) as a link to every task without a URL key. - Setup fixes build real state through supported product routes: routine history, agent grants through the board grant route, budget holds with a real cap, and a storage-layer archive fault for `inbox-search.archive-partial-failure`. - Browser boundary journeys (sign-in, chat-disabled, onboarding) now write the standard pass evidence files. - Retargeted cases, approved by the maintainer: `agent-chat.chat-disabled` (Conference Room gate), `agents.restore-instructions` (no saved instruction history), `access.manage-membership` (member permission edits stay board-only). - The private evidence trusted files (`manual-private-handoff.ts`, `manual-private-retention.ts`, `manual-private-report-assets.ts`) are byte-identical to master. - Product files outside `tests/` (43 files, +1,328 −49) carry candidate-only changes that master does not yet have. Most are tests. The main code changes are in `request-human-input.ts`, `heartbeat-run-summary.ts`, `native-completion-feedback.ts`, `native-run-finalizer.ts`, `openapi.ts`, `secrets.ts` and `IssueChatLongThreadPerf.tsx`. ## Verification - `pnpm vitest run chat-catalog-domains chat-catalog-contract.test.ts chat-catalog-consistency.test.ts unit-launcher.test.mjs --testTimeout=180000` in `tests/runner-e2e`: 269 files pass, 3,550 tests pass. - `node --test chat-full-catalog-testops-manifest.test.mjs`: 10 pass. - `node tests/runner-e2e/source-sdk-public-inputs.mjs <HEAD> --check`: no change. - `tsc` for `server` and `tests/runner-e2e`: clean. - Live single-case verification: 139 of 242 cases passed on paid runs, and a person read the saved evidence for each. These are single-case runs, not one full-cohort run. ## Risks - Large diff. Most of it is test-only, but the 43 non-test files change product code paths. Review those files first. - Some graders are now less strict on reply wording. A person must read replies for meaning on those cases. - 42 cases need disposable outside accounts (Notion, Slack, GitHub, Daytona and others) and stay blocked. `access.invite-teammate` stays failing until local agent isolation lands, because the runner's durable state keeps a live invite token. - The paid workflow compares the trusted private-evidence files with master. They are unchanged, so this risk is low. ## Model Used Anthropic Claude Opus 5.5 (`claude-opus-5-5`), with tool use and code execution, in Claude Code. Subagents used the same 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 - [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 - [ ] 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 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Paperclip <noreply@paperclip.ing>
70 lines
5.5 KiB
TypeScript
70 lines
5.5 KiB
TypeScript
import { pollUntil } from "../api.js";
|
|
import { isDeepStrictEqual } from "node:util";
|
|
import { collectRunEvents } from "../run-observations.js";
|
|
import type { ChatRun } from "../chat-flow.js";
|
|
import type { ChatCatalogContext } from "../chat-catalog-contract.js";
|
|
import type { RuntimeAgent } from "./runtime-support.js";
|
|
interface NativeFailureRun extends ChatRun {
|
|
nativeIssueId?: string | null;
|
|
runnerProfileJson?: { nativeExecutionInput?: { provider?: { kind?: string; model?: string | null } } };
|
|
}
|
|
interface NativeEvent { seq?: number; eventType?: string; payload?: { prpEvent?: { eventType?: string } } }
|
|
export function admittedModelFailure(input: { run?: NativeFailureRun; events: NativeEvent[]; companyId: string; agentId: string; issueId: string; model: string; provider?: "codex" | "opencode" }) {
|
|
const r = input.run, provider = r?.runnerProfileJson?.nativeExecutionInput?.provider;
|
|
return !!r && r.runtimeMode === "native" && r.status === "failed" && !!r.startedAt && !!r.errorCode && !!r.error
|
|
&& r.companyId === input.companyId && r.agentId === input.agentId
|
|
&& [r.nativeIssueId, r.contextSnapshot?.issueId, r.contextSnapshot?.taskId].includes(input.issueId)
|
|
&& provider?.kind === (input.provider ?? "codex") && provider.model === input.model
|
|
&& input.events.some(event => [event.eventType, event.payload?.prpEvent?.eventType].some(type => type === "run.terminal" || type === "turn.failed"));
|
|
}
|
|
/** A real public configuration/invocation fixture, never a synthetic failure.
|
|
* The caller supplies exactly one ordinary task/routine invocation and retains
|
|
* its own invocation receipt. HTTP/preflight rejection is a setup blocker.
|
|
* This source-qualified recipe is not live qualification until this predicate
|
|
* observes an actually admitted native failed row and durable terminal events.
|
|
*/
|
|
export async function unavailableNativeModelFailure(ctx: ChatCatalogContext, worker: RuntimeAgent,
|
|
invoke: () => Promise<{ issueId: string }>) {
|
|
const endpoint = `/api/agents/${worker.id}`;
|
|
const original = await ctx.input.api.get<RuntimeAgent>(endpoint);
|
|
if (original.lifecycleState !== "ready" || !["active", "idle"].includes(original.status ?? "")
|
|
|| original.adapterType !== "paperclip_runner" || !["codex", "opencode"].includes(String(original.adapterConfig?.provider)))
|
|
throw new Error("Native failure prerequisite requires an actually ready selected native Codex or OpenCode worker");
|
|
const provider = original.adapterConfig!.provider as "codex" | "opencode";
|
|
const before = await ctx.allRuns(); const model = unavailableNativeModel(provider, ctx.input.nonce);
|
|
for (const scope of [{ scopeType: "company", scopeId: ctx.input.fixtures.company.id }, { scopeType: "agent", scopeId: worker.id }])
|
|
await ctx.input.api.post(`/api/companies/${ctx.input.fixtures.company.id}/budgets/policies`, { ...scope, amount: 500, hardStopEnabled: true,
|
|
isActive: true, unpricedUsagePolicy: "allow" });
|
|
let failed: NativeFailureRun | undefined;
|
|
try {
|
|
const configured = await ctx.input.api.patch<RuntimeAgent>(endpoint, { adapterConfig: { ...original.adapterConfig, model } });
|
|
if (configured.adapterConfig?.model !== model || configured.lifecycleState !== "ready")
|
|
throw new Error("Unavailable-model configuration was not retained on the admitted worker; no failed-native coverage");
|
|
const receipt = await invoke();
|
|
if (!receipt.issueId) throw new Error("Native failure invocation did not return an actual source task");
|
|
const settled = await pollUntil({ label: "Actual admitted unavailable-model native failure", deadlineAt: Date.now() + 150_000,
|
|
load: async () => (await ctx.allRuns()).filter(run => !before.some(prior => prior.id === run.id)),
|
|
accept: rows => rows.length === 1 && rows[0]?.status === "failed",
|
|
reject: rows => rows.length > 1 ? "Failure prerequisite admitted more than its one declared invocation"
|
|
: rows.some(run => ["succeeded", "cancelled", "timed_out"].includes(run.status)) ? "Invocation did not produce genuine native failed history" : undefined });
|
|
failed = await ctx.input.api.get<NativeFailureRun>(`/api/heartbeat-runs/${settled[0]!.id}`);
|
|
const events = await collectRunEvents<NativeEvent>((afterSeq, limit) => ctx.input.api.get(`/api/heartbeat-runs/${failed!.id}/events?afterSeq=${afterSeq}&limit=${limit}`));
|
|
if (!admittedModelFailure({ run: failed, events, companyId: ctx.input.fixtures.company.id, agentId: worker.id, issueId: receipt.issueId, model, provider }))
|
|
throw new Error("Actual invocation lacks admitted model-bound native failure evidence; no failure coverage claimed");
|
|
return { failedRun: failed, failedRunEvents: events, failedIssueId: receipt.issueId, failedModel: model,
|
|
failureKind: "native-failure-under-explicit-unavailable-model-configuration", originalConfiguration: original };
|
|
} finally {
|
|
await ctx.input.api.patch(endpoint, { adapterConfig: original.adapterConfig });
|
|
const current = await ctx.input.api.get<RuntimeAgent>(endpoint);
|
|
if (current.status === "error") await ctx.input.api.post(`${endpoint}/clear-error`, {});
|
|
const restored = await ctx.input.api.get<RuntimeAgent>(endpoint);
|
|
if (!isDeepStrictEqual(restored.adapterConfig, original.adapterConfig)) throw new Error("Original admitted model configuration was not restored");
|
|
}
|
|
}
|
|
|
|
/** OpenCode requires a provider-qualified ID; actual admission and persisted failure remain mandatory. */
|
|
export function unavailableNativeModel(provider: "codex" | "opencode", nonce: string) {
|
|
const id = `paperclip-eval-unavailable-${nonce}`;
|
|
return provider === "opencode" ? `openrouter/paperclip-eval/${id}` : id;
|
|
}
|