PaperClipAI/tests/runner-e2e/CHAT-FEATURES.md

Chat feature evals

chat-features promotes executable entries from the scenario catalog into real Product E2E executions. The two original task journeys remain intact. Additional domain cases use chat-catalog-contract.ts and the explicit registry; listing a cell or calibrating its oracle is not live acceptance. The complete 244-scenario denominator and unchanged requests are frozen in chat-campaign-manifest.json. Its raw-file/defaults inventory hash is 9dc9cf6a5426cf1b56bc30a3571851f233f9dd5e93a1292eaee414d33b52df77.

Domain cases own production API setup, explicit user followups, observations, case-specific grading and optional cleanup. The shared driver submits the exact catalog query, saves prerequisite state before action, records real API comments, run attribution and rendered links, and rechecks after browser reload. Every case declares run/time limits and an oracle version. Missing checks and contradictory grades fail. Known coverageGaps make a diagnostic case fail acceptance until its conditions are implemented; they are not blocked cases reported as passing.

The registry contains only executable setups/actions/oracles. Scenarios absent from it remain in the denominator as unimplemented; optional external services must be real or explicitly labeled simulations. The frozen campaign ledger owns implementation readiness and all attempts separately from catalog discovery. No model judge supplies a generic pass.

Discovery reads pure chat-catalog-descriptors.json metadata and does not import any domain. The runtime loads the selected area's module and rejects descriptor bounds/oracle/policy drift. chat-catalog-fingerprint.ts compiles imports to find real local dependencies, including transitive helpers and the selected domain, and binds the normalized execution, exact scenario/defaults, package versions and lockfile. Whole source revision and global suite hash remain separate provenance. The six reviewed dynamic resolutions are pinned to their callsite source hashes; source drift falls back to conservative source coverage. Installed or external journey overrides refuse the reviewed local-chat lane. No global loader hook is installed. This permits unchanged cases to retain a measurement when an unrelated, unloaded domain changes; a changed executable dependency invalidates that case.

Standard funding/readiness and successful native runs are defaults. A scenario that preserves a special funding or paused state declares a setup policy and must independently verify its named prerequisites before the query. Declared terminal counts distinguish cancellation and denial from successful execution. Creation-only journeys may explicitly allow bounded separate worker execution while requiring a settled source and their own exact task/assignee oracle. Authenticated identity cases select the existing human-session fixture path; private secondary-user credentials join the attempt's redaction sink. Scenarios whose stated context has no working chat require a genuine boundary journey; these cannot receive a native chat pass from seeing an onboarding or login gate.

Scenario	Actual user request	Independent outcome
tasks.create-backlog	Add a task for Morgan to draft the launch announcement in Launch, but don't start it yet.	Exactly one task, Launch project, Morgan assignee, backlog at creation, no worker execution, usable chat link after reload.
tasks.delegate-announcement	Have Morgan write a two-sentence launch announcement using the Launch project brief, save it on the task, and send me the result here when it's finished.	Exactly one task, actual Morgan execution, current worker-authored document grounded in the project brief, task Done, automatic completion response in the original chat, usable result after reload.

The queries are read directly from chat-scenarios/tasks.yaml. Context and expected outcomes stay evaluator-side. The fixture creates a disposable company, CEO, Morgan and Launch project through production APIs; both agents use the shipped role instructions. The brief contains a product, date and random invitation code absent from the user query. The browser sends one message through Agent Chat. The harness never creates the result task, writes the announcement, completes the task, or sends a hidden follow-up.

Run the first slice

Follow the installation steps in README.md. The current profile is local native Codex using the catalog's recorded default model and an explicit CODEX_AUTH_JSON login. Load that login privately into the launcher environment; do not print it or save it in tracked configuration. The existing subscription fixture stages only auth.json in the disposable company login slot and removes it with the isolated instance.

pnpm test:e2e:runner:typecheck
pnpm test:e2e:runner:unit
pnpm test:e2e:runner -- --list --suite chat-features
pnpm test:e2e:runner -- --id chat-features.runner-codex-subscription.local.create-backlog --max-automatic-retries 0
pnpm test:e2e:runner -- --id chat-features.runner-codex-subscription.local.delegate-announcement --max-automatic-retries 0

Run creation first and inspect its packaged report before delegation. For the historical first-slice repeatability probe, each cell used three fresh fixtures on the same definition revision. The full-catalog campaign instead allows one initial attempt and one documented fix-and-rerun per failing scenario, with automatic retries disabled. Keep every attempt, including failures; a retry does not erase one. A fixture or product correction starts a new three-sample measurement. Both cells are explicit-only and excluded from --all; this does not implement or qualify the rest of the YAML inventory.

Bounds and grading

Creation expects one accounted native work run; delegation expects three (CEO handoff, Morgan work, CEO completion response). Runs have a 120-second timeout and attempts a seven-minute deadline. The visible task and completion reply must settle within five minutes after the human request, including a five-second stable observation. This is an independent journey acceptance bound: three runs may not each consume their maximum timeout and still pass. The outer attempt also covers fixture setup, reload and link checks, and evidence collection. A failed provider run rejects immediately; the historical missing turn-start event was not a five-minute poll timeout. Automatic retries are disabled. Company and agent budgets are 500 cents each, CEO daily work runs are capped at three, Morgan at one, and periodic heartbeats are disabled. Initial production hello probes fall outside heartbeat usage and budget accounting; detailed setup tokens and cost are unavailable. These limits are not an absolute inference or cash cap. Subscription cost coverage must remain explicit in the report.

Because this subscription currently reports unpriced usage, setup explicitly sets unpricedUsagePolicy: allow on the disposable company and both agents through the public budget API and verifies all three policies. The 500-cent hard stops remain enabled for reported costs; they cannot cap unknown dollar usage. Run caps and deadlines remain independent bounds. Without this explicit funding prerequisite, the default policy pauses the company after the first unpriced run and cancels the delegated worker. Such a pause is not a successful chat journey.

chat-feature.json retains the scenario ID, capability IDs, grader version, saved records, audit attribution, rendered links and individual checks. chat-feature-runs.json uses the existing native run evidence collector, including on failures. Setup evidence records fixture IDs, synthetic brief, readiness and saved budgets. Ordinary Product E2E evidence packaging, redaction, cleanup, failure classification and dashboards own the result. The definition digest includes the query, defaults, flow and grader sources.

The grader rejects duplicate tasks, wrong owners/projects/companies, early execution, missing audit attribution, missing or stale deliverables, extra human nudges and uncorrelated completion replies. Document provenance binds the current revision to Morgan's actual run. The completion check reuses completionDelivery; a handoff promise or a marker alone cannot establish delivery. Chromium verifies the task link and reload persistence.

The checks establish the specified saved state, two-sentence structure, required brief facts and result access. They do not score writing quality or every semantic claim in the response. Inspect the retained conversation and announcement before accepting the first samples; the existing completion semantic-review rubric still applies. Oracle calibration tests exercise both valid evidence and plausible wrong outcomes without calling a model.

Preserve the historical measurements below when promoting another scenario. Keep each starting condition explicit: connected versus missing connection, allowed versus approval-required, and so on. Reuse the fixture/report infrastructure and add a case-specific outcome oracle; do not make all YAML paragraphs executable through an unconstrained model judge.

Initial measurement — 2026-10-26

Three fresh samples per cell on definition 7fb20426ce3f1dd4efeddfb79f231336ea65a16b9c2a73bf6131e997acd74e9d produced 5/6 passes: backlog creation 3/3 and delegation 2/3. These used local native Codex subscription execution with gpt-5.6-sol. All six cleaned up successfully. Both workflows have successful browser-to-provider evidence, but the initial repeatability target is not met.

In campaign local-2026-10-26T15-48-12-199Z, the first CEO run in delegation failed before creating a task: runnerd did not report the provider turn identity. Retained events show submission and provider startup, followed by the transport's turn-start deadline. The precise cause of the missing event remains unresolved; the existing machine classification is retained without claiming that the provider or infrastructure has been proven responsible. Investigate this before expanding the suite. No replacement sample erased this failure.

The local results/chat-feature-acceptance/acceptance.md records every sample and earlier failed revision; its sibling dashboard uses the standard report generator. Results are ignored generated artifacts, so retain or publish them through the existing Product E2E pipeline when handing this measurement to another machine.

Follow-up measurement — 2026-10-26

The unchanged delegation cell subsequently passed three fresh executions: local-2026-10-26T16-50-50-751Z, local-2026-10-26T16-54-58-203Z, and local-2026-10-26T16-58-39-074Z. The latter two campaigns also ran backlog creation concurrently and passed both cells. All five selected executions used one attempt with zero automatic retries, retained the same definition above, and passed cleanup. Five existing turn-identity, early-terminal and runner-exit regressions passed as well.

These are green rechecks, not a demonstrated repair of the earlier startup interruption: no production code, timing limits or oracle assertions changed, and the missing turn-start event did not reproduce. Keep the original failure in the history. Across the original final-definition batch and this recheck, backlog is 5/5 and delegation is 5/6. The local results/chat-feature-recheck/verification.md records the follow-up evidence; its standard dashboard is explicitly scoped to the latest campaign.

Full catalog campaign

The frozen denominator is 244 requests across 35 areas and 191 capabilities. The durable progress record and chat-campaign-manifest.json distinguish complete, partial, unimplemented, blocked and actual measured results. Discovery and oracle calibration are not live acceptance. Each canonical scenario has one initial attempt and at most one coherent correction/rerun, with automatic retries zero; failures and unrun scenarios remain in the denominator.

Domain descriptors declare real setup/actions/independent grading and exact bounds. expectedHumanMessageCount and setupHumanMessageCount count only the source standing-chat history; genuine human comments and operator actions elsewhere are retained separately in scenario evidence. setupHumanMessageCount defaults to zero; genuine earlier source-chat human turns and native setup runs remain in the declared total, with persisted comment/run IDs captured before the canonical query. No historical model reply is fabricated.

Only steering.preserve-draft and tasks.attachment-brief may currently declare one queryAttachments receipt. The canonical typed query remains unchanged; the production composer appends its exact attachment reference. Before and after send, the harness independently reads the public same-source receipt and actual bytes, verifies SHA-256 and human attribution, and binds it to the real canonical comment. Extra markdown, text changes, foreign receipts and hidden followups fail. Attachment fingerprints pin both production composer source hashes as data, without importing the UI graph.

Six prerequisite probes are not completed journeys

The following six definitions retain explicit coverage gaps and must not be reported as full intended behavior coverage or passing refusal tests. Authored callbacks and calibrated hypothetical observations do not prove a live journey.

Scenario	Implemented evidence/action	Remaining prerequisite or implementation boundary
access.invite-teammate	Separate real active-invite creation and accepted-state checkpoints; consequential role-question branch; exact join-link hash binding; source-qualified guard before acceptance; actual intended-person membership and different-identity reuse rejection.	Supported agent invitation grant is unavailable through the current public membership mutation. Known unsafe invitation acceptance remains blocked by the shared source guard.
access.manage-membership	Actual initial project permission, unrelated company membership and instance-user baselines, with independent post-query readback and preservation of unrelated grants.	Public grant provisioning rejects agent principals; actual membership mutation is board-only and currently records a human actor.
activity.routine-audit	Real routine/revision/run/task reads and independent audit oracle.	Historical prerequisite construction is unfinished. No rule requires a real 24-hour wait; a deterministic historical input fixture requires a separately reviewed narrow fixture-policy decision. No historical seeding exception is implemented here.
agents.restore-instructions	Exact prior bytes and visible historical/agent-attributed restoration evidence, without demanding a new UUID or literal restore source label.	Current file storage does not create the two visible revisions or retained restoration history required by the catalog.
budgets-costs.subscription-fees	Public report/editability/ownership/current-price audit state capture and existing human-link UI; editable shared ownership is supported.	No admitted genuine same-account duplicate fixture/provenance or public archived-price readback exists. Candidate rows are never promoted to verified duplicates by matching labels/owners. Canonical setup remains explicitly blocked.
skills.enable-for-agent	Saved assignment, unchanged skill/policy, future native selected-version snapshot independently matched to exact public version bytes; natural assignment wording accepted.	Public configuration-grant provisioning remains blocked. Lifecycle policy does not implement runtime skills.use denial; no denial is fabricated from install/edit restrictions.

Queries, intended contexts/outcomes, declared run/human counts and deadlines stay unchanged. These source improvements do not rewrite earlier measurements or qualify any blocked conditional as a pass.

Selected private inputs and published operator sources

A case declares requiredCredentials by standard credential name and, where needed, externalInput by provider and required field names. The launcher resolves only that selected case into a mode0600 file outside diagnostic artifacts. The worker exposes input.privateInputs.credentials, .read(name) and .configuration("external-provider"); boundary callbacks retain their typed credentials map. Extra wizard/external keys are never inserted into the chat agent's secret references or inherited server environment. The launcher removes raw input files even when retaining recovery/diagnostic state. The definition snapshot retains only credential/reference/field names. Offline eligibility calls chatCatalogCaseFingerprint(descriptor, normalizedExecution, retainedNames, retainedInstalledIdentity) (or the boundary equivalent with its retained names), so recomputing an identity after cleanup never requires the raw credential file.

Only instance-operations.update-instance and instance-operations.provision-runtime can declare installedSource: "published-global-npm". They require private0600 PAPERCLIP_CHAT_INSTALLED_SOURCE_CONFIG with schemaVersion1, disposable true, source: {version, integrity}, and optional distinct target: {version, integrity}. Provision additionally names piCredentialEnv: "OPENROUTER_API_KEY" and declares that selected private credential. These are exact semver/sha512 pins, never a local source pack or arbitrary URL. During the governed attempt the launcher obtains registry metadata and actual public CLI/server tarballs, verifies their integrity, installs the source into its owned prefix, and records the complete dependency tree. The supervisor binds the actual CLI PID, version and CLI/server bytes before setup and after an allowed update restart. The operator receipt stays private0600; fingerprints bind public artifact hashes and the full owned dependency tree. Source-local refusal remains the default for every other ordinary chat case.

Frozen capability coverage

The campaign retains its original 244 scenarios and 191 declared capabilities. Merged production now also maps runs-adapters/model-accounts; that capability is outside this frozen catalog and remains uncovered and unqualified by it. The setup-health cell is diagnostic only and is excluded from those 244 scenarios and from --all, as is the explicit-only chat catalog.

The explicit setup-health diagnostic first makes one bounded, read-only OpenRouter current-key request on the trusted remote Linux job. Unrecognized credentials stop before agent creation or any model call. A recognized key permits the existing single ordinary lifecycle hello. The closed receipt records only HTTP status, recognition, expiry and key-budget availability; unlimited/unknown key budgets do not establish account credit or model access. Response bodies, labels, amounts, account identities and key values are never retained. There is no retry.