[Workflow] Use when E2E tests need to be written, updated or fixed: resolves the scope, authors tests only when needed, then verifies them through the configured runner in a bounded fix/retest loop.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
Read the registry → workflows.workflow-e2e together with this skill. Call /start-workflow workflow-e2e to resolve the selected mode, pre-actions and fingerprint.
Todo FIRST: before triage, analysis or step execution, create one todo per phase of the selected mode, in order, listing every occurrence of that phase as a line, including conditional/optional ones (a mode that declares no phases gets one todo per occurrence); preserve occurrence IDs, roles and barrier groups. Use native todo tools or an equivalent persistent ledger. Then mark the first todo in_progress; close each step line with its run or skip evidence, and complete a phase todo only when every line is closed and its applicable exit gates hold.
Mode selection and registry loading prepare tracking; any 'first action' below means the first substantive action after this bootstrap.
Call each step skill: read its linked SKILL.md and execute its protocol through the active host with the registry args. Reading or naming a skill alone is not execution. Required gates and core/optional flex follow the linked start-workflow Step Execution Protocol; keep this skill's quality gates, loops and evidence requirements.
Conditions: load each occurrence's registry applicability.when and skipReason verbatim, and record its run/skip evidence through the linked start-workflow protocol. Do not silently omit a todo or turn a conditional skill into an unconditional call; close, in one batch at triage, the todos whose condition is already false there. Declared parallel groups retain their all-return barrier.
Explicit step-skill calls by mode (registry order; roles and conditions remain owned by the registry):
default: /investigate (optional; conditional) → /e2e-test (optional; conditional) → /e2e-test --mode=verify --fix-loop (gate) → /docs-manager --mode=update (optional; conditional) → /workflow-end (optional; conditional) → /watzup (optional; conditional)
Phase todos: Discover (1) → Build (1) → Verify (2; exit tests-pass) → Close (2; exit run-closed)Regenerate this block with node .claude/scripts/lib/workflow-skill-contract.cjs --write after registry edits; /sync-codex refreshes it before mirroring.
<!-- PROMPT-ENHANCE:STEP-TASK-ANCHOR:END -->[BLOCKING] Workflow steps follow the guided contract in
/start-workflow→ Step Execution Protocol:gatesteps are fixed; other steps may flex with a logged reason. [BLOCKING] Before each step or sub-skill call, update todo tracking: setin_progresswhen step starts, setcompletedwhen step ends. [BLOCKING] Every completed/skipped step MUST include brief evidence or explicit skip reason. [BLOCKING] If Task tools are unavailable, create and maintain an equivalent step-by-step plan tracker with the same status transitions.
Goal: One E2E lifecycle for every request — resolve the source and scope, author or update tests only when needed, then converge the fixed scope to an honest green through the configured runner. E2E quality/visual review is capped at one pass per verification attempt; execution/fix/retest rounds remain a separate bounded convergence mechanism.
Use this to write, update, run, verify or fix E2E/browser/user-flow coverage. Use workflow-integration-test (--mode=green / --mode=write) for integration tiers, /e2e-test --mode=verify alone for a one-shot report-only check, and workflow-bugfix when the task is a product bug whose E2E test is only the proof.
IMPORTANT MANDATORY Steps: /investigate -> /e2e-test -> /e2e-test --mode=verify --fix-loop -> /docs-manager --mode=update -> /workflow-end -> /watzup
Step contract: steps follow /start-workflow → Step Execution Protocol — gate steps always run, a step that runs invokes its Skill tool, and every other deviation is logged. NEVER batch-complete validation gates.
--source)--visual-review)Before choosing steps, classify the target and record the result in the run report:
--source (below) and --visual-review (below). State both before any step.whole run (hundreds of cases). Size sets batching and delegation, never whether the green gate runs.prompt|context|whole with no expected edits) · product fix expected (a failing journey on changed behavior) · data/auth/environment-sensitive (seed, accounts, external services).The triage picks which recommended skills run and how deep. An unresolvable source or scope stops with an ambiguity record — never a silently chosen maintenance mode.
| Gate | Evidence that proves it |
|---|---|
tests-pass — the fixed scope is green in THIS run | /e2e-test --mode=verify --fix-loop report: exact runner output (counts, failing names, exit status), the configured consecutive fresh green runs, Round Integrity Check pass, no scope shrink, no skipped/weakened/deleted test |
| Case traceability | Each selected owner-qualified case/variant (or strict-default TC/§8 when no native specArtifacts profile applies) maps to an actual test, assertion and observed result; ID presence alone is not proof; unresolved links stay UNVERIFIED |
| Honest verdicts | Every failure is classified SOURCE-WRONG / TEST-WRONG / TEST-NOT-OPTIMAL / ENVIRONMENT-BLOCKED / AMBIGUOUS before any edit; a fix lands at the owning layer and is reviewed |
| Explicit acceptance | No snapshot, baseline, fixture or expectation changes without a named acceptance record; otherwise ACCEPTANCE-PENDING |
| Visual (when it applies) | One review pass per verification attempt: every required capture opened/read case by case via /experience-review --rounds=0; blocking findings return to the execution/fix/retest loop. A false value cannot waive a required gate |
| Docs synced (when the run changed tests, baselines, product source or documented behavior) | /docs-manager --mode=update triage result with terminal evidence |
run-closed (top-level run) | /workflow-end verifies every gate above; a nested run is closed by its parent workflow, whose /workflow-end checks this run's tests-pass evidence |
N/A needs evidence: E2E is APPLICABLE only with a configured framework, runner and command; a relevant but unavailable capability is ENVIRONMENT-BLOCKED, never green.
| Skill | Earns its cost when | Proves / feeds |
|---|---|---|
/investigate (conditional) | Scope, runner, case profile or test organization is not already evidenced in this session; M+ size | Applicability, case map, commands, data strategy |
/e2e-test (e2e-author, conditional) | --source is changes, recording or update-ui | Authored/updated artifact + exact scope handed to the loop |
/e2e-test --mode=verify --fix-loop (gate) | Always | tests-pass, verdicts, visual gate, fresh reruns |
/experience-review --rounds=0 | Called by the loop when visual review applies | Per-capture records, blocking vs advisory findings |
/docs-manager --mode=update (conditional) | The run changed tests, baselines, product source, specs or documented behavior | Docs synced |
/workflow-end (e2e-end, conditional) | Top-level run only; a nested run is closed by its parent workflow | run-closed |
/watzup (e2e-watzup, conditional) | Top-level run only; a nested run leaves the recap to its parent | Handoff summary |
/test is not a second top-level run and no separate visual-fix loop exists — /e2e-test --mode=verify --fix-loop is the single convergence and remediation owner after preparation. Skipping a recommended skill is fine when triage shows it does no real work; log it as a deviation (intent-skip / when-false) with evidence.
Conditional step notes (registry skipReason, verbatim):
/investigate — skip when: The parent run or this session already evidences the scope, runner, case map and test organization./e2e-test — skip when: Resolved --source is prompt, context, or whole; e2e-test --mode=verify --fix-loop owns scenario selection or generation for that source./docs-manager --mode=update — skip when: Verification-only run: no test, baseline, product source, spec, or documented behavior changed, so docs-manager --mode=update has nothing to record; cite the unchanged diff and the terminal verify report./workflow-end — skip when: Nested inside a parent workflow, whose own workflow-end closes the run and checks this workflow's tests-pass evidence./watzup — skip when: Nested inside a parent workflow (workflow-feature, workflow-bugfix, workflow-big-feature, workflow-implement-spec); the parent's tail (workflow-end then watzup) owns the whole run's recap, so this nested run skips its own workflow-end and records its recap as covered by the parent workflow's watzup instead of printing a second one.--source)An explicit value wins. When omitted, infer in order: recording JSON present → recording; explicit whole-project request → whole; run/verify/QC a feature, bugfix, journey or current context → prompt or context; UI/style/template diff with baseline intent → update-ui; confirmed code/spec/API change → changes. No evidenced signal → stop with an ambiguity record.
--source | Use when | /e2e-test mode it drives |
|---|---|---|
changes | Test specs, source or API changed and E2E coverage must follow | from-changes / from-spec: update only cases the changed intent requires |
recording | A browser recording JSON exists and a runner test is wanted | from-recording: convert with the project's configured recorder/converter (else author by hand), add outcome assertions, follow the configured test organization |
update-ui | UI changed and screenshot baselines need review | update-ui: collect candidate evidence; keep accepted expectations until an explicit acceptance record |
prompt / context | A feature, bugfix, journey, QC request or current context defines the scope | none — the loop selects or generates cases |
whole | Project-wide E2E/QC verification | none — the loop runs the configured full suite |
Authoring sources run only a focused authoring check at most; the final full green run belongs to the loop. Every source hands the loop one exact scope plus its case map.
--visual-review)Resolve independently from an explicit value, the task request and the project contract. true (explicit or project-required) on an applicable surface: capture the project-declared state × viewport matrix per the resolved uiStateCapture.mode, index captures, and have /experience-review --rounds=0 read every required capture in exactly one review pass for that verification attempt. Validated BLOCKING findings join the execution failure set and return to owning-layer repair; a fresh rerun may collect fresh evidence but never starts a second review pass for the prior attempt. ADVISORY findings are recorded, never looped on. A false value cannot waive a project-required gate. A non-visual scope records N/A.
Read .claude/skills/shared/e2e-quality-protocol.md before authoring or verification, and .claude/skills/shared/ui-state-capture-protocol.md when visual review applies; they own the scenario/invariant, isolation, auth, wait, evidence, cleanup and capture rows. This workflow owns only sequencing.
Resolve one record before the first authoring or verification step and carry it to every step (the loop owns execution evidence; /docs-manager --mode=update receives the terminal result):
| Field | Required content |
|---|---|
applicability | APPLICABLE with evidence of a configured framework, runner and command; otherwise N/A — <evidence>. Skill-local browser helpers are not project evidence |
caseMap | Configured owner/case/scenario/variant + evidence fields, or strict-default TC/§8 when no native profile applies; a malformed declared profile blocks and never falls back |
fullCommand / focusedCommand | Copy-ready configured commands; invalid or zero-match selections exit non-zero |
runIdentity / dataStrategy | Unique non-sensitive run identity, public-path setup, declared seed/accumulation mode, parallel-worker isolation |
testOrganization | The declared local pattern and its canonical locator/action/wait owners; never require a Page Object Model or abstract base the project does not select |
result | Exact counts, failing names, exit status, repeat evidence; one fresh green full run, or the project's configured consecutiveGreen count |
Browser journeys follow observe → act → observe with the configured runner's native waits or an evidenced project helper; action pacing applies only when the project contract configures it and never substitutes for a real settle signal.
You choose inline vs sub-agent, batching and wave layout, optimizing wall-clock and token cost at equal quality. Fixed constraints only: a test exists before it is verified; fixes are re-verified after they land; /workflow-end runs last in a top-level run; nested in a parent workflow, this run records its evidence in the parent's Goal Contract instead of creating a second one, and the parent closes the run; gates awaiting user approval never run in parallel; each loop round runs the default verify pass inline (never a nested --fix-loop). Recommended: XS/S inline without sub-agents; L/XL partition authoring and failure triage per module or feature batch (one report per batch), while the green gate still runs the full fixed scope; delegate authoring to the e2e-runner agent when the batch is large.
tmp/reports/ FIRST, then append per step, round and batch; visual records are appended one capture at a time.TaskList before continuing; sub-agent briefs make report writing their first deliverable and return only the summary envelope.Owned by /e2e-test --mode=verify --fix-loop; this workflow only enforces its bounds. Classify before editing, trace root cause with /investigate --mode=debug, fix at the owning layer with /fix, review the round's diff with /changes-review — nested in a parent workflow whose review gate re-runs after this step, that per-round review defers to the parent's re-review, which receives the round's changed files — run the Round Integrity Check every round, then rerun fresh over the same scope. Converge on one fresh green run (default 1; the project may configure more consecutive greens) within the round cap (default 3). A non-shrinking or rising failure count, a cap hit with failures open, scope shrink or test loss → NOT-CONVERGED and escalate via ask user question tool with exact evidence. Never weaken, skip, narrow, delete, silence or auto-accept to obtain green.
Protocol guides — Apply each protocol only when its read-when condition holds. If a protocol's text is not in your context, read its file below before you act on it.
e2e-visual-design-contract — Evidence and baseline rules for visual review in E2E and human QC; handling visual-review evidence or visual baseline updates → .claude/skills/shared/protocols/e2e-visual-design-contract.mdincremental-persistence — Persist results per file or section while the work proceeds; a sub-agent or heavy step processes more than three files → .claude/skills/shared/protocols/incremental-persistence.mdsession-goal-ledger — Keep the original goal and every user prompt of the session; running a long or multi-prompt session → .claude/skills/shared/protocols/session-goal-ledger.mdsubagent-return-contract — Sub-agents return a structured envelope and a report path, never an inline report; spawning a sub-agent → .claude/skills/shared/protocols/subagent-return-contract.mdtest-architecture-execution-contract — Testability as an architecture condition: required test types and execution modes; setting up or reviewing a test architecture → .claude/skills/shared/protocols/test-architecture-execution-contract.mdworkflow-registry-binding — Read the workflow registry entry and the workflow skill together, since they must agree; executing or editing a workflow → .claude/skills/shared/protocols/workflow-registry-binding.mdMUST ATTENTION Each assertion-bearing test names the behavior or technical contract it protects and asserts an outcome it owns. Use the project's configured/native test format — Given/When/Then is one valid format, never a framework-wide requirement. When specArtifacts is valid, link configured owner/case/variant identity and intent/contracts evidence; when absent, name the guarded business intent or technical contract. A malformed declared profile BLOCKS without fallback. Broad test-format migration → assign an owner and next step, never rewrite cases outside scope.
MUST ATTENTION Before implementation record evidence-backed applicability for the test types and modes the task/project contract requires: copy-ready full and focused commands where available, zero-match behavior, a useful platform-appropriate entry point, supported execution modes and environments, state-isolation requirements, exact results, and repeat evidence where persistent state makes it relevant. Exercise claimed modes; report a missing required capability as ENVIRONMENT-BLOCKED. Never invent production targets or impose a test format. Browser/UI E2E uses the configured runner's waits or project helper for observable readiness and outcomes; apply action pacing only where the project contract specifies it. Reuse the project's evidenced test organization — require a POM, base class, or component taxonomy only when the project actually selects it.
MUST ATTENTION visual E2E/QC resolves the project design authority first, applies project design-system and frontend decisions plus applicable UI-*/DD-*/CL-* roles, records component ownership using the project's taxonomy or observed boundaries, sends static source findings to /ui-design --mode=review and runtime image evidence to /experience-review, captures states and transitions required by the configured evidence contract, reloads the convention docs then reads and records each required capture before synthesizing findings with coverage gaps, treats UIX/UI/accessibility-floor findings as blocking and UIX-POLISH/DD identity as advisory, never invents measurements, and never auto-promotes baselines; non-visual runs state N/A.
Task Planning Rules: pin Original goal:, keep User prompts this session: P1…Pn, and map the result to every prompt before claiming done; full text: .claude/skills/shared/protocols/session-goal-ledger.md.IMPORTANT MUST ATTENTION Goal: one E2E lifecycle — resolve --source and visual mode, author only when the source needs it, then converge the fixed scope to an honest green through /e2e-test --mode=verify --fix-loop or escalate with exact evidence.
tests-pass = exact runner output for the fixed scope, configured consecutive fresh green runs within the round cap, Round Integrity Check pass — never weaken, skip, narrow, delete or auto-accept a test or baseline to get there./experience-review --rounds=0; a false value cannot waive a project-required gate.tmp/reports/ first, keep one phase task per phase, one row per selected step, re-read both after compaction; /workflow-end runs last in a top-level run, and a nested run is closed by its parent.No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
[Documentation] Use when generating, smart-merging or refactoring the root AI context (CLAUDE.md, AGENTS.md) and its Claude/Codex projections. Config file only: project-config.
The English summary is being prepared. Showing the original description.
[Code Quality] Use when a workflow step or the user asks for AI code or plan review: LLM calls, prompts, agents, RAG, evals and guardrails. --mode={code|plan}; --report-only is read-only.
The English summary is being prepared. Showing the original description.
[Architecture] Use when a workflow step or the user asks for architecture design, compliance review, scalability grading or whole-project audit via --mode={design|review|scalability|full}.
The English summary is being prepared. Showing the original description.
[Content] Use when a workflow step or the user asks for product/feature ideation with ranked opportunities and validation tests. --mode={roadmap|scope}.
The English summary is being prepared. Showing the original description.
[Content] Use when a workflow step or the user asks for business viability: Business Model Canvas, financials, risks and go-to-market recommendation.
The English summary is being prepared. Showing the original description.
[Code Quality] Use when a workflow step or the user asks for a focused code-change review: staged, unstaged, branch diff or simple pull request. --fix-loop fixes and re-reviews. Complex pull request: workflow-review-changes; rationale: why-review.
The English summary is being prepared. Showing the original description.