worker-implement

Worker: Implement Vertical Slice

Command: pasture:worker:implement — Implement assigned vertical slice following TDD layers

-> Full workflow in PROCESS.md <- Phase 9

[wimpl-plan-backwards]

Given: vertical slice task
When: implementing
Then: plan backwards from production code path
Should not: start with types without knowing the end

[wimpl-vertical-ownership]

Given: production code path
When: implementing
Then: own full vertical (types → tests → impl → wiring)
Should not: implement only horizontal layer

[wimpl-import-production]

Given: tests
When: writing
Then: import actual production code
Should not: create test-only export or dual code paths

[wimpl-verify-production]

Given: implementation complete
When: verifying
Then: confirm production code path is wired (via code inspection or safe testing)
Should not: rely only on unit tests passing

[wimpl-inject-deps]

Given: dependencies
When: designing
Then: inject all deps for testability
Should not: hard-code new

[wimpl-validate-input]

Given: external input
When: processing
Then: validate with schema/validation tooling
Should not: trust raw input

[frag--validation-cases]

Given: any REQUEST (every request, not only fix-intent ones)
When: eliciting (URE), acceptance-testing (UAT), or implementing
Then: elicit concrete validation cases — a definition of done plus correct and incorrect behaviours (inputs/behaviors that must pass or must fail), confirm the case set with the user in UAT, evaluate the implementation against them, and store failing real-data cases as test fixtures
Should not: ship without validation cases; treat validation cases as applying to fix-intent requests only; introduce a request-type axis or enum to gate them

Task Recovery

The tracker backend is recorded once in .agents.local/pasture/EPOCH.md; read that file instead of re-probing. If EPOCH.md is absent, stop and ask the orchestrator to run detection.

Recover from live tracker state at session start and after compaction; never trust cached task rows. Read the rec: path from the header and use it as the one entry point: "${EPOCH_REC}" <op>. The operation set is closed: create, label, block, note, show, query, ready, status, close, whoami, signal.

Store selection: the adapter reads PASTURE_DB_PATH when set, otherwise the header's db:; the probe wrote that value from the same precedence (XDG_DATA_HOME/pasture/pasture.db, HOME/.local/share/pasture/pasture.db, then .pasture/pasture.db). Use the same selected store for every command. If a recovery query fails, report its actual store path, failed operation, impact, and permission/schema/configuration repair; failure is not an empty work queue.

The adapter passes --namespace only when EPOCH.md records namespace:. That pin wins on both backends. Otherwise it omits the flag. Pasture then derives the namespace itself: PASTURE_NAMESPACE, else the git remote, then file:// of the working directory. The ledger calls derive_namespace for the header and for every write that has no pin. That function is the one list of remote forms; do not restate it. With no origin, both backends use file:// of the working directory, so invocations from different subdirectories use different namespaces. Queries without a pin follow that backend's derivation; pass --all-namespaces to opt out. When neither a pin nor PASTURE_NAMESPACE is set, pasture reads are unscoped and creation falls back to the git remote, then file:// of the working directory. Inspect the returned full URI before choosing repository work. Separate reads are not an atomic snapshot.

Run "${EPOCH_REC}" ready, "${EPOCH_REC}" query --blocked, and "${EPOCH_REC}" query --status in_progress. For each referenced active task, run "${EPOCH_REC}" show "$TASK_URI", "${EPOCH_REC}" show "$TASK_URI" --notes, and "${EPOCH_REC}" show "$TASK_URI" --history. All *_URI variables in examples are inputs bound to full task URIs from an assignment, a verified tracker read, or the URI create printed; never use legacy short IDs. Resolve each variable before execution and quote it as one operand.

Create prints the new task URI directly. Persist it before any label or note: create-plus-label is non-atomic, so on failure retain the URI and retry only the failed label/note operation, never create again. Only open, in_progress, and closed are statuses; blockage is represented by live dependencies and notes, not a blocked status.

An owner is recorded with "${EPOCH_REC}" label "$TASK_URI" owner:<role>. That is a claim, not an allocation: an unassigned task stays unassigned until the supervisor allocates it, and repeating an identical claim is not allocation. Never invent an owner.

Parent stays open and is blocked by child: "${EPOCH_REC}" block "$PARENT_URI" --by "$CHILD_URI". Reference documents belong in description frontmatter under references:, never fabricated blockers. Workers report evidence and handoffs without closing their slices or leaves; the supervisor closes only after independent review and satisfied gates. Use git agent-commit for local commits. Never install, enable, or modify Git hooks, Git hook path configuration, or pre-commit integration without explicit user approval. Tracker writes are durable; Git history lands separately.

If the header records backend: error, stop before the first mutation: the store exists but cannot be used, and falling back could lose earlier epochs. A mid-epoch write failure retries 3 times. After the third failure, write the attempted record to .agents.local/pasture/pending/<timestamp>-<op>.md, keep the recorded backend, and surface the pending record to the user at the next gate. Never switch backends and never drop the record.

Task Attribution

Select the current registered author with "${EPOCH_REC}" whoami. It prints the author recorded in the header or PASTURE_AUTHOR, and on the pasture backend it confirms that actor with agents show. It does not list agents and it does not invent one. If neither source names an author, it stops: set author: in EPOCH.md or export PASTURE_AUTHOR. Never auto-register or guess an identity during recovery. Every note supplies --author explicitly; machine examples use the pre-registered pasture-system/00000000-0000-0000-0000-000000000000 actor. If it is absent, stop and request registration from the operator rather than claiming registration is fixed. Preserve human intent verbatim and quote historical attribution as evidence, not as a forged new author.

When to Use

You have a Pasture task ID for a vertical slice and are ready to implement end-to-end.

Steps

Step 0: Plan backwards from production code path (before implementing)

Given Pasture task when starting then identify production code path first

"${EPOCH_REC}" show "${TASK_ID_URI}" --notes
# Look for: "productionCodePath": "cli-command subcommand" or "api-endpoint"

Trace backwards through call stack:

End: User runs production command
  ↓ Entry: CLI command.action(...) or API endpoint handler
  ↓ Service: createXService({ deps }).method(...)
  ↓ Types: InputType → OutputType

Identify what you own in each layer:

L1 Types: Which types does YOUR slice need? (not other slices)
L2 Tests: Import actual production code (CLI/API), not test-only export
L3 Implementation: Service method + wiring with real dependencies (not TODO)

Step 1: Read Pasture task for full context

"${EPOCH_REC}" show "${TASK_ID_URI}" --notes

Step 2: Update status

"${EPOCH_REC}" status "${TASK_ID_URI}" --set in_progress

Step 3: Implement your vertical slice in layers

Layer 1: Types (your slice only)

Create only types YOUR slice needs
Don't add types for other slices

Layer 2: Tests FIRST (import production code)

Write the tests before the implementation. The tests ARE the executable verification of the validation-case contract agreed with the user during URE and Plan UAT (the universal validation cases — see [frag--validation-cases] and C-validation-cases): definition of done plus correct/incorrect behaviours.
Import actual CLI/API package: import "myproject/cmd/feature"
NOT test-only handler: ~~import "myproject/internal/testhelpers/feature"~~
Tests will FAIL initially — red-first is expected (no impl yet). As you implement Layer 3, progressively fewer tests fail until all are green.

Layer 3: Implementation + Wiring

Service method for your slice
CLI/API wiring with real dependencies: NewService(ServiceDeps{ FS: fs, Logger: logger })
NOT TODO placeholders: ~~// TODO: Wire service~~

Follow:

validation_checklist items
acceptance_criteria (BDD Given/When/Then)
tradeoffs from ratified plan

Step 3b: Evaluate against validation cases (R6, every request)

For every request (not only fix-intent ones), the URE/UAT captured concrete validation cases — the definition of done plus the correct/incorrect behaviours that must pass or must fail. Per [frag--validation-cases] and C-validation-cases:

These validation cases are the contract you wrote your Layer-2 tests against (tests-first); evaluate the implementation against each confirmed case.
Store the failing real-data cases as test fixtures so the behaviour is locked in.
The slice is not done until its validation cases pass (red → green).

There is no request-type axis or enum gating this — what a request needs is recognized from the REQUEST/URD, not classified.

Step 4: Verify quality gates

Type checking passes
Tests pass

Step 5: Commit safely in a shared worktree

Stage only the files belonging to your slice, by name:

git add cmd/feature/list.go pkg/feature/service.go pkg/feature/types.go
git agent-commit -m "feat(feature): add list subcommand"

Never use git add ., git add -A, or git commit -am ... — they sweep peer-worker WIP into your commit.

Never use destructive git operations (git reset --hard, git checkout HEAD -- <path>, git stash pop, git stash apply, git clean -fd, git branch -D) on the shared worktree. A PreToolUse hook blocks these for worker agents; if you find peer work in your way, post "${EPOCH_REC}" note and wait for supervisor coordination instead. See Shared-Worktree Git Discipline in /pasture:worker for the full rationale and the escape hatch.

Checklist

[ ] Planned backwards from production code path
[ ] Read Pasture task for validation_checklist
[ ] Each validation_checklist item satisfied
[ ] BDD acceptance_criteria met
[ ] Tests import actual production code (not test-only export)
[ ] No dual-export anti-pattern (one code path for tests and production)
[ ] No TODO placeholders in production code
[ ] Service wired with real dependencies (not mocks in production)
[ ] Quality gates pass (type checking + tests)
[ ] Production code path verified (via code inspection: no TODOs, real deps wired, tests import production code)
[ ] Files staged individually by name (no git add . / git add -A)
[ ] No destructive git operations (reset --hard, checkout HEAD -- <path>, stash pop/apply, clean -fd, branch -D) used on the shared worktree

Follow-up slices

If your Pasture task is a follow-up slice, the implementation procedure is identical. Additionally:

Check for a "DEFER'd-Item Leaf Tasks" section in "${EPOCH_REC}" show "${TASK_ID_URI}" — these are user-DEFER'd UAT items you must resolve
Your implementation must address each DEFER'd-item leaf task's acceptance criteria
On completion, report which DEFER'd-item leaf tasks were resolved

Review-Fix Cycle (within the chosen review-effort budget)

Your slice is not finished when the first pass lands. Code review runs in two phases. During a wave, only an open BLOCKER finding gates progress; IMPORTANT and MINOR findings stay open. After the last wave, convergence queries every open pasture:finding across all waves and targets 0/0/0 within maxRounds. Stay available to fix findings of every severity — IMPORTANT and MINOR reach the convergence round too, not just BLOCKER. Findings close only per C-finding-lifecycle; a worker cannot close one. Iterate up to the review-effort budget chosen at Phase 8. If the budget is exhausted before convergence, the outstanding findings are surfaced to the user at a gate (not proceeded-past silently). Do not treat "tests pass once" as wave completion.

Complete: /pasture:worker-complete
Blocked: /pasture:worker-blocked

01 前言 Front Matter

03 SKILL.md 正文