baphuongna/pi-crew/requirements-to-task-packet
Claude Code and agent skill by baphuongna/pi-crew
- community
Checking that the SKILL.md file is still in the repository.
Use when a goal, issue, roadmap item, review finding, or user request must become actionable worker tasks. When NOT to use: already-decided work that just needs execution; tasks so simple they fit one sentence.
Install
npx vskill@latest install baphuongna/pi-crew/requirements-to-task-packetSKILL.md
requirements-to-task-packet
Core principle: workers need explicit task packets, not inherited ambiguity. Ask only when ambiguity changes architecture, safety, public behavior, or data loss risk; otherwise record assumptions.
Distilled from detailed reads of clarification, spec-to-implementation, subagent-driven development, and skill-authoring patterns.
Clarify or Proceed
Ask before implementation when ambiguity affects:
- security boundary, permissions, ownership, or secret handling;
- destructive operations, migrations, publishing, or public API behavior;
- architecture or data model;
- acceptance criteria or rollback expectations.
Proceed with explicit assumptions when ambiguity is local, reversible, and testable.
Task Packet Template
Subagent Context Rules
- Give each worker fresh, curated context; do not rely on hidden parent history.
- Include exact upstream artifact paths and summaries when needed.
- Keep implementation tasks independent or explicitly sequenced.
- Require workers to report one of: DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, BLOCKED.
- For BLOCKED/NEEDS_CONTEXT, change context/model/scope before retrying.
Acceptance Criteria
Use observable checks:
- command output, state transition, UI/status text, artifact contents;
- regression tests or named test files;
- security properties such as containment/ownership/no secrets;
- compatibility requirements such as Windows paths or Pi CLI flags;
- rollback notes.
Spec Pairs (pi-crew v0.10.1+, ADR-6)
When the target runs pi-crew, author the SpecRecord + TaskPacket pair instead of prose-only acceptance:
- SpecRecord (workspace
state/specs/<id>.json):requirements[]with
must|should|could priority + stable ids; acceptance[] entries each tied to a requirementId. Machine-checkable acceptances carry command, expectedDigest (sha-256 hex of stdout) or expectedExitCode, and idempotent: true — only idempotent commands are ever re-run.
- PROVENANCE — critical: specs you (an agent/skill path) write are persisted
generated and NEVER re-executed by the orchestrator. manual+trusted (the only specs strict mode re-runs) are minted exclusively by USER-facing import actions — a worker cannot author a command the root re-executes.
- Wire the workflow: add
specRefs: [<spec ids>]to steps held to the spec;
specStrict: true in frontmatter opts the whole workflow into strict mode (requires a verifier role step — the run rejects at start otherwise).
- Executor footer contract: workers must END results with
``text SPEC-EVIDENCE: <acceptanceId>: <one-line evidence> ``
Non-strict = mechanical coverage only (unverified badge on gaps, never blocks). Strict = coverage AND machine-check; failures fail the run.
Enforcement — Requirements to Task Packet Gate
Before dispatching workers, verify task packet has:
- [ ] Objective clearly stated (goal in one sentence)
- [ ] Scope and paths defined (what is/isn't in scope)
- [ ] Allowed vs forbidden edits specified
- [ ] Inputs/dependencies and expected output artifacts listed
- [ ] Acceptance criteria are observable (command output, state transition, test)
- [ ] Verification commands provided
- [ ] Escalation conditions defined
If ANY answer is NO → Stop. Complete task packet before dispatching.
Anti-patterns
- Broad "fix everything" prompts.
- Buried assumptions.
- Expanding scope because context remains.
- Treating tests as proof when the requirement was never asserted.
Self-restraint
"Creating nothing is a valid result." If the evidence does not support a meaningful change, say so explicitly rather than inventing one. The next attempt may find stronger evidence; an invented change now damages trust in every future report.
"Creating nothing" here means the requirements are already actionable as-is — no packet needed. Inventing scope or splitting trivial work into packets adds overhead without value.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/requirements-to-task-packet/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/baphuongna/pi-crew/requirements-to-task-packet)