plan
Use this skill when decomposing complex work into verifiable tasks and dependencies before or during implementation. Activates on mentions of write a plan, create a plan, break this down, task decomposition, implementation plan, what are the steps, plan the work, spec this out, or decompose this feature.
hyperb1iss · 已有静态检查记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add hyperb1iss/hyperskills --skill plan技能说明
name: plan description: Use this skill when decomposing complex work into verifiable tasks and dependencies before or during implementation. Activates on mentions of write a plan, create a plan, break this down, task decomposition, implementation plan, what are the steps, plan the work, spec this out, or decompose this feature.
Structured Planning
Build a plan that another session can execute and verify without rediscovering the decision. Specify outcomes and dependencies tightly enough to prevent drift while leaving implementation choices to the code and evidence encountered during execution.
The user's instructions take precedence over this skill's guidelines. Planning does not add an approval requirement. If execution is already authorized, continue after the plan is ready; if the user requested only a plan, deliver the plan without starting implementation.
Size the Plan by Uncertainty
| Work shape | Planning depth |
|---|---|
| Clear, reversible change with a direct check | Execute with a brief mental or inline plan |
| Several dependent behaviors with known patterns | Record tasks, interfaces, and acceptance checks |
| Uncertain architecture or migration | Resolve decisive unknowns and plan a representative vertical slice |
| Work spanning agents, sessions, or environments | Preserve ownership, state, dependencies, and resumption instructions |
File count is a poor proxy for risk. A one-line authorization change can need more analysis than a mechanical rename across many files. Plan enough to expose the uncertain or irreversible parts, not enough to predict every edit.
Establish the Actual Starting Point
Inspect repository status, relevant diffs, and the code path under discussion. Locate an existing implementation pattern and the real verification commands in scripts, hooks, or CI. Read relevant project memory through the installed Sibyl skill when configured.
Record the facts that constrain execution:
- The observable user or system outcome.
- The current baseline and revision or environment being inspected.
- The interfaces, ownership, and compatibility obligations affected.
- The uncertainty that could invalidate the proposed work.
- The explicit constraints, non-goals, and user corrections.
Treat an inherited plan as a hypothesis. Recheck consequential claims against the current tree before executing them. Distinguish verified facts, assumptions awaiting a check, and obsolete premises. Update the plan itself when a premise fails; a detached comment leaves the executor following stale instructions.
Choose a Useful Slice
Look for existing capabilities that can satisfy the request before decomposing a new mechanism. Simplification must preserve the requested result. Do not replace a difficult product goal with a convenient demonstration of plumbing.
Prefer a vertical slice through the uncertain boundary when it can prove feasibility early: a real input passing through the changed logic to the consumer that needs it. Plan migrations around compatibility and deployment order. Source-file order does not determine safe rollout order.
Keep implementation and its behavioral verification in the same accountable task. A title containing "and" is not evidence that the task should split. Separate tasks when they have independently reviewable outcomes, distinct owners, or real dependency boundaries. A test-infrastructure prerequisite may be separate; feature tests should not become an orphaned follow-up.
Give Each Task an Executable Contract
Use only the fields the task needs:
| Field | Content |
|---|---|
| Outcome | The behavior or artifact delivered |
| Scope | Known files/modules and permitted expansion rules |
| Dependencies | Required inputs or interfaces, with their producer |
| Verification | Exact runnable check plus the behavior it proves |
| Completion evidence | Expected artifact, assertion, observation, or report |
| Risk and recovery | Compatibility, rollback, or unresolved external dependency when material |
Example:
### Task: Reject expired sessions at the request boundary
Outcome: An expired session receives the established unauthorized response.
Scope: The session validator and its request-level tests.
Depends on: The existing clock injection interface, confirmed in the repo.
Verify: [repository's exact focused test command].
Cases: Expired, valid, and boundary-time sessions preserve expected behavior.
Completion: Tests exercise the request path and report executed cases.
Prefer real file paths where known; label proposed paths instead of pretending exploration is complete. A copyable command still needs an assertion: "renders" does not prove that the generated policy matches the intended resources.
Map Dependencies and Ownership
Parallelize independent outcomes. File separation helps avoid collisions but does not prove independence: two modules can share a schema, generated file, test database, port, or output directory. Name those shared resources and their owners.
Choose dependency-ready work rather than forcing every task into synchronized waves. A wave is useful when tasks share a meaningful integration checkpoint. Designate an integration owner and verify combined behavior after branches or patches meet. Consult orchestrate when delegation adds useful capacity and current instructions permit it.
Review the
阅读完整来源 ↗