build
The "build" skill defines a single-threaded plan-execute process for an AI agent to implement specifications. It involves reading specs, parsing tasks, formulating detailed implementation plans (citing invariants, interfaces, files, and tests), then executing code modifications and running verifications. Upon failure, the agent triggers a backpropagation mechanism to analyze the root cause, deciding whether to fix the code or update the specification, ensuring each commit is verified and adheres to a strict policy.
npx skills add https://github.com/juliusbrussee/cavekit --skill buildBefore / After Comparison
1 组Manual developer process for task planning, coding, testing, and debugging based on specifications, which is time-consuming and error-prone, leading to longer task completion cycles and inconsistent quality, averaging 4 hours for a medium-complexity task.
AI agent automates the specification implementation workflow, including intelligent planning, code modification, automatic verification, and error backpropagation, significantly reducing task completion time and improving code quality and consistency, completing tasks in 1 hour.
build — implement spec
Single-thread native plan→execute. You are main Claude. No swarm.
LOAD
- Read
SPEC.md. If missing → tell user to invoke the spec skill first. Stop. - Read
FORMAT.mdonce if not loaded. - Parse invocation args:
§T.n→ that task only--next→ lowest-numbered row with status.or~--allor empty → every.row in §T order
PLAN
Native plan mode. For chosen task(s):
- Cite every §V invariant that applies. Plan must respect all.
- Cite every §I interface touched. Plan must preserve shape.
- List files to create / edit.
- List tests to add or update (one per invariant touched).
- Name verification command (test, build, lint).
Show plan. Wait for user OK unless auto mode.
EXECUTE
Per task in order:
- Flip §T.n status cell
.→~. Just write to SPEC.md. - Edit code per plan.
- Run verification command.
- Pass → flip
~→x. Next task. - Fail → invoke backprop skill. Do NOT retry blindly.
FAIL → BACKPROP
On test/build failure:
- Read failure output.
- Ask: is failure (a) my code bug, (b) spec wrong, or (c) unspecified edge case?
- If (a) → fix code, re-run. No spec change.
- If (b) or (c) → invoke spec skill with
bug: <cause>first, let it update §V and §B, then resume build against updated spec.
Rule: never silently fix root-cause without considering backprop. §B is the memory that stops recurrence.
WRITE POLICY
- Only flip §T status. No other SPEC.md edits from build.
- Other spec edits → invoke spec skill.
- Commit after each §T completes. Message:
T<n>: <goal line>+ §V cites.
VERIFICATION
Task x only if:
- Verification command exits 0.
- New test(s) added per plan.
- No §V invariant regressed (run full test suite at end).
NON-GOALS
- No sub-agents. No parallel workers. Main thread only.
- No progress dashboards.
cat SPEC.md | grep §Tis the dashboard. - No speculative work beyond chosen task scope.
User Reviews (0)
Write a Review
No reviews yet
Statistics
User Rating
Rate this Skill