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.
