Build
Claude Code plugin marketplace — sdd (spec-driven dev) and gh (GitHub workflow) plugins
npx -y skills add kborovik/pilot-skills --skill buildAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Plan-then-execute impl vs SPEC.md. Triggers when user asks to build, implement, or execute spec or specific §T task. Phrasings: "build §T.<n>", "build --next", "implement next task", "run the build", "does the implementation run?", "is §T.<n> done?".
SKILL.md
3.9 KB, as published. Nobody here has run it
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. - Parse invocation args:
§T.n→ that task only--nextor empty → lowest-numbered row with status.--all→ 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).
Emit plan inline every task — transparency, not wait-state — then proceed to EXECUTE.
EXECUTE
Per task in order (status flips . → x direct):
- Edit code per plan.
- Run verification command.
2'. On staged diff touching PUBLISHED, probe
.claude/check-extras.mdand run audit recipes it ships; bail per recipe message every surviving match, not commit until match-free. Probe not match → no-op. - Pass (verification exits 0 and tests added per plan and not §V regressed via full-suite re-run) → flip §T.n status
.→x; auto-commit. Surface/sdd:checkas Next-block item #1 per §V.<n> (cascade scan; not silent close). Next task. - Fail → invoke backprop skill. Do NOT retry blindly. Status stays
..
FAIL → BACKPROP
On test/build failure:
- Read failure output.
- Classify failure: (a) code bug → step 3, (b) spec wrong → step 4, (c) unspec edge → step 4. Confident → proceed direct. Low-confidence (ambiguous or multiple plausible) → AskUserQuestion per decision-gate invariant w/ 3 options keyed (a)/(b)/(c), labels is action descriptions ("Code bug — fix and re-run", "Spec wrong — /sdd:spec bug:", "Unspec edge — /sdd:spec bug:"), header
Verify-fail class. - (a) → fix code, re-run. No spec change.
- (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 records bug-class precedent so recurrence-class blocked.
WRITE POLICY
- Only flip §T status. No other SPEC.md edits from build.
- Other spec edits → invoke spec skill.
- Auto-commit on
.→xper §T row; not user prompt. Message:T<n>: <goal line>+ §V cites. - Stage explicit
git add <listed-paths>per plan; notgit add -A— pre-existing dirty working-tree state not bundled. /sdd:build --allchains plan-once → every §T row {edit → verify → commit} autonomously.- FAIL → not commit; FAIL→BACKPROP runs first.
OUTPUT — "Next" block
Heading ## Next; 1–5 atomic items (one sentence each, no Reply prefix); positional dispatch (run <int> or run /<plugin>:<cmd> [args]). Optional ## Hint (≤ 3 lines) precedes when item selection needs hidden state. PLAN not wait-state so not execute/revise/abort items; EXECUTE-pass auto-commits so Next leads w/ /sdd:check post-§T-close cascade; after backlog cleared, swap /sdd:build --next for /sdd:spec seed.
Canonical example after EXECUTE pass (commit already auto-fired):
## Next
1. /sdd:check — cascade scan over the just-closed §T row
2. /sdd:build --next — start the next pending §T row
3. /sdd:spec amend §T.<n> — re-scope before continuing
Variant: backlog cleared (terminal state) → swap item 2 for /sdd:spec (seed new row), drop item 3.
NON-GOALS
- No progress dashboards.
cat SPEC.md | grep §Tis the dashboard. - No speculative work beyond chosen task scope.