Sheepdog loop
Parent-orchestrated review/fix/review loop for boundary-law violations in a diff or target. Use when asked to run sheepdog loop, fix architecture drift until clean, or desheepdog a god orchestrator. Edits only after accepted fixes.From its SKILL.md
npx -y skills add filipores/sheepdog --skill sheepdog-loopAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 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.
SKILL.md
1.2 KB, 223 tokens by cl100k_base, as published. Nobody here has run it
Sheepdog Loop
Loop until no blocking boundary leaks remain or the cap is reached. Default cap: 3 rounds.
Loop
- Run sheepdog-review on the diff/target.
- Synthesize only blocker/major findings worth fixing now.
- Apply the smallest safe moves: move responsibility home, keep callers thin, add the smallest boundary check.
- Run focused tests/static checks.
- Review again if the fix materially changed structure.
Stop and ask the user when:
- a fix needs a product/scope/architecture decision
- the only path is accepted boundary debt
- the next move is broad refactor unrelated to the requested change
- max rounds reached
Accepted debt
Only with explicit user approval:
# sheepdog: accepted boundary debt, <reason>; home: <module/layer>; trigger: <when to move>
Final output
sheepdog-loop: clean | stopped | debt-accepted | cap-reached
changed: <files>
checks: <commands>
remaining: <none or blockers>
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.