Junhanliu-dev/espalier-engineering / skills/espalier-init
espalier-init
VerifiedAnalyze any existing codebase, discover its patterns and best practices, and generate an Espalier structure (rules, skills, wiki, pipeline) so AI coders produce code matching that project's language and conventions. Greenfield repos take the two-pass Decide-Then-Bind path (/espalier-map charts the conventions, init binds them); boilerplate repos run init normally, then map the product. Wires into Claude Code, Codex, and/or GitHub Copilot (.claude/, .agents/skills/ + .codex/, .github/).
templates/agents/modes/fix-round.md
View on GitHub# Mode: fix-round
Read by `harness-coder` when its prompt carries `FIX ROUND {n}:` — a Stage 4 re-spawn on panel findings. Named on that prompt line by the orchestrator;
not loaded otherwise. Verbatim procedure — the agent body keeps the heading
and a one-line pointer; nothing here is a budget.
## Fix Rounds: Fix the Class, Not the Instance
When your prompt carries `FIX ROUND {n}:` you are being re-spawned on review
panel findings. Field data: most second and third panel rounds find the SAME
defect one hop away from the line just fixed — the reviewer named one
instance, the coder fixed that instance, the sibling survived. Each such
round costs a full 2-agent panel. Close the class in ONE round:
1. **Name the class.** For every P0/P1 finding, write one line stating the
property that was violated, not the line that violated it — e.g.
"generated `*WhereInput` exposes a hidden reverse relation as filterable",
"money columns recomputed outside the acceptance transaction",
"client-supplied return URL reaches a navigation call unvalidated".
2. **Enumerate the siblings.** Search for every other place the same
construct occurs: the same helper / decorator / access pattern / generated
surface across every touched layer AND the generated artifacts it feeds
(schema output, route tables, barrels). Record the search you ran and the
occurrence count — the reviewer re-runs it.
3. **Fix every sibling inside the change's scope in this round** — scope =
the layers this change touches plus the generated surfaces they feed. For
each sibling you leave, name it and say which: (a) NOT an instance of the
class — one reason, checked by the reviewer; or (b) OUT OF SCOPE — an
instance in a layer this change never touched. List (b) under
`- Out-of-scope siblings:`; the orchestrator files them as a follow-up
change. Never widen a feature change into a repo-wide refactor to close a
class — Solution Selection Ladder and "Modify files outside the task
scope" still bind. Each sibling you DO fix gets its own read: same class
does not mean same fix — a sibling with a different transaction boundary,
actor, or caller may need a different change or none.
4. **Re-check the seeds on the new code**: the fix's own external calls,
list reads, and error paths still follow Production-Aware Coding, and the
fix has its own test (folded mode) — a fix round is the most likely place
for a new defect to enter.
Append to your coding report, one block per P0/P1 finding:
```
### Class Sweep
- Finding: {P-sev} {≤80-char summary as the panel worded it}
- Class: {one line — the violated property}
- Search: {command or scope you enumerated with}
- Occurrences: {N} — fixed: {list}; not affected: {list — one reason each}
- Out-of-scope siblings: {list — layer never touched by this change; or "none"}
```
A fix round whose report carries no `### Class Sweep` block for a P0/P1 is
an instance-only fix; the reviewer files it as a P1 and the round repeats.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |