[ LA TECHNIQUE · 36 skills ]
A skillet full of agent skills.
Production-tested workflow skills for AI coding agents: repo audits with leverage-sorted backlogs, bug hunting, security sweeps, publish gates, releases, and memory handoffs. Plain SKILL.md files for Claude Code and compatible harnesses.
Part of the Brigade fleet · brigade.tools · escoffierlabs.dev
[ 01 · THE AUDIT TRIO ]
One report contract, three lenses
Run any of them alone or all three over time. The findings compose into a single leverage-sorted backlog. The trio is read-only by design, and expedite is the separate, opt-in step that drives the backlog to done.
/line-check
The flagship. A chef’s line check for your repo: seven stations scored (docs, agent-readiness, tests/CI, hygiene, structure, release hygiene, TODO mining) and a backlog sorted by impact relative to effort.
/bug-hunt
Correctness sweep across five lenses with mandatory adversarial verification. Only bugs that survive a refutation attempt make the report, each with a concrete trigger.
/security-sweep
Defensive security audit: secrets in tree and history, dependency CVEs, injection surfaces, authn/authz, exposure. Every finding ships with its remediation.
Report contract: docs/audit-report-format.md on GitHub.
[ 02 · THE LINE ]
The daily build chain
Design before code, plan before execution, execution with the checkboxes ticked live. Each handoff is written so a fresh session or a zero-context engineer can pick it up.
/taste
Test-first discipline that survives pressure. The failing test is watched failing before any implementation. Nothing leaves the kitchen untasted.
/check
The taste before you call it done. Verify the claim against reality before saying it works, passes, or shipped, especially when relaying a subagent or tool success report.
/refire
Root-cause-first debugging. Find out why the plate came back before cooking it again, three failed fixes means question the architecture.
/stations
The expeditor’s fan-out for parallel agents: cluster by root cause, check write sets, give each station a complete ticket, taste the integrated result.
[ 03 · OUT THE DOOR ]
Nothing leaves the kitchen unchecked
Gates for pull requests, public prose, repo publication, releases, and the knowledge a session leaves behind.
/pass
The gate before a pull request leaves your hands: real fix, tested, one concern, self-reviewed diff, and a PR body you approved.
/publish-readiness
The gate before a repo goes public: working-tree and git-history leak scans, hygiene checks, and the history-rewrite recipe when something already leaked.
/release-cut
Changelog roll-up, semver bump, tag, GitHub release, drafted announcement. Releases on request, never per feature.
/memory-handoff
Ends a session by writing durable knowledge into a structured handoff a memory owner can review and file. Pairs with Brigade, works standalone.
[ 04 · TRY IT IN 60 SECONDS ]
Any npx skills-compatible harness can install the whole roster without a plugin manager.
$ npx skills add escoffier-labs/skillet --list # see every skill without installing
$ npx skills add escoffier-labs/skillet # install the roster
$ /line-check # seven-station repo audit, leverage-sorted backlog
$ /security-sweep # defensive security audit, each finding with its fix
$ /publish-readiness # leak scan before a repo goes public[ 05 · WHY NOT SOMETHING ELSE ]
A single mega-prompt or one big CLAUDE.md works until it bloats past the context budget and goes stale, and it loads every instruction on every turn whether the work matches or not. Skillet is one self-contained file per job that triggers only when the work matches, so the audit procedure does not cost context while you are cutting a release.
Hand-rolling the procedure each session means the gotcha you learned last month, the repo that published with hostnames in its history, the audit that produced a wall of unprioritized findings, is gone by the next session. Each skill encodes the procedure plus the hard-won gotchas so the floor stays raised.
A tool-locked agent framework ties the workflow to one harness. Skillet skills are plain SKILL.md files: they install into Claude Code, Codex, or any SKILL.md-compatible harness, per-machine or per-repo, the same way.
Native harness skill libraries are great, and skillet is not a replacement for the harness. It is the roster of process skills you drop on top, with a shared report contract so the read-only auditors and the execution skills compose instead of each inventing its own format.
[ 06 · WHAT SKILLET IS NOT ]
It does not run anything on its own, ship a binary, or apply changes automatically. You stay in the loop for anything destructive, breaking, or public.
Not a runtime
Does not run anything on its own or install a daemon, scheduler, or background process. It carries no runtime dependency and does not call out to the network on its own.
Not a CLI
Ships no binary. The skills are markdown the harness reads, and they do not replace your harness, your model, or your editor.
Not an auto-fixer
The read-only audit skills (line-check, bug-hunt, security-sweep, special) never apply changes themselves. Applying findings is the separate, opt-in expedite step, and release-cut runs on request, never per feature.
[ 07 · INSTALL ]
Install Skillet, then wire it into Brigade when you want the station on your line:
Attach via Brigade:
brigade add skills
Brigade itself:
brigade.tools
·
pipx install brigade-cli
More from Escoffier Labs
Local-first tools for AI coding agents
One kitchen, one brigade. Each tool does one job well, and they share one reviewed source of memory, config, and receipts.