agentleFS
fmindfmind/dotfiles/skills/mise/SKILL.md

mise

Skill13 starsChanged 4 days agoMITView on GitHub

Manage mise tool pins, lockfiles, environments, and tasks for hooks and CI.

  • Reads credentials

9 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every project exposes the same core tasks with short aliases so agents, hooks, and CI stay portable:

    line 27

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Mise

One project mise.toml owns tool pins and commands; hooks and CI decide when to invoke them. Reuse the appropriate language stack's configuration instead of creating parallel command definitions.

Workflow

  1. Select versions per the tool version rules before installing copied scaffold selectors; never copy tool binaries into projects.
  2. Keep tasks simple: prefer direct commands, short sequential run arrays, and declarative dependencies; small setup/cleanup sequences are acceptable. Prefer native flags to conditions and explicit tasks to mode detection. Do not wrap commands in bash -c or sh -c, nest conditions, compress a program onto one line, or move a large shell block into TOML; keep unavoidable branching, retries, and response parsing in a small maintained script, preferably Python.
  3. Keep the shared vocabulary below; task conventions own subtask names, concise output, argument forwarding, dependency order, and tool updates.
  4. Pin and install the project toolchain, then validate task definitions with mise tasks validate; use mise run <task> in automation.
  5. Verify the changed task and its callers; record a concrete reason for held-back pins.

Task Vocabulary

Every project exposes the same core tasks with short aliases so agents, hooks, and CI stay portable:

TaskAliasPurpose
installiSync dependencies and install git hooks (lefthook install).
formatfFormat all sources (fans out to format:*).
checkcAll static checks in parallel (fans out to check:*).
testtRun the test suite.
buildbCompile or package artifacts (fans out to build:*).
watchwLive reload, or re-run tests; omitted by stacks with neither.
allacheck, test, build in sequence without rewriting source files: the full gate.

Python projects start from python-stack; specialized content and infrastructure projects use the task files owned by their stack skills, such as infra-as-code.

Gotchas

  • Load .env only when needed: [env] with _.file = ".env" loads dotenv values for tasks; use it only when the project needs that file. _.source expects a shell script.
  • Keep local builds free of side effects: builds and checks must not publish, deploy, or spend by default; expose consequential operations only as explicit on-demand paths.
  • Isolate the full gate on dirty trees: mise run all runs whole-tree checks, tests, and builds without rewriting source files, while mise run format writes the tree. When unrelated changes are present, use git-worktree to materialize the current candidate in isolation and check that the tested files match before transferring proof.
  • Declare project tools in the project: the workstation's global config serves agent coding (connectors, cross-project CLIs such as ruff, ty, lefthook, and dprint, and a few review and editor tools), and a global copy never replaces a project's pin. Declare every tool a project needs, such as kubectl or OpenTofu, in the project's mise.toml with mise use --pin <tool> per the tool version rules. In a client repository without mise, prefer the client's own toolchain and ask before adding a mise.toml.
  • Never bypass the locked global config: the global config runs in locked mode, so mise x <tool> for a tool it does not list fails outside a project with No lockfile URL found ... (--locked mode). Do not retry with MISE_LOCKED=0, which rewrites the deployed global lock; declare the tool in the project instead.

Documentation

  • upgrade-tools for per-repository upgrades
  • mise · Tasks · Settings
  • Releases: mise · changelog
  • Companion skills: lefthook (hooks call these tasks), github-actions (CI installs the toolchain with mise-action and runs mise run all).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.