`reyn lint`¶

Run deterministic structural checks on a skill directory: graph, frontmatter, artifact references, and Python preprocessor steps (when mode: safe, the AST is also validated). Detects most authoring mistakes before runtime.

Synopsis¶

reyn lint SKILL

Positional arguments¶

Name	Description
`SKILL`	Skill name. Same resolution as `reyn run`: `reyn/project/` → `reyn/local/` → stdlib.

What is checked¶

Graph: every key references a phase file in phases/; every value is a known phase, sub-skill (@name), or end.
Reachability: every phase reachable from entry; phases with can_finish: true have a path to end.
Frontmatter: required keys (type, name, entry, final_output).
Artifact references: every input and final_output_schema resolves to an artifact file.
Preprocessor: each python step has a matching permissions.python entry, the .py file exists, and the function is defined. In mode: safe, the AST is checked against the allowlist (no open, eval, exec, __import__, subprocess, etc.).

Exit codes¶

Code	Meaning
`0`	No errors (warnings may still be present)
`1`	One or more errors found

Output¶

Each issue is printed on its own line:

[error]   reyn/local/my_skill/phases/draft.md
          graph references unknown phase 'reveiw' (typo for 'review'?)

[warning] reyn/local/my_skill/phases/draft.md
          phase 'draft' has can_finish: true but no path to 'end'

A summary follows: N error(s), M warning(s).

When clean: No issues found.

Examples¶

Lint a project skill:

reyn lint article_writer

Lint a stdlib skill (sanity check after editing):

reyn lint eval

Use in CI:

reyn lint my_skill || exit 1

reyn lint¶