metravel-docs-maintainer

name: metravel-docs-maintainer description: Maintain metravel project documentation and Codex operating rules. Use when Codex needs to update docs/, AGENTS.md, .codex/skills, project instructions, workflow rules, prompts, skill metadata, or documentation structure in this repository.

Metravel Docs Maintainer

Read AGENTS.md, docs/RULES.md, docs/README.md, and docs/CODEX.md before changing documentation or Codex skills.

Use the AI task triage and self-check in docs/CODEX.md to keep docs changes scoped before editing.

Keep documentation compact and authoritative:

• Prefer updating existing canonical docs over creating new files.
• Use docs/RULES.md for mandatory project rules and policies.
• Use docs/README.md for quick navigation, setup, and API reference.
• Use docs/DEVELOPMENT.md, docs/TESTING.md, and docs/RELEASE.md for workflow-specific details.
• Use docs/CODEX.md for Codex workflow, project skill selection, and skill maintenance rules.
• Update docs/INDEX.md whenever adding, removing, or renaming a docs file.

Maintain skills in .codex/skills/<skill-name>/:

• Keep each SKILL.md short, procedural, and specific to one job.
• Put triggering guidance in the frontmatter description, because Codex sees that before loading the body.
• Do not add README, CHANGELOG, or other auxiliary files inside a skill folder.
• Add references/ only for detailed material that should be loaded on demand.
• Keep agents/openai.yaml aligned with the skill name, purpose, and default prompt.

Avoid documentation drift:

• Do not duplicate the same rule in many places unless one location is a short pointer to the canonical source.
• When changing rules, update both the canonical source and the short Codex map if the skill selection or agent workflow changes.
• When changing Codex behavior rules, update the matching skill prompts/metadata so agents see the rule before starting work.
• Preserve existing project constraints for external links, UI guardrails, server path safety, release checks, and e2e secrets.
• Keep Codex/debug artifact rules aligned across AGENTS.md, docs/RULES.md, docs/CODEX.md, and relevant skills: temporary debugging output belongs only in ignored local folders such as .codex-temp/ or .codex-debug/, and stale artifacts should be removed before handoff.
• Do not print secrets from .env, .env.e2e, or deployment configs.
• Read Markdown as UTF-8; if PowerShell displays Cyrillic as mojibake, reread with Get-Content -Encoding UTF8 before assuming the file is corrupt.

Validation for docs-only changes:

• Check changed Markdown/YAML files for readable structure and valid frontmatter.
• Run the skill validator for changed skills when the skill-creator scripts are available.
• Run broader project checks only when documentation changes also alter code, commands, governance behavior, or release policy.