The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes third-party content, specifically issue bodies and comments from GitHub, GitLab, and Linear, to identify unresolved project decisions.
Ingestion points: External issue content retrieved via gh, glab, and Linear MCP tools.
Boundary markers: The skill implements a comprehensive 'Author Authority' logic (detailed in REFERENCE.md) that distinguishes between authorized instructions from maintainers and unauthorized context from external contributors. It explicitly monitors for text addressing the agent directly.
Capability inventory: The skill has access to forge APIs (gh, glab) and local shell execution (Bash).
Sanitization: The workflow enforces human-in-the-loop confirmation before any external writes (e.g., updating issue bodies) and treats unauthorized text as non-instructional context.
[COMMAND_EXECUTION]: The skill uses Bash to execute utility commands such as jq for JSON processing, git for repository analysis, and forge CLIs (gh, glab) for metadata retrieval. These are standard operations for its stated purpose.
[EXTERNAL_DOWNLOADS]: The skill references a configuration schema hosted on GitHub at https://raw.githubusercontent.com/TitusKirch/skills/main/tituskirch-skills.schema.json. This is a vendor-owned resource used for static validation.

refine-issue