The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local binaries (e.g., gstack-config, gstack-team-init, gstack-slug) located in the ~/.claude/skills/gstack/bin/ directory to manage tool configuration and project initialization.
[REMOTE_CODE_EXECUTION]: The workflow employs eval to execute shell commands generated dynamically by the gstack-slug utility. This dynamic execution pattern is used for environment setup but represents a security risk if the local utility's output is manipulated.
[PROMPT_INJECTION]: The skill performs an automated audit of repository documentation and git logs, creating a surface for indirect prompt injection where malicious instructions in the code or documentation could influence the agent's behavior.
Ingestion points: Repository documentation files (.md), git log output, and git diff output.
Boundary markers: The skill does not use explicit boundary markers to isolate ingested content from its internal instructions.
Capability inventory: File system modification, repository manipulation (git commit, git push), and platform interaction via GitHub (gh) and GitLab (glab) CLI tools.
Sanitization: No sanitization or validation of the audited documentation content is performed before it is analyzed to determine documentation updates.

document-release