The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to scan sensitive local files including .env, application.yml, and the ~/.config/ directory to extract service endpoints and configuration data (SKILL.md Phase 0.3).
[EXTERNAL_DOWNLOADS]: The skill automatically retrieves external source code by executing 'gh repo clone' for any repositories discovered during the integration tracing process (SKILL.md Phase 0.3).
[COMMAND_EXECUTION]: The agent executes shell commands and GitHub CLI operations ('gh api', 'gh repo clone') to search for and download third-party codebases (SKILL.md Phase 0.3).
[PROMPT_INJECTION]: The skill possesses a significant surface for indirect prompt injection as it ingests and analyzes unvalidated content from external repositories to generate implementation stories and epics. Ingestion points: Discovered codebases and documentation (SKILL.md Phase 0.3). Boundary markers: Not implemented in the provided prompt instructions. Capability inventory: Filesystem access, network operations via GitHub CLI, and subprocess execution. Sanitization: No sanitization or validation of external content is specified before processing (SKILL.md Phase 6).

integration-analysis