The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/repo_snapshot.py executes the local git command using subprocess.run to retrieve the current HEAD and branch name.
[EXTERNAL_DOWNLOADS]: The documentation suggests using openskills, an external Node.js tool from the same author, to facilitate installation and discovery of the skill.
[COMMAND_EXECUTION]: The script scripts/render_md_to_html.py performs file system operations to write the generated HTML report to the user's local disk.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and analyze untrusted data from the repository being reviewed.
Ingestion points: The agent reads source code, documentation, and configuration files from the target repository as part of its analysis phases.
Boundary markers: The SKILL.md file includes a specific 'Security and Desensitization' section that instructs the agent to ignore and mask sensitive values like keys and tokens.
Capability inventory: The skill includes Python scripts that use subprocess.run for git commands and file writing operations.
Sanitization: The scripts/render_md_to_html.py script utilizes html.escape to sanitize content before rendering it into the HTML report.

repo-deep-dive-report