The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted source code and repository manifests to generate summaries and documentation. This creates a surface for indirect prompt injection where malicious instructions hidden in the analyzed codebase could influence the agent's output. * Ingestion points: Local source code files, package manifests (e.g., package.json, Cargo.toml), and git diff data. * Boundary markers: The skill does not define explicit delimiters to isolate untrusted code content from agent instructions during processing. * Capability inventory: The agent has permissions to read local repository files and write markdown documentation to the file system. It does not have network access or arbitrary command execution capabilities. * Sanitization: No explicit logic is provided to sanitize or escape strings extracted from source files before inclusion in the generated documentation.
[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands for git operations to support the update and status sub-commands. * Evidence: The update-workflow.md file specifies the use of git diff --name-only and git log to identify changed files and commit hashes.

codedocs