The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git commands (ls-files, rm, add, commit) using the Bash tool for repository state management and cleanup of tracked violations. It also uses chmod to set execution permissions on the generated pre-commit hook.
[REMOTE_CODE_EXECUTION]: The skill generates a bash script from a template and writes it to the repository's .git/hooks/pre-commit directory. This script is executed automatically by the local git environment during the commit process to enforce file exclusion rules.
[PROMPT_INJECTION]: The skill analyzes project configuration files to detect the tech stack and apply appropriate ignore patterns.
Ingestion points: .fullstack-dev/config.json, package.json, tsconfig.json, .mcp.json, and other stack-specific marker files.
Boundary markers: None identified in the prompt interpolation logic.
Capability inventory: File system writing (Write tool) and repository modification (git commands via Bash tool).
Sanitization: Tech-stack detection relies on file existence and whitelisted keyword matching against a predefined internal catalog rather than direct interpolation of untrusted file contents into prompts.

gitignore