The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a Python script to perform repository audits which involves executing local Git commands.- Evidence: In scripts/audit-higgsfield-game.py, the git_commit function utilizes subprocess.run with a list of arguments ["git", "-C", str(repo), "rev-parse", "HEAD"] to resolve the current repository state.- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external repository files, creating a surface where malicious content in those files could influence agent behavior.- Ingestion points: The audit-higgsfield-game.py script reads the content of SKILL.md, INSTALL.md, and references/game-flow.md from the target repository directory provided by the user.- Boundary markers: The auditor script outputs structured JSON or text results for the agent to interpret, providing some separation, though the instructions do not mandate specific filtering of the ingested content.- Capability inventory: The skill is configured with Bash, Read, Glob, Grep, and WebFetch tools, and it executes shell commands through the Python helper script.- Sanitization: The script reads and processes file content directly without specific NLP sanitization or instruction-stripping logic.- [EXTERNAL_DOWNLOADS]: The skill documentation includes examples for installing components from a remote source.- Evidence: The SKILL.toon file contains an installation example using npx skills add targeting a GitHub repository hosted by the skill author.

higgsfield-game-generation