The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes local files in the target repository to calculate metrics and generate a project overview. This ingestion of potentially untrusted local data creates a surface for indirect prompt injection.
Ingestion points: scripts/collect-site-metrics.py reads all .md files within the docs/ directory to aggregate counts and content summaries.
Boundary markers: The workflow does not utilize specific delimiters to isolate ingested file content from the agent's instructions.
Capability inventory: The skill is scoped to local file system read operations and writing the README.md output; it does not have network access or system administration privileges.
Sanitization: Ingested text is processed for metrics (word counts, item counts) but is not sanitized for potential instruction strings before being used to generate the project description.
[COMMAND_EXECUTION]: The skill uses the shell tool to execute two local Python scripts, collect-site-metrics.py and validate-readme.py. These scripts are part of the skill's distribution and perform read-only analysis of the repository content.
[DATA_EXPOSURE]: The skill reads the .git/config file to identify the project's repository name and remote URL. This data is used solely to populate the generated README with appropriate links and badges.

readme-generator