The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes existing repository files to inform its configuration process, which presents a surface for instructions embedded in data to influence agent behavior.
Ingestion points: Reads files from the repository root including CLAUDE.md, AGENTS.md, CONTEXT.md, and package.json (as seen in SKILL.md).
Boundary markers: No explicit delimiter markers are used when reading these files.
Capability inventory: The skill has the capability to write new markdown files to the docs/agents/ directory and modify existing agent instruction files.
Sanitization: The skill includes a human-in-the-loop mitigation, requiring the agent to present a draft of all changes to the user for confirmation and potential editing before writing to the filesystem.
[COMMAND_EXECUTION]: The skill uses local system commands and provides guidance for utilizing external CLI tools to manage repository resources.
Evidence: Executes git remote -v to identify the hosting provider. It also provides comprehensive templates for using the gh (GitHub) and glab (GitLab) CLI tools for issue and pull request management.
[DATA_EXFILTRATION]: The skill accesses local repository metadata to determine its operating context.
Evidence: Accesses .git/config and git remote -v output to identify repository origin URLs. This information is used locally to suggest the appropriate issue tracker configuration (GitHub vs. GitLab).

setup-matt-pocock-skills