The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of shell commands through its configuration rules and user-defined commands.
Evidence: SKILL.md and references/workflow-examples.md detail sh and commands fields for running arbitrary shell scripts like make dev-setup or git operations.
[EXTERNAL_DOWNLOADS]: The configuration examples suggest automated package management which involves downloading external code.
Evidence: references/workflow-examples.md includes npm install as a standard setup command in rule definitions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the repository matching system.
Ingestion points: Repository URLs and directory paths are matched against regex patterns in SKILL.md to determine which commands to execute.
Boundary markers: The system lacks explicit markers to distinguish between trusted and untrusted repository metadata during matching.
Capability inventory: Full shell access is available through usercommands, rules.commands, and recycle hooks defined in the YAML configuration.
Sanitization: The skill implements a shq filter for Go templates to ensure safe shell quoting of variables like .Path or .Name.

config