The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads and installs new agent capabilities from the vendor's repository using the npx @ericrisco/rsc add <id> command. This is a core management feature and requires user confirmation for each installation.
[COMMAND_EXECUTION]: The skill invokes various command-line tools to maintain project health, including npx @ericrisco/rsc doctor for diagnostics, npx @ericrisco/rsc repair for fixing harness configuration issues, and npx @ericrisco/rsc sync to align the environment with project metadata.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes all user inputs to detect missing capabilities and classify intent, creating a surface where malicious prompts could influence suggestions.
Ingestion points: The skill logic processes every user turn at the beginning of the session and after every compaction event (SKILL.md).
Boundary markers: The skill does not implement specific delimiters or explicit warnings to the agent to ignore instructions embedded within the user data it processes.
Capability inventory: The skill has the ability to execute management-level CLI tools, perform package installations via npx, and modify project configuration files like .rsc.json.
Sanitization: The skill does not perform explicit filtering or validation of user-provided task descriptions before using them to query the skill catalog for matches.

suggest