The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to run npx -y skills-ref validate, which downloads and executes a package from the npm registry at runtime without version pinning or integrity checks.
[EXTERNAL_DOWNLOADS]: The validation workflow fetches the skills-ref package from the public npm registry during execution.
[COMMAND_EXECUTION]: The skill invokes the npx utility to execute external validation logic on the filesystem.
[PROMPT_INJECTION]: The skill ingests content from existing SKILL.md files to drive its improvement logic, which creates a surface for indirect prompt injection.
Ingestion points: Reads SKILL.md files from ~/.agents/skills/, ~/.pi/agent/skills/, and project-local directories.
Boundary markers: The instructions lack delimiters or specific directives to ignore or sanitize instructions embedded within the files being processed.
Capability inventory: The skill has the ability to read from and write to the local filesystem and execute shell commands via npx.
Sanitization: There is no evidence of sanitization or validation of the content read from external skill files before it is processed by the agent.

improve-skills