The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to run the npx opensrc command to investigate package internals. This involves executing a CLI tool with arguments that may be derived from the local environment or user queries.
[EXTERNAL_DOWNLOADS]: The use of npx to execute the opensrc package involves fetching the package from the NPM registry at runtime if it is not already cached locally.
[INDIRECT_PROMPT_INJECTION]: The skill is built to ingest and process data from external, untrusted sources including open-source code repositories and web search results.
Ingestion points: External content is pulled into the subagent's context via the output of the opensrc tool and results from web searches as specified in SKILL.md.
Boundary markers: The instructions do not provide explicit delimiters or warnings to prevent the agent from accidentally executing instructions that might be embedded within the retrieved code or documentation.
Capability inventory: The subagent has capabilities to read local project metadata (lockfiles, manifests, workspace configs) and perform network-based research.
Sanitization: No specific sanitization or filtering logic is defined for the content retrieved from external sources before the agent distills and reports it.

librarian