The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes content from arbitrary public GitHub repositories.
Ingestion points: Untrusted data enters the agent's context through the read-wiki-contents and ask-question tools implemented in scripts/deepwiki.ts when they retrieve documentation derived from external repositories.
Boundary markers: The script lacks explicit boundary markers or instructions to the agent to ignore potentially malicious instructions embedded within the retrieved repository documentation.
Capability inventory: The skill is limited to making network requests to the vendor's API (mcp.deepwiki.com). No capabilities for local file system modification, sensitive data access, or arbitrary command execution were found in the analyzed scripts.
Sanitization: The client-side proxy script does not perform sanitization or filtering of the content returned by the DeepWiki service before presenting it to the agent.

deep-wiki