deep-research
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from a wide variety of untrusted external sources, creating an attack surface for instructions hidden in those sources to influence the agent's synthesis or actions.
- Ingestion points: Data is ingested from
WebSearch,WebFetch,agent-browser, and thelast30daysskill (fetching from Reddit, X, YouTube, and Hacker News). - Boundary markers: The instructions lack specific boundary markers or 'ignore' directives to separate untrusted external content from the agent's core instructions.
- Capability inventory: The agent possesses capabilities to write to the local file system (archiving research to
~/.config/makerskills/deep-research/archive/), call the Notion API, and generate synthesis for the user. - Sanitization: No sanitization or filtering logic is specified for the data retrieved from external web sources.
- [COMMAND_EXECUTION]: The skill uses
grepto perform local searches on the file system. Specifically, it targets~/.claude/memory/and the${MAKERSKILLS_CONFIG:-$HOME/.config/makerskills}/deep-research/archive/directory to retrieve user context and past research runs.
Audit Metadata