deep-read

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface because its core function is to systematically read and reason about untrusted source code from a repository. This could allow malicious instructions embedded in source code files to influence the agent's reasoning process.
  • [PROMPT_INJECTION]: Evidence Chain:
  • Ingestion points: The agent reads codebase files including source code, configuration files (e.g., package.json, requirements.txt), and environment schemas during the scoping and deep-reading phases.
  • Boundary markers: The protocol does not define specific delimiters or defensive prompts to distinguish between data and instructions within the processed files.
  • Capability inventory: The skill is limited to information gathering and reasoning tools (Glob, Grep, Read, AskUserQuestion, and sequential-thinking). It does not have access to shell execution, file-system modification, or network-enabled tools.
  • Sanitization: No sanitization or escaping mechanisms are implemented for handling the content of the files being read.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 08:54 AM