github-deep-research

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Prompt Injection (HIGH): The skill is highly susceptible to Indirect Prompt Injection. It ingests untrusted data from GitHub repositories and external web pages using web_fetch and scripts/github_api.py. Evidence: 1. Ingestion points: scripts/github_api.py (fetching readme, issues, prs) and web_fetch. 2. Boundary markers: Absent (no delimiters or instructions to ignore embedded commands). 3. Capability inventory: Executes python scripts and performs file-write operations. 4. Sanitization: Absent. Malicious instructions in a repository or webpage could override agent logic.
  • Command Execution (MEDIUM): The skill instructs the agent to execute shell commands to run scripts/github_api.py. Since the script content is not included in the skill files, it is an unverifiable dependency. There is a risk of command injection if the script does not properly sanitize the <owner>, <repo>, or <topic> arguments derived from user input.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 01:28 AM