resolving-merge-conflicts

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to ingest and analyze untrusted external data to understand the intent behind code changes.
  • Ingestion points: SKILL.md (Step 2) instructs the agent to read git history, conflicting files, commit messages, pull request descriptions, and issue tickets.
  • Boundary markers: No specific delimiters or instructions are provided to help the agent distinguish between legitimate documentation and potentially malicious instructions embedded in git metadata.
  • Capability inventory: The agent has the capability to execute shell commands for git operations and run automated project checks (SKILL.md, Step 4).
  • Sanitization: There is no mention of sanitizing or filtering the content read from git logs or PR descriptions.
  • Analysis: An attacker could craft a malicious commit message or PR description that includes hidden instructions designed to trick the agent into performing unauthorized actions during the resolution process.
  • [COMMAND_EXECUTION]: The skill requires the agent to run project-specific automated checks.
  • Evidence: SKILL.md (Step 4) instructs the agent to "Discover the project's automated checks and run them, typically typecheck, then tests, then format."
  • Analysis: This capability is standard for a developer agent, but it presents a risk if the agent is operating on an untrusted repository. A malicious actor could configure the project's test suite to execute harmful commands when the agent attempts to run the expected "automated checks."
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 09:23 PM
Security Audit — agent-trust-hub — resolving-merge-conflicts