resolving-merge-conflicts

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to read and resolve conflicts within source code containing Git conflict markers (<<<<<<<, =======, >>>>>>>). This establishes an ingestion point for untrusted external data which could potentially contain embedded instructions intended to influence the agent's resolution logic.
  • Ingestion points: Conflicted files in the worktree during merge, rebase, cherry-pick, or stash-pop operations (SKILL.md).
  • Boundary markers: Absent; there are no specific instructions to ignore malicious text within the conflict hunks, though the skill mandates resolution against primary sources (PRs, tickets).
  • Capability inventory: The agent is authorized to use git, git-commit, file-pr, and internal triggers like /done and preflight-mutations (SKILL.md).
  • Sanitization: Not explicitly mentioned for the content being resolved.
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute standard and project-specific command-line tools to manage repository state and verify resolutions.
  • Evidence: Use of git status, git diff, git-commit, file-pr, and workflow tools like preflight-mutations and /done.
  • Context: These tools are used for their intended development purposes, including status checking, change verification, and finalization of repository operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 09:35 AM
Security Audit — agent-trust-hub — resolving-merge-conflicts