developer-agent

Pass

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): Indirect Prompt Injection Surface.
  • Ingestion points: The skill ingests untrusted user-provided links and attachments in SKILL.md (Stage 5) and references/workflow-details.md.
  • Boundary markers: Minimal delimiters are used (e.g., "Attached Resources" header in references/workflow-details.md), but there are no explicit instructions to the receiving agent to treat the content as untrusted data or ignore embedded instructions.
  • Capability inventory: The skill has high-impact capabilities including local shell command execution (pnpm build) and repository modification via git operations (SKILL.md, Stages 8 and 9).
  • Sanitization: No sanitization is performed. In fact, references/cursor-guidelines.md explicitly mandates including "ALL user-provided links" and "ALL user-provided attachments" when prompting the sub-agent.
  • [COMMAND_EXECUTION] (SAFE): The skill performs shell command execution to manage the development lifecycle.
  • Evidence: pnpm build, git checkout, git commit, and git push commands are found in SKILL.md (Stages 2, 8, and 9).
  • Context: These operations are necessary for the skill's primary purpose as a developer orchestrator and do not involve immediate execution of remote or untrusted scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 23, 2026, 07:01 AM