oma-debug

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill has a defined surface for ingesting untrusted data through error messages, logs, and stack traces provided by the user. This is a standard requirement for debugging tasks. Ingestion points: SKILL.md (Expected inputs), resources/execution-protocol.md (Gather symptoms). Boundary markers: No explicit delimiters are specified for log content. Capability inventory: The skill can write source code and execute shell commands for reproduction and testing (CALL_TOOL, WRITE). Sanitization: No explicit sanitization of log content is mentioned.
  • [COMMAND_EXECUTION]: The skill is designed to execute local development commands, including test suites, linters, and type-checkers, as well as reproduction scripts. This is consistent with the skill's primary purpose.
  • [DYNAMIC_EXECUTION]: The skill generates and runs new regression tests and applies code patches at runtime. These actions are fundamental to the agent's role in bug fixing and are constrained by a self-verification checklist to ensure no new issues are introduced.
  • [SAFE]: The skill includes comprehensive educational content in resources/common-patterns.md that helps the agent identify and avoid security risks like SQL injection and XSS during code modification. References to external tools like Serena MCP and Antigravity Browser appear to be internal or platform-integrated utilities.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 04:52 PM