oma-search

Warn

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The resources/error-playbook.md file includes recovery steps that instruct the agent to run pip install curl_cffi if the dependency is missing. Runtime installation of packages is a security risk as it can introduce unverified code into the execution environment.
  • [COMMAND_EXECUTION]: The skill relies on shell commands to interact with various tools, including gh (GitHub CLI), glab (GitLab CLI), and a custom oma CLI. The oma search fetch utility implements bypass strategies such as impersonation and browser automation (Puppeteer) to retrieve content from sites that might otherwise block access.
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to fetch and process data from external, untrusted sources like web pages and public code repositories. This behavior creates an attack surface where malicious instructions hidden in the retrieved content could be interpreted by the agent, potentially leading to unauthorized actions.
  • Ingestion points: Content retrieved via the web, docs, and code routes described in resources/execution-protocol.md (e.g., via oma search fetch, gh search code).
  • Boundary markers: The instructions do not specify any sanitization or isolation protocols to prevent the agent from following instructions embedded within the fetched search results.
  • Capability inventory: The skill has the ability to execute shell commands, perform network operations, and read local project files.
  • Sanitization: No explicit sanitization or validation of external content is mentioned before it is processed by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 6, 2026, 04:52 PM