report-analyzer-issue

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh issue create) to automate the filing of issue reports. This command execution is a functional requirement for the skill's intended purpose of reporting confirmed engine issues.\n- [DATA_EXFILTRATION]: Information extracted from local diagnoses and test projects is transmitted to the seqra/opentaint repository. As this repository belongs to the tool's vendor and the data transfer is part of the documented reporting workflow, it is considered legitimate functional behavior rather than malicious exfiltration.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and interpolates untrusted data from engine diagnoses and test projects into its report output without explicit sanitization.\n
  • Ingestion points: Untrusted data enters the agent context via the <diagnosis>, <test-project>, and <artifact> inputs.\n
  • Boundary markers: The instructions do not specify any delimiters or warnings to prevent the agent from accidentally executing instructions contained within the ingested data.\n
  • Capability inventory: The agent possesses the capability to write to the local filesystem (.opentaint/issues/) and execute network-enabled commands (gh issue create).\n
  • Sanitization: There is no requirement for input validation, escaping, or filtering of the external content before it is processed and uploaded to the remote repository.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 05:57 AM
Security Audit — agent-trust-hub — report-analyzer-issue