ln-513-regression-checker

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill builds and executes shell commands for various test frameworks (pytest, jest, vitest, go test). It specifically prioritizes exact commands found in 'docs/project/runbook.md', allowing for arbitrary command execution based on file content.
  • [DATA_EXFILTRATION]: The worker reads project infrastructure documentation ('docs/project/infrastructure.md') and runbooks to extract service endpoints, port allocations, and environment variables. These details are used to configure the test environment but involve processing potentially sensitive technical data.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through untrusted project files.
  • Ingestion points: Reads instructions and commands from 'docs/project/runbook.md', 'docs/project/infrastructure.md', and framework-specific configuration files (e.g., pytest.ini, package.json).
  • Boundary markers: None implemented; commands from the runbook are given priority over automated safety checks or detection logic.
  • Capability inventory: Capability to execute arbitrary shell commands via the detected test framework CLI and post results to the Linear API.
  • Sanitization: No evidence of input validation or escaping for commands extracted from project documentation before execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 01:43 PM