autopilot
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
Bashtool to perform complex repository management tasks includinggit switch,git branch, andgit log. It also uses shell redirection to write JSON files to the local file system usingcatand heredocs, and executes local scripts via thenoderuntime.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and parses configuration fromcontract.mdandcontract-data.jsonfiles within the user's project directory. This data is used to populate the manifest for the workflow engine and directly influences the tool execution flow.\n - Ingestion points: Project configuration files (
contract.md,contract-data.json) and various specification files identified viaspecPath.\n - Boundary markers: The skill relies on specific Markdown headers and JSON structure for parsing but lacks explicit safety checks for the content within those structures.\n
- Capability inventory: The skill has access to shell command execution (
Bash), background workflows (Workflow), and multi-stage agent spawning (Agent).\n - Sanitization: No explicit sanitization or input validation is performed on the data parsed from project files before it is used in subsequent tool calls or manifest generation.\n- [DYNAMIC_EXECUTION]: The skill dynamically constructs and executes logic by passing arguments derived from project files to local JavaScript and TypeScript utilities (
run-report-gen.ts,verify.mjs) and a specialized workflow engine (execute-contract.mjs).
Audit Metadata