transitions-agent
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
npxto execute thetransitions-agentCLI tool for several automated workflows including scanning the codebase, applying code fixes, and initializing CI configuration. - [EXTERNAL_DOWNLOADS]: The skill fetches the
transitions-agentpackage from the npm registry at runtime usingnpx transitions-agent@latest. The tool also communicates withapi.transitions.devto fetch motion recipes and perform fixed-based transformations. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the codebase by processing the JSON output of the scan command.
- Ingestion points: SKILL.md (via
npx transitions-agent --jsonoutput processing). - Boundary markers: Absent; the agent is instructed to relay results directly to the user.
- Capability inventory: The skill has the ability to write to files (
fixcommand), create PRs/branches (fix --pr), and modify CI workflows (init-cicommand) across all its operational scripts. - Sanitization: None mentioned; the agent relies on the CLI's deterministic scan output.
- [DATA_EXFILTRATION]: The skill accesses sensitive configuration paths to manage its licensing, specifically reading from
~/.transitions-agent.jsonand theTRANSITIONS_AGENT_LICENSEenvironment variable. It may also transmit this key to theapi.transitions.devservice via the--licenseflag during fix operations.
Audit Metadata