workflow-spec-sync
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill includes instructions to run a local Node.js script (
.claude/scripts/lib/workflow-skill-contract.cjs) for regenerating workflow call blocks. This is a standard project maintenance task and does not involve downloading or executing code from external sources. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests data from external project files and user prompts.
- Ingestion points: Reads project configuration (
docs/project-config.json), workflow definitions (.claude/workflows.json), and project principles (spec-principles.md), alongside the user prompt. - Boundary markers: The protocol emphasizes resolving a "native artifact contract" and mandates that unresolved components remain
UNKNOWN/BLOCKEDrather than being guessed, which acts as a logical boundary for instruction interpretation. - Capability inventory: The skill can spawn sub-agents, execute local Node.js scripts, write reports to the
tmp/reports/directory, and invoke a suite of other functional skills ($spec,$test,$investigate, etc.). - Sanitization: The skill implements multi-stage quality gates, including a
review-convergedstate and mandatory fault adjudication for test failures, which serves to validate and sanitize the impact of processed data before task completion.
Audit Metadata