rhdh-spec-driven-schema

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes the openspec CLI to fetch artifact instructions and status (e.g., openspec instructions <id> --json). It also executes a local Python script, scripts/install_project_schema.py, to initialize the project directory by copying configuration files into an openspec/ folder.
  • [INDIRECT_PROMPT_INJECTION]: The skill reads existing project artifacts and CLI output to generate new documentation and implementation tasks. This creates a potential surface where malicious content in a project's documentation could attempt to influence the agent's behavior during subsequent turns.
  • Ingestion points: The agent is instructed to read all files listed in the dependencies field of the CLI response before drafting new artifacts (references/artifact-loop.md).
  • Boundary markers: The skill instructs the agent to treat context and rules blocks as constraints rather than content, though it lacks explicit structural delimiters to separate untrusted document content from instructions.
  • Capability inventory: The skill performs file system writes to the product repository and has the ability to spawn subagents for implementation and verification phases (schema.yaml).
  • Sanitization: The skill does not specify any sanitization, filtering, or escaping for the content ingested from the project's markdown files.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 08:19 PM
Security Audit — agent-trust-hub — rhdh-spec-driven-schema