The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local script using node plugins/ruflo-adr/scripts/verify.mjs. This is the primary mechanism for performing graph validation logic and is consistent with the skill's stated purpose.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external namespaces (adr-patterns and adr-edges) retrieved via memory tools.
Ingestion points: Data is read from the ruflo__memory namespaces using mcp__plugin_ruflo-core_ruflo__memory_retrieve.
Boundary markers: The skill does not explicitly define prompt boundaries in the markdown, as data processing is offloaded to the Node.js script.
Capability inventory: The skill has access to Bash and Node.js for execution.
Sanitization: Validation and integrity checks are performed by the verify.mjs script to identify mismatches and cycles, acting as a structural filter.

adr-verify