The Agent Skills Directory

[SAFE]: The skill utilizes vendor-specific tools for metadata storage and search. The mcp__plugin_ruflo-core_ruflo__agentdb_* and mcp__plugin_ruflo-core_ruflo__memory_* tools are used to maintain a hierarchical graph of decisions.
[SAFE]: All file system operations are scoped to the creation and numbering of documentation files within the docs/adr/ directory.
[INDIRECT_PROMPT_INJECTION]: The skill incorporates user-supplied input (title and context) into the generated ADR files and database entries.
Ingestion points: The <title> argument and the user-provided context for the ADR are interpolated into file templates and JSON values.
Boundary markers: The input is encapsulated within Markdown structures and JSON-encoded strings.
Capability inventory: The skill uses Write to create files and Bash for command-line operations.
Sanitization: The instructions do not specify explicit sanitization logic for the title slugification beyond lowercase conversion and hyphenation, though the risk is negligible as these are intended for documentation records.

adr-create