The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes data from external files to determine agent state and behavior.\n
Ingestion points: The agent reads configuration from {project-root}/_bmad/skf/config.yaml, preferences from {sidecar_path}/preferences.yaml, and knowledge content from fragment files referenced in skf-knowledge-index.csv.\n
Boundary markers: There are no explicit delimiters or instructions provided to the LLM to ignore or separate instructions potentially embedded in these external files.\n
Capability inventory: The agent can invoke various other skills (e.g., skf-setup, skf-create-skill, skf-analyze-source) that perform filesystem and repository operations.\n
Sanitization: No sanitization or validation of the content read from external files is described before it is interpolated into the agent context.\n- [COMMAND_EXECUTION]: The skill implements a 'Pipeline Mode' that dynamically parses and executes a sequence of other tools based on user input.\n
Evidence: The skill processes user strings containing multiple workflow codes (e.g., BS CS TS EX) or aliases (e.g., forge, onboard) to dynamically sequence the invocation of registered agent capabilities.

skf-forger