The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the handoff of user instructions to the build-agent tool by requiring a faithful and verbatim transfer of user wording. This design creates a path for indirect prompt injection where adversarial instructions in the input could influence the downstream builder tools.
Ingestion points: The user's request is captured and passed as a message parameter to the build-agent tool within SKILL.md instructions.
Capability inventory: The skill utilizes tools like build-agent, build-workflow, and data-tables which have the power to create and configure executable artifacts and database structures.
Boundary markers: The instructions specify that only explicit user requirements should be included in the tool call, but do not provide specific delimiters or ignore-instructions for the LLM to use when interpolating the data.
Sanitization: There are no instructions provided for sanitizing or validating the content of the user's request before forwarding it to the builder.

agent-builder