The Agent Skills Directory

[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection. It reads user-provided requirement and design documents from the file system and uses them to generate follow-up documentation like technical plans and task lists. Malicious instructions embedded in the input documents could influence the generated output or subsequent agent actions.
Ingestion points: Accesses files such as .dwf/specs/{spec.name}/01-需求/需求文档.md and /02-设计稿/设计稿.md.
Boundary markers: The skill does not define specific delimiters or instructions to prevent the agent from obeying commands found within the source files.
Capability inventory: The agent has the ability to read from and write to the local file system (e.g., creating .md and _meta.json files).
Sanitization: There is no evidence of validation or filtering for potentially malicious content in the ingested documents.

dwf-development