The Agent Skills Directory

[SAFE]: The skill implements "Hard Gates" that strictly limit its operations to generating requirement documents and prevent it from prematurely advancing workflow stages or creating unauthorized files.
[SAFE]: The skill includes a confirmation mechanism that requires user approval before overwriting existing files, mitigating the risk of accidental data loss.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface because it processes content from external web links and local files.
Ingestion points: Content fetched via web-access and lark-doc tools, as well as local file paths provided by the user in SKILL.md (Step 1 and 3).
Boundary markers: The instructions do not specify the use of delimiters or "ignore embedded instructions" warnings to isolate ingested data from the agent's core instructions.
Capability inventory: The skill has file system write access (restricted to the .dwf/specs/ directory) and can update the project's .dwf/state.json file.
Sanitization: No explicit sanitization, escaping, or validation of the ingested content is performed before it is incorporated into the generated Markdown requirement document.

dwf-requirement