writing-workspace
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by ingesting external content for processing.
- Ingestion points: The skill processes user-provided prose, author voice samples, and external source links as described in
references/current-guidance.md. - Boundary markers: The instructions require the agent to distinguish between 'writing workspaces' and 'publishing targets' and to check public boundaries before sharing references or source snippets.
- Capability inventory: The skill involves reading and writing files in local workspaces (e.g.,
drafts/,articles/,voice/) and potentially interacting with publishing repositories. - Sanitization: Verification steps are mandated to confirm that factual claims are preserved and frontmatter/metadata is validated before publication.
Audit Metadata