skills/szoloth/skills/quill/Gen Agent Trust Hub

quill

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill includes instructions to install the 'quill-tui' CLI tool via 'cargo install' from the author's GitHub repository (github.com/szoloth/quill-tui). It also instructs the user to build a local MCP server using 'npm install' in the '~/quill-mcp' directory. These are standard installation procedures for the vendor's ecosystem.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill manages document state through a shared application directory at '~/.quill/'. It uses specific tools to read ('quill_get_document') and write ('quill_update_document') document content and annotations. This access is scoped to the application's data and is necessary for the bi-directional writing assistant workflow.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process document content ('draft.md') to provide feedback. This represents a standard surface for indirect prompt injection common to all document analysis skills. The skill manages this via structured MCP tools ('quill_get_prompt', 'quill_get_annotations') which provide a controlled interface for the agent to interact with the user's text.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 08:25 PM
Security Audit — agent-trust-hub — quill