catch-and-linger

Fail

Audited by Gen Agent Trust Hub on Nov 16, 2026

Risk Level: HIGHDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill accesses sensitive local files including Chrome and Edge bookmark databases from the user's AppData directory via scripts/read-bookmarks.mjs. This information, along with API keys retrieved from ~/.config/catchlinger/env by scripts/catchlinger-env.mjs, is transmitted to an external service endpoint using scripts/catchlinger-api.mjs.
  • [INDIRECT_PROMPT_INJECTION]: The skill fetches untrusted content from the web and GitHub for processing by the AI agent, creating an injection surface. 1. Ingestion points: External URLs fetched by scripts/fetch-content.mjs and GitHub metadata from scripts/github-stars.mjs. 2. Boundary markers: None; the skill does not use delimiters or instructions to ignore embedded content. 3. Capability inventory: API write access for creating, modifying, or deleting bookmark entries and tags. 4. Sanitization: None; raw text is passed directly to the agent.
  • [COMMAND_EXECUTION]: The skill operates by executing multiple local Node.js scripts that perform file system reads and network operations.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Nov 16, 2026, 06:50 AM