bear-notes

Warn

Audited by Gen Agent Trust Hub on Jul 16, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's metadata includes an installation step for the grizzly CLI tool from a third-party GitHub repository (github.com/tylerwince/grizzly) using go install. This repository is not maintained by a recognized trusted vendor or the skill author.
  • [COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands using the grizzly binary to interact with the local Bear application on macOS.
  • [PROMPT_INJECTION]: The skill processes content from Bear notes, which are external data sources that could contain malicious instructions intended to influence the agent's behavior.
  • Ingestion points: Note content and tags are retrieved into the agent's context through grizzly open-note and open-tag commands.
  • Boundary markers: The instructions do not provide explicit delimiters or warnings to the agent to treat note content as untrusted data.
  • Capability inventory: The agent is granted capabilities to read, create, and modify notes, and it can access the local file system through command arguments (e.g., token-file paths).
  • Sanitization: There is no evidence of filtering or sanitization of note content before it is processed by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 16, 2026, 09:14 PM
Security Audit — agent-trust-hub — bear-notes