auto-update
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONPERSISTENCE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses Git commands such as
git fetch,git status,git log, andgit pullto manage repository state. - [REMOTE_CODE_EXECUTION]: The skill is designed to pull and apply updates to the agent's own code and instructions from a remote repository. This constitutes a mechanism for remote logic modification.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests instructions from an external source (the Git remote) which are then reloaded into the agent's context. This creates an attack surface for indirect prompt injection. 1. Ingestion points: Remote content is pulled via
git pull origin maininto the local workspace. 2. Boundary markers: None are present; the skill reloads the environment immediately after pulling. 3. Capability inventory: The skill can execute shell commands and reload agent skills. 4. Sanitization: No sanitization or validation of the fetched content is performed before activation. - [PERSISTENCE]: The skill utilizes a
scheduletool to create recurring tasks (e.g., cron jobs), allowing the update mechanism to run automatically and persist across agent sessions.
Audit Metadata