hindsight-upgrade
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches version information from the vendor's GitHub repository at https://raw.githubusercontent.com/vectorize-io/hindsight-skills/main/VERSION using curl or wget.
- [EXTERNAL_DOWNLOADS]: The upgrade flow downloads updated skill code from the vendor's repository at https://github.com/vectorize-io/hindsight-skills.git via git clone or git pull.
- [COMMAND_EXECUTION]: The skill automatically executes a setup script (./setup) located within the updated repository after a successful download or update.
- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests and displays external data including version numbers and CHANGELOG.md content.
- Ingestion points: Remote version strings fetched via network; text content from the CHANGELOG.md file.
- Boundary markers: The content is presented as informative text to the user, but lacks explicit delimiters to prevent the agent from interpreting embedded text as instructions.
- Capability inventory: The skill is capable of performing network requests, file system modifications (rm, mv), and arbitrary shell command execution via the setup script.
- Sanitization: The update script validates the remote version string against a semantic versioning (semver) regular expression before processing.
Audit Metadata