here-now
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes provided bash scripts (
publish.shanddrive.sh) to perform file management tasks such asfind,wc, andsha256sum, and to handle network operations usingcurlfor site deployment and storage management. - [CREDENTIALS_UNSAFE]: Instructions direct the agent to store API keys in a local file (
~/.herenow/credentials) and set restrictive permissions (chmod 600). This is a standard practice for CLI-based authentication management, which the agent is tasked with implementing on behalf of the user. - [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it processes and publishes user-provided files and directories.
- Ingestion points: Local file paths and Drive storage snapshots passed to the
publish.shanddrive.shscripts. - Boundary markers: No specific delimiters are defined in the instructions to isolate user-provided data from agent processing logic.
- Capability inventory: The skill can perform network requests and local file system reads/writes via its helper scripts.
- Sanitization: While the scripts perform URL encoding for paths, they do not validate or sanitize the content of the files being uploaded.
- [EXTERNAL_DOWNLOADS]: The skill interacts with the following external resources for its core functionality:
- The vendor API at
https://here.nowandhttps://api.here.now. - Direct file uploads to Cloudflare R2 storage at
https://*.r2.cloudflarestorage.com. - Installation and update instructions reference
npx skills add heredotnow/skill. These interactions are consistent with the skill's purpose and target established vendor or well-known service endpoints.
Audit Metadata