here-now

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes local bash scripts (publish.sh, drive.sh) and standard system utilities (curl, jq, file) to interact with the service API and manage local files. Input sanitization for URL paths is performed using jq to prevent command or path injection during network operations.
  • [DATA_EXFILTRATION]: The skill is designed to upload local files and directories to the here.now servers for web hosting and cloud storage. While this involves sending user data to a remote service, it is the primary function of the tool. The scripts implement a security check (ALLOW_NON_HERENOW_BASE_URL) to ensure API keys are not accidentally sent to unauthorized domains.
  • [EXTERNAL_DOWNLOADS]: The agent is instructed to fetch and read documentation from an external URL (https://here.now/docs) and perform API calls to the vendor's service. All network interactions are directed towards the service's official domain.
  • [PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection because it mandates that the agent fetch and follow instructions from an external website. A compromise of the documentation site could allow an attacker to influence the agent's behavior.
  • Ingestion points: Dynamic content is fetched from https://here.now/docs at runtime during the first interaction and when answering user questions.
  • Boundary markers: No specific delimiters or safety instructions are provided to separate the external documentation from the agent's core instructions.
  • Capability inventory: The agent has access to tools for network requests with authentication, file system reading and writing, and shell script execution.
  • Sanitization: No sanitization or verification of the external content is performed before the agent processes it.
  • [CREDENTIALS_UNSAFE]: The skill instructs the agent to store API keys in ~/.herenow/credentials and set file permissions to 600. While this creates a persistent sensitive file on the system, it follows standard practices for local secret management and avoids more dangerous patterns like hardcoding secrets or passing them through world-readable command history.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 03:31 AM
Security Audit — agent-trust-hub — here-now