skills/steipete/openclaw/oracle/Gen Agent Trust Hub

oracle

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches the @steipete/oracle package from the NPM registry as specified in the installation metadata.
  • [COMMAND_EXECUTION]: Executes the oracle CLI tool to interact with the local filesystem and manage analysis sessions. The tool also includes a server mode (oracle serve) for remote model interaction.
  • [DATA_EXFILTRATION]: Transmits repository context to external model providers (API or browser-based) for analysis. The documentation includes explicit warnings to redact secrets and avoid including .env or key files in the bundled payload.
  • [PROMPT_INJECTION]: Acts as a surface for indirect prompt injection by ingesting repository data for processing by secondary models. Evidence chain: 1. Ingestion points: Local repository files selected via --file globs. 2. Boundary markers: None mentioned for isolating ingested context. 3. Capability inventory: File system read, network communication, and local server hosting. 4. Sanitization: No evidence of content filtering for the bundled files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 02:11 PM
Security Audit — agent-trust-hub — oracle