opencode-dev

Pass

Audited by Gen Agent Trust Hub on Oct 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches and executes @memlab/cli from npm and opencode-ai via bunx. These are recognized tools from trusted or project-related sources.\n- [COMMAND_EXECUTION]: Uses system utilities such as kill, pgrep, ps, and sqlite3 to interact with local processes and state files, which is required for debugging the client/server architecture of the application.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local logs and server API endpoints (e.g., packages/cli/src/commands/handlers/serve.ts).\n
  • Ingestion points: Reads from ~/.local/share/opencode/log/ and server responses via bun dev api.\n
  • Boundary markers: None identified in the instructional text.\n
  • Capability inventory: Includes shell command execution, file system access, and network requests to localhost.\n
  • Sanitization: Not explicitly defined, as the output is intended for manual developer inspection.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 11, 2026, 09:04 AM