Openwork Debug

What it does

The Openwork Debug skill enables AI agents to identify and suggest fixes for subtle code logic errors using advanced symbolic execution and analysis through the OpenCode engine. It provides access to debugging capabilities within a workspace, allowing inspection of sessions and messages. This skill relies on interacting with an external OpenWork server via HTTP requests or directly accessing a local SQLite database when the API is unavailable.

When to use it

• When you need to identify potential logic errors in code beyond what standard linting can detect.
• To inspect recent audit logs related to debugging sessions within a workspace.
• If the OpenCode engine API is temporarily unavailable, to directly query the underlying SQLite database for session and message data.

Key capabilities

• Accesses OpenWork server endpoints via HTTP requests (health checks, capabilities, workspaces, configuration, audit logs).
• Utilizes the opencode command-line tool for debugging tasks (pinging, listing MCPs, debugging specific entities).
• Provides a fallback mechanism to inspect SQLite database contents directly.

Example prompts

• "Run opencode debug <name>."
• "Show me the latest 5 sessions from the OpenWork audit log."
• "Check the OpenWork server's capabilities."

Tips & gotchas

• Requires environment variables OPENWORK_SERVER_URL and OPENWORK_SERVER_TOKEN to be set before use. OPENWORK_WORKSPACE_ID is optional but recommended for workspace-specific operations.
• Audit logs are stored locally in the workspace root at .opencode/openwork/audit.jsonl.
• OpenWork server operates within approved workspace roots, limiting write access.