$ show darrenhinde/OpenAgentsControl
OpenAgents Control (OAC) is a framework for managing, testing, and orchestrating AI coding agents. It provides a CLI tool (`oac`) to install, update, and manage agents, context files, and skills from a registry. The system includes a plugin architecture for OpenCode and a compatibility layer that converts agent definitions between OAC, Claude Code, Cursor, and Windsurf formats. A comprehensive evaluation framework tests agent behavior against critical rules (approval gates, context loading, stop-on-failure, delegation, tool usage) using YAML-defined test suites and timeline-based analysis. The repository is primarily a software development toolchain for AI-assisted coding workflows, with strong emphasis on deterministic safety rules and multi-agent coordination.
The LLM identified 5 files worth highlighting for reuse:
Central orchestrator for converting agent configurations between OAC, Cursor, Claude, and Windsurf. Coordinates mappers (ToolMapper, PermissionMapper, ModelMapper, ContextMapper) and returns warnings for lossy conversions. Demonstrates the adapter pattern with graceful degradation: each mapper returns a result object with exact/fuzzy conversion status and warnings, enabling the engine to report compatibility issues without failing. Reusable pattern for any cross-platform configuration translation.
Abstract base class for all evaluators in the testing framework. Provides helper methods for filtering tool calls from timelines (`getToolCallsByName`, `getEventsBefore/After`), creating evidence and violation objects, and building standardized evaluation results with scores. Concrete evaluators (ApprovalGate, ContextLoading, StopOnFailure, etc.) extend this class, implementing only the `evaluate` method. This template-method pattern enables consistent violation collection and scoring across all agent behavior tests.
Pure functions for managing a JSON manifest file that tracks installed components (agents, contexts, skills). Supports CRUD operations (`createEmptyManifest`, `addFileToManifest`, `removeFileFromManifest`, `updateFileHash`) with SHA256 hash verification for drift detection and clean removal. The functions are deterministic, have no side effects beyond the manifest object, and are fully testable. Reusable pattern for any file-based component installation system.
Implements a smart approval strategy that can auto-approve or auto-deny tool calls based on configured allowed/denied patterns (e.g., paths, tool names). Supports a maximum approval count limit and fallback to manual approval for uncategorized requests. This is a composable, testable strategy that can be swapped with `AutoApproveStrategy` or `AutoDenyStrategy`. Demonstrates the Strategy pattern for approval decisions in agent evaluation.
Manages concurrent execution of ability steps with support for limiting concurrent executions, cleanup on abort, and status tracking. The `ExecutionManager` can start, stop, and track multiple ability workflows, ensuring proper cleanup and preventing resource leaks. This is a self-contained concurrency control module that could be reused in any workflow engine requiring step-based execution with cancellation support.