Future Roadmap¶

MCPGuard Integration¶

Record MCP tool usage, permissions, policy decisions, and blocked operations as first-class evidence in a run.

Attach benchmark task IDs, regression results, and evaluation scores to AgentTrace reports.

Validate runs against prompt contracts, required evidence checklists, and organization-specific completion rules.

Add a local or hosted dashboard for browsing runs, risky files, failing tests, review coverage, and agent tool adoption.

Generate PR comments, attach reports to pull requests, and link AgentTrace run IDs to commits or branches.