Work in your agent. Review with Refario.
The same plugin includes Codex and Claude Code manifests, hooks, and a review skill. Availability of model, reasoning, tool, and stop-boundary evidence depends on the host.
Verified.
Accepted.
Both checks matter.
- Objective verification
- Passed
- Human acceptance
- Yes
- Observed failed tools
- 2
- Monetary cost
- Unavailable
Clarify the failed tool’s usage instructions, then compare the next similar task.
Sample evidence, not a customer result or measured improvement.
One outcome model across two hosts.
Native hooks capture session boundaries and sanitized tool activity. On hosts without compatible hooks, the bundled scripts provide an explicit start and review path.
Codex
Install the local marketplace, enable Agent Performance, and start a fresh task to pick up the review skill.
Claude Code
Install the same download through its Claude marketplace manifest. Start a fresh session before measuring work.
Keep detailed evidence on your machine.
The plugin does not record prompts, transcripts, source contents, or tool inputs and outputs. Local records include task text, repository paths, and sanitized tool metadata. Optional uploads require an endpoint and key; use --no-upload to keep reviews local.
Already sending AI telemetry?
The existing API and SDK platform remains available for usage attribution and customer economics. That account is separate from installing the local performance plugin.
Make the next task a useful experiment.
Capture the evidence, verify the result, and decide what to try next.