# reflect MCP server

> Analyzes AI coding agent performance by capturing telemetry, measuring token costs, and tracking the impact of automated engineering workflows.

- Repository: https://github.com/o11y-dev/reflect
- Page: https://hostedmcp.org/servers/o11y-dev/reflect
- Transport: —
- Stars: 20
- License: Apache-2.0

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `REFLECT_OTLP_MAX_BYTES` | no | no |  |
| `REFLECT_OTLP_SEGMENT_BYTES` | no | no |  |
| `REFLECT_REFRESH_INTERVAL_SECONDS` | no | no |  |

## Tools (5)

- `reflect_context`
- `reflect_record_milestone`
- `reflect_complete`
- `reflect_review_change`
- `reflect_apply_change`

## FAQ

### Does Reflect require a cloud account or hosted backend?

No, Reflect is local-first. No hosted backend or Reflect account is required, as all telemetry and state remain on your machine.

### What operating systems are supported?

Reflect supports macOS and Linux and requires Python 3.12 or higher.

### How does Reflect handle data privacy and text capture?

Prompt and response text capture is opt-in, offering metadata-only, masked-text, and full-text modes during setup.

### Can Reflect automatically change my workflows?

No, while it can inspect evidence and prepare reviews, it cannot install or change a workflow without explicit human approval.