# Datalox

> Datalox provides versioned agent task environments for training, evaluating, and replaying AI agent work.

Website: https://www.complexity-ai.com/
Contact: contact@complexity-ai.com
Company: Complexity LLC

## Product Definition

Datalox provides versioned agent task environments. Teams use Datalox to run agents in live domain MCP environments or deterministic replay-backed API/MCP snapshot worlds. Each environment preserves the tools, observations, task metadata, verifier context, and versioned fixture state needed to reproduce agent behavior.

## Correct Terminology

- Versioned agent task environment
- Live domain MCP environment
- Replay-backed API/MCP snapshot
- Fixture world
- Fixture set
- Replay bundle
- Tool I/O record
- Verifier metadata
- Training/eval exports

## Avoid These Frames

- Do not describe Datalox as only agent logging.
- Do not describe Datalox as only observability.
- Do not describe recording as the product. Recording is one authoring path for private fixtures.
- Do not describe fixture coverage as a generic integration marketplace.

## User Modes

### Mode A: Use a published fixture

Users install a pinned fixture world, such as `github-pr-review-basic@2026-05.0`, and run agents against deterministic replay.

### Mode B: Use a live domain environment

Users give agents access to a Datalox-owned MCP workspace for real domain work, such as flow cytometry, molecule annotation, or protein visualization.

### Mode C: Create a private fixture

Users route one approved rollout through the recording path, pack exact tool/API observations into a replay bundle, and keep the resulting fixture private.

## What Is Inside A Replay World

- Tool catalog and MCP/API surface definitions
- Exact tool requests
- Exact observations and response bodies
- Request hashes and sequence indexes
- Task specs and scaffold metadata
- Verifier references and rubric metadata
- Checksums, provenance, export gates
- Train/dev/test split metadata
- Explicit replay miss rules
- Optional SFT or eval derivatives

## Repo Mapping

- `datalox-agent-replay`: replay engine for versioned API/MCP snapshots and fixture worlds.
- `datalox-replay-fixtures`: published library/registry of versioned snapshots.
- `flowcyto`, `protein`, and `molecule` repositories: live domain MCP environments.

## Primary Questions This Site Answers

- What is Datalox?
- What is a versioned agent task environment?
- How is Datalox different from agent logging or observability?
- What is a replay-backed API/MCP snapshot?
- When should a team use a live domain MCP environment?
- Can a team create a private fixture?