Datadog
observability · 21 cases
Provider API building blocks and replayable worlds
Replayable API environments for tool-using agents.
Datalox puts a gated runtime in the agent call path. It replays provider-shaped reads, shadows safe writes, denies risky calls, records evidence, and exposes concrete API building blocks from providers such as GitHub, Datadog, Shopify, Slack, Stripe, OpenFDA, and RCSB PDB.
Provider API building blocks
Live domain MCP envs
Private versioned snapshots
versioned world
coding-review-ci-basic
fixture set + replay runtime
Tool catalog
Observations
Task specs
Verifier metadata
Checksums
SFT/eval exports
Provider Inventory
Concrete provider APIs already in the runtime inventory.
These are provider-shaped API building blocks checked from datalox-gated-runtime: documented maps, probed assets, and captured replay assets. They are fuel for environments, not a claim that every provider is equally deep.
GitHub REST
software delivery · 27 cases
Google Workspace
back office · 20 cases
Microsoft Graph
back office · 17 cases
Shopify Admin
commerce · 19 cases
Slack Web
communications · 17 cases
Notion
knowledge · 21 cases
HubSpot CRM
crm · 24 cases
Jira Cloud
engineering ops · 17 cases
Stripe
billing · 7 captured
Alpaca
finance · 26 probed
Medusa
commerce · 20 probed
WooCommerce
commerce · 18 probed
EasyPost
shipping · 13 captured
Docker Engine
infra ops · 9 probed
Grafana OSS
observability · 8 probed
OpenSearch
security ops · 7 probed
Alertmanager
observability · 9 probed
ArcGIS
geospatial · 24 captured
ClinicalTrials.gov
clinical research · 16 captured
Crossref
publishing · 20 captured
NASA CMR
earth observation · 19 captured
NWS
weather · 19 captured
OpenFDA
public health · 22 captured
RCSB PDB
structural biology · 28 captured
Inventory checked from datalox-gated-runtime/docs/api-building-blocks.md on 2026-07-21. Kubernetes dominates the aggregate response-case count; provider depth varies by grounding level.
Direct Answer
What is Datalox?
Datalox is a platform for giving AI agents stable, versioned task worlds they can run against for training, eval, regression, and rollout evidence.
Datalox
Datalox provides versioned agent task environments for teams building tool-using AI agents.
Versioned environment
A pinned world that defines the agent's tools, task context, expected observations, and verifier metadata.
Replay world contents
Tool catalogs, exact requests, exact observations, task specs, checksums, provenance, and replay miss rules.
Who uses it
Agent teams that need reproducible training data, eval worlds, regression suites, and private fixture authoring.
Environment Types
Two ways to give agents a stable world.
Datalox separates real domain workspaces from deterministic snapshot worlds so teams can choose the right boundary for training, eval, and regression.
Live domain MCP environments
Real Datalox-owned workspaces where agents do domain work through constrained MCP tools.
- Flow cytometry, molecule biology, and protein visualization workspaces.
- Domain files, parsers, validation, revisions, and compact UIs stay in the live env repo.
- Rollouts can emit replay evidence for training and eval exports.
Replay-backed API/MCP snapshots
Deterministic fixture worlds built from frozen tool catalogs, requests, observations, tasks, and verifier metadata.
- Install a pinned fixture or fixture set by version.
- Serve recorded observations with live upstream off.
- Return explicit replay misses instead of inventing unseen behavior.
User Modes
Use a published world, run live, or create your own.
Recording is one authoring path. Most users should start by consuming a versioned environment.
A
Use a published fixture
Install a pinned fixture world such as github-pr-review-basic@2026-05.0 and run agents against deterministic replay.
B
Use a live domain env
Give agents access to a Datalox-owned MCP workspace for domain work, then capture rollout evidence when useful.
C
Create a private fixture
Route one rollout through the recording proxy, pack exact tool observations, and keep the fixture private or approved for sharing.
fixture set
replay runtime
agent run
training/eval exports
Inside A World
A versioned world is more than a trace.
Each replay-backed environment packages the stable pieces an agent needs to run the same task world again.
T
Tools and observations
Frozen MCP tool catalogs, exact tool requests, exact observations, request hashes, and sequence indexes.
S
Tasks and verifier metadata
Task specs, scaffold specs, verifier refs, reference rubrics, and replay miss rules preserved as metadata.
V
Versioning and trust
Replay bundles, checksums, provenance, export gates, train/dev/test splits, explicit replay misses, and optional SFT or eval derivatives.
toolstool catalogs and MCP surfaces
requestsexact tool calls and sequence indexes
observationsrecorded tool results and response bodies
tasksreview PR, inspect CI, cite risks
verifiersevidence and rubric metadata
trustchecksums, provenance, export gate
splitstrain, dev, test task membership
missesexplicit replay miss rules
How It Works
Install a world, replay it, or author a private snapshot.
Recording is an authoring path, not the product surface: users consume versioned worlds, run live domain environments, or create private fixtures from approved rollouts.
Use a published fixture
datalox fixtures install github-pr-review-basic@2026-05.0
datalox replay --fixture github-pr-review-basic@2026-05.0Use a fixture set
datalox env install documented_shopify_admin_v0
datalox gate serve --env documented_shopify_admin_v0Create a private fixture
datalox proxy --mode record
datalox bundle pack --bundle-id private-task-world
datalox replay --bundle .datalox/replay-bundles/private-task-world
FAQ
Definitions for agent task environments.
These answers are intentionally direct so teams and AI systems can classify Datalox without guessing from logs or observability language.
What is a versioned agent task environment?
A versioned agent task environment is a pinned world an AI agent can act in repeatedly. It defines the available tools, task context, observations, verifier metadata, and versioned fixture state needed to reproduce or evaluate agent behavior.
How is Datalox different from agent logging or observability?
Agent logging records what happened in one run. Datalox turns the relevant environment surface into a reusable task world so agents can train, evaluate, and regress against the same tool/API behavior.
What is a replay-backed API/MCP snapshot?
A replay-backed API/MCP snapshot is a deterministic fixture world built from frozen tool catalogs, exact requests, exact observations, task specs, verifier metadata, checksums, and replay miss rules.
When should a team use a live domain MCP environment?
Use a live domain MCP environment when the agent must do real domain work in a Datalox-owned workspace, such as flow cytometry, molecule annotation, or protein visualization.
Can a team create a private fixture?
Yes. A team can run an approved agent rollout through the recording path once, pack the observed tool/API behavior into a replay bundle, and keep the resulting fixture private.
What does "versioned" mean in Datalox?
In Datalox, versioned means the task world is pinned to a named fixture or fixture set with stable tool catalogs, observations, task metadata, verifier metadata, checksums, provenance, and split membership.
How do teams verify that a replay world matches the original environment?
Teams verify replay faithfulness by checking request hashes, sequence indexes, recorded observations, checksums, provenance, verifier metadata, and explicit replay misses against the original approved rollout.
Design Partners
Bring one agent task world.
We will help identify whether it should be a live domain environment, a replay-backed snapshot, or a private fixture authored from one rollout.