← All implementations

Cloudflare · Platform

Internal AI engineering stack

Cloudflare's Dev Productivity team runs an internal AI engineering stack built on the company's own products. It puts MCP servers behind one OAuth portal, routes every model request through a gateway, and generates context files across thousands of repos. Every merge request gets an automated multi-agent review.

1 Supports2 Contextualizes

Supporting infrastructure

This entry describes supporting infrastructure that other work builds on. The catalog classifies it as a platform. The record also reports a workflow.

Approach type
Platform
Work
Coding, Code review
Human involvement
Drafts reviewed
Invocation
Event driven, Interactive
Interfaces
Cli, Ci, Web
Deployment stage
Scaled
Evidence strength
Detailed primary
Entry reviewed

How it works

The workflow the sources report for this implementation.

One OAuth aggregation point for all MCP tools

1

Collapse N tool schemas into 2 calls to hold token overhead constant at scale

1

Structured, generated repo context (runtime, nav, conventions, boundaries, deps)

1

Multi-agent CI review: risk tiering, specialist agents, Codex-rule citations

1

Where people stay involved

  • pull request → AI review findingsWork product review · Level 3

Level 3 for pull request → AI review findings; human attention boundary: work-product-review.

1

Observation date
2026-04-20

Implementation details

Sandbox

Dynamic Workers for sandboxed code execution; Sandbox SDK to clone/build/test

1

Harness

OpenCode + Windsurf clients; Agents SDK (McpAgent + Durable Objects) for stateful sessions

1

Model

Workers AI (open-weight, on-platform) + frontier models (Opus, GPT), routed by task

1

Interfaces

cli, ci, web

1

Tool access

MCP Server Portal; one OAuth point aggregating 182+ tools from 13 servers; AI Gateway for routing, cost, BYOK, ZDR

1

Knowledge

Backstage catalog (2,055 services) + AGENTS.md generated across ~3,900 repos

1

Credentials

Zero API keys on client machines; a Worker injects keys server-side; Cloudflare Access (Zero Trust) auth

1

Context management

Code Mode collapses upstream tool schemas into search + execute, holding token overhead constant at scale

1

Reported results and limitations

The catalog records what the sources report, with the scope and the denominator of every figure. A qualification below limits the figure it sits under.

Reported metrics

Headline claim

47.95 million AI requests in 30 days across the internal AI engineering system

1

The source does not report the denominator of this figure.

Reported by
Cloudflare
Scope
Internal AI engineering requests in the 30 days preceding the report
Method
Company-reported AI Gateway count for the preceding 30 days
Observation date
2026
Key observation

3,683 internal users (60% of company, 93% of R&D)

1

Reported by
Cloudflare
Scope
Active internal AI coding-tool users in the preceding 30 days
Denominator
Approximately 6,100 employees for company share; R&D organization for R&D share
Observation date
2026
Key observation

47.95M AI requests and 241.37B tokens via AI Gateway in the preceding 30 days

1

The source does not report the denominator of this figure.

Reported by
Cloudflare
Scope
Internal AI requests and AI Gateway tokens in the preceding 30 days
Method
Reported request counts and AI Gateway token counts
Observation date
2026
Key observation

10,952 merge requests in the week of March 23, 2026, nearly double the Q4 baseline; four-week average above 8,700

1

The source does not report the denominator of this figure.

Reported by
Cloudflare
Scope
Company merge requests in the week of March 23, 2026, versus Q4 baseline; not agent-authored PRs
Method
Weekly merge-request count; distinct from the four-week rolling average
Observation date
2026-03
Key observation

295 teams using agentic AI tools

1

The source does not report the denominator of this figure.

Reported by
Cloudflare
Scope
Teams using agentic AI tools and coding assistants in the reported 30-day snapshot
Observation date
2026

Lessons and interpretation

Centralize through a proxy early; direct-to-gateway looks simpler but blocks per-user attribution, model cataloging, and policy later

1

Without structured data, agents are working blind; they read code but can't see the system around it (Backstage / AGENTS.md)

1

Tool schemas eat context (34 GitLab tools ≈ 7.5% of a 200K window); collapse them at the portal

1

Frontier + open-source hybrid: route a growing share of workloads to cheaper self-hosted models

1

Sources and research details

Citations link to the original publisher. Each source also keeps a preserved copy in the repository, so a changed or removed page stays checkable.

  1. The AI engineering stack we built internallyhttps://blog.cloudflare.com/internal-ai-engineering-stack/Engineering blog · First party · Last source verification: 2026-08-31
    Preserved copy ↗
  2. Hacker News discussion of Cloudflare's internal AI engineering stackhttps://news.ycombinator.com/item?id=47837240Hn thread · Community · Last source verification: 2026-08-31
    Preserved copy ↗
Research details for every claim on this page
  1. Summary
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  2. Headline claim
    Statement type
    Metric
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
    Reported by
    Cloudflare
    Scope
    Internal AI engineering requests in the 30 days preceding the report
    Denominator
    Not reported
    Method
    Company-reported AI Gateway count for the preceding 30 days
    Observation date
    2026
  3. Sandbox
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  4. Harness
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  5. Model
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  6. Interfaces
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  7. Tool access
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  8. Knowledge
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  9. Credentials
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  10. Context management
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  11. Supporting component
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  12. Supporting component
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  13. Supporting component
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  14. Supporting component
    Statement type
    Fact
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
  15. Key observation
    Statement type
    Metric
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
    Reported by
    Cloudflare
    Scope
    Active internal AI coding-tool users in the preceding 30 days
    Denominator
    Approximately 6,100 employees for company share; R&D organization for R&D share
    Method
    Not reported
    Observation date
    2026
  16. Key observation
    Statement type
    Metric
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
    Reported by
    Cloudflare
    Scope
    Internal AI requests and AI Gateway tokens in the preceding 30 days
    Denominator
    Not reported
    Method
    Reported request counts and AI Gateway token counts
    Observation date
    2026
  17. Key observation
    Statement type
    Metric
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
    Reported by
    Cloudflare
    Scope
    Company merge requests in the week of March 23, 2026, versus Q4 baseline; not agent-authored PRs
    Denominator
    Not reported
    Method
    Weekly merge-request count; distinct from the four-week rolling average
    Observation date
    2026-03
  18. Key observation
    Statement type
    Metric
    Provenance
    Reported
    Confidence
    High
    Confidence reason
    A linked first-party source states the claim.
    Reported by
    Cloudflare
    Scope
    Teams using agentic AI tools and coding assistants in the reported 30-day snapshot
    Denominator
    Not reported
    Method
    Not reported
    Observation date
    2026
  19. Lesson
    Statement type
    Inference
    Provenance
    Catalog judgment
    Confidence
    Medium
    Confidence reason
    The catalog derives this observation from the linked sources.
  20. Lesson
    Statement type
    Inference
    Provenance
    Catalog judgment
    Confidence
    Medium
    Confidence reason
    The catalog derives this observation from the linked sources.
  21. Lesson
    Statement type
    Inference
    Provenance
    Catalog judgment
    Confidence
    Medium
    Confidence reason
    The catalog derives this observation from the linked sources.
  22. Lesson
    Statement type
    Inference
    Provenance
    Catalog judgment
    Confidence
    Medium
    Confidence reason
    The catalog derives this observation from the linked sources.
  23. Operating model assessment
    Statement type
    Inference
    Provenance
    Catalog judgment
    Confidence
    Medium
    Confidence reason
    The source documents automated review findings, but the record combines several platform workflows.
    Observation date
    2026-04-20
Back to the directory →