Anthropic checks Anthropic.
Claude, Codex, Gemini, and Grok can review the same PR. Delimit decides the AI-assisted merge and emits a signed, replayable attestation. Humans only see exceptions.
10 seconds. No API keys. No account. Works immediately.
Snapshots the diff. Runs lint + tests. Emits a signed, replayable attestation you can point an auditor at.
Try it right here
Edit the specs below or paste your own OpenAPI YAML. No backend required.
Same policies, same audit trail, same enforcement. No matter which assistant writes the code.
Claude Code
governed
Codex
Gemini CLI
Cursor
~/.delimit/ on disk. Every model reads the same policies, every session.
Your AI drafts. You approve. The daemon executes. Every action is typed, whitelisted, and logged — so scheduled work doesn't mean blind work.
Workers scan repos, PRs, and deliberation transcripts.
A bounded worker produces a typed work order.
You click once in the inbox. Audit row written.
The daemon runs a whitelisted action — never arbitrary shell.
Why this matters now
Other agent platforms cap scheduled runs at 15 a day. Delimit runs on your infrastructure with your quotas — hourly, every 5 minutes, or only on a governance signal. Pause from the dashboard when you need to, resume when you're ready. The audit trail keeps going either way.
One file. No config. Breaking change detection on every pull request.
name: API Governance
on: [pull_request]
jobs:
check:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: delimit-ai/delimit-action@v1
with:
spec: api/openapi.yamlEvery pull request gets a governance report. Breaking changes, semver classification, and a migration guide.
Models self-verify their outputs. Delimit provides the independent audit trail that proves it.
Claude forgets your rules when the session ends. Delimit persists them on disk in ~/.delimit/ -- every model reads the same policies, every time.
Which model deployed that breaking change? Delimit records every action with evidence -- who changed what, when, and which assistant did it.
Lint, semver classification, security audit, deploy gate. Automatic enforcement chain. Every time. No agent can skip a step.
npx delimit-cli setup creates a .delimit/policies.yml with your rules. Choose strict, default, or relaxed -- or write custom policies.
When any AI assistant modifies your API spec, Delimit auto-chains: diff, semver classification, governance evaluation, security audit, evidence collection. No manual steps.
The GitHub Action catches 27 types of API changes on every PR. Semver classified, migration guide posted, policy violations flagged -- before anyone clicks merge.
Every governance decision, every blocked deploy, every security finding -- recorded with evidence. Know which model changed what, and when.
Real numbers. Real infrastructure. No vaporware.
Whether you ship alone or at scale.
Catch breaking changes before they ship. 27 detectors, automatic semver, policy enforcement. Free tier covers everything a solo dev needs.
Get started freeShared governance policies across every developer and every AI assistant. One set of rules. Full audit trail. Scale from solo dev to 100-person team without changing tools.
View team plansApproval gates, compliance reporting, SSO, and data residency. Governance at the scale your security team demands.
Contact salesCore tools run entirely locally. Deliberation uses your own API keys if you configure them -- optional.
Your work lives in ~/.delimit/ -- local files, not cloud state. Switch models anytime. Cancel any subscription. Your data stays.
One command. Works with whatever AI assistant you already use.
Get notified when we ship new governance rules. No spam.