Chokepoint

Solutions / LLM Infrastructure & Governance / Chokepoint

Active — updated Aug 16, 2026

The single point every LLM call passes through

A drop-in reverse proxy for OpenAI, Anthropic, and other LLM APIs that enforces governance at the one place every request has to pass: PII and secret detection with redaction, policy-based block/redact/route rules, per-team cost tracking, immutable audit logs, and latency/error monitoring — shipped as a single self-hosted Go binary.

LicenseMIT
LanguageGo
DeploymentSelf-hosted binary
CategoryLLM infrastructure

What it does

PII & secret redaction

Scans every request and response for personal data and credentials, redacting before it ever reaches — or leaves — the model provider.

Policy-based routing

Define rules to block, redact, or route calls based on content, team, or model — enforced centrally instead of per-application.

Per-team cost tracking

Attribute spend across OpenAI, Anthropic, and other providers by team or project, without changing how your apps call the API.

Immutable audit logs

Every call is logged in a tamper-evident trail — what was sent, what was redacted, and what came back.

How it fits your stack

Your app
→
Chokepoint
→
OpenAI / Anthropic / any LLM API

Point your existing API base URL at Chokepoint instead of the provider directly. No SDK changes, no code rewrites — governance is applied in transit.

Get started

# pull and run
docker run -d -p 8080:8080 aihelpers/chokepoint

# point your app at Chokepoint instead of the provider
export OPENAI_BASE_URL=http://localhost:8080/v1

Related solutions

Distillery

No-code fine-tuning platform — train, deploy, and secure right-sized models with one click.

Just247Pipes

Hosted automation platform for wiring your tools and workflows together.