{"id":34726793,"url":"https://github.com/getaxonflow/axonflow","last_synced_at":"2026-04-05T15:00:53.742Z","repository":{"id":329436448,"uuid":"1108435661","full_name":"getaxonflow/axonflow","owner":"getaxonflow","description":"AxonFlow: Runtime control layer for production AI","archived":false,"fork":false,"pushed_at":"2026-04-05T13:02:08.000Z","size":11935,"stargazers_count":43,"open_issues_count":1,"forks_count":2,"subscribers_count":1,"default_branch":"main","last_synced_at":"2026-04-05T13:09:07.290Z","etag":null,"topics":["agent-runtime","agentic-ai","ai-control-plane","ai-governance","ai-observability","ai-orchestration","ai-workflow","deterministic-workflows","enterprise-ai","execution-engine","llm-security","mcp","multi-agent-systems","responsible-ai","runtime-infra","workflow-engine"],"latest_commit_sha":null,"homepage":"https://docs.getaxonflow.com","language":"Go","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/getaxonflow.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":"CODE_OF_CONDUCT.md","threat_model":null,"audit":null,"citation":null,"codeowners":".github/CODEOWNERS","security":"SECURITY.md","support":null,"governance":"docs/governance/cost-controls.md","roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-12-02T12:54:19.000Z","updated_at":"2026-04-05T13:00:46.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/getaxonflow/axonflow","commit_stats":null,"previous_names":["getaxonflow/axonflow"],"tags_count":45,"template":false,"template_full_name":null,"purl":"pkg:github/getaxonflow/axonflow","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/getaxonflow%2Faxonflow","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/getaxonflow%2Faxonflow/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/getaxonflow%2Faxonflow/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/getaxonflow%2Faxonflow/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/getaxonflow","download_url":"https://codeload.github.com/getaxonflow/axonflow/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/getaxonflow%2Faxonflow/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31439442,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-05T13:13:19.330Z","status":"ssl_error","status_checked_at":"2026-04-05T13:13:17.778Z","response_time":75,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agent-runtime","agentic-ai","ai-control-plane","ai-governance","ai-observability","ai-orchestration","ai-workflow","deterministic-workflows","enterprise-ai","execution-engine","llm-security","mcp","multi-agent-systems","responsible-ai","runtime-infra","workflow-engine"],"created_at":"2025-12-25T02:30:28.288Z","updated_at":"2026-04-05T15:00:53.729Z","avatar_url":"https://github.com/getaxonflow.png","language":"Go","readme":"# AxonFlow\n\n**AxonFlow is the execution authority and system of record for AI decisions in production workflows.**\n\nIt operates inside the execution path between your workflow logic and model or tool calls. Gateways can help at the request boundary and observability tools can tell you what happened later. AxonFlow records why an action was allowed, blocked, paused, or resumed while the workflow is running.\n\nIt runs self-hosted (Docker or Kubernetes), with SDKs for **Python**, **TypeScript**, **Go**, and **Java**.\n\n## Why AxonFlow Exists\n\nProduction AI systems are multi-step, non-deterministic, and increasingly regulated. In practice:\n\n- Prompt filters alone do not control downstream tool execution.\n- Orchestration frameworks coordinate steps, but do not decide whether a risky step should execute.\n- Routing gateways improve connectivity, but do not give you workflow-level approvals or safe resume semantics.\n- Logs show that something happened, but not always why it was allowed or who owned the decision path.\n\nAxonFlow addresses this with a runtime execution layer that enforces policy at the step boundary and records decision context at execution time.\n\n## What AxonFlow Is\n\nAxonFlow is composed of:\n\n- **Agent runtime** for inline policy evaluation and execution checks (`:8080`)\n- **Orchestrator runtime** for workflow execution state and routing (`:8081`)\n- **Policy engine** for tenant and org governance policies\n- **Workflow Control Plane (WCP)** for gated, step-level workflow execution\n- **Decision record and evidence layer** for replay, export, and compliance workflows\n\nExecution modes:\n\n- **Gateway Mode**: Pre-check + your own LLM call + audit\n- **Proxy Mode**: AxonFlow enforces and proxies model/tool execution\n- **WCP Mode**: Governed multi-step workflow execution with step gates\n\nAxonFlow is not a workflow engine, observability dashboard, or prompt gateway. Your application or orchestrator still decides what to do next. AxonFlow decides whether the next model or tool action can run and records that decision.\n\n## What AxonFlow Does\n\n**Policy Enforcement** — 60+ built-in policies across multiple categories:\n- **Security**: SQL injection detection (37 patterns), unsafe admin access, schema exposure\n- **Sensitive Data**: PII detection (SSN, credit cards, PAN, Aadhaar, email, phone), salary, medical records\n- **Compliance**: GDPR, PCI-DSS, HIPAA basic constraints (Community); EU AI Act, SEBI/RBI, MAS FEAT, DORA frameworks with retention and exports (Enterprise)\n- **Runtime Controls**: Tenant isolation, environment restrictions, approval gates\n- **Cost \u0026 Abuse**: Per-user/team limits, anomalous usage detection, token budgets\n\nAll policies are configurable. Teams typically start in observe-only mode and enable blocking once they trust the signal.\n\n\u003e **[Full policy documentation](https://docs.getaxonflow.com/docs/policies/overview)** · **[Community vs Enterprise](https://docs.getaxonflow.com/docs/features/community-vs-enterprise?utm_source=readme_eval)**\n\n**Human-in-the-Loop Approval Gates** — Require explicit approvals for high-risk workflow steps. Configurable expiry, pending limits by tier, and automatic workflow abort on expiration.\n\n**Policy Simulation \u0026 Impact Reporting** — Dry-run policy changes against historical traffic before deploying. See which requests would be blocked, allowed, or changed.\n\n**Evidence Export** — Generate compliance-ready audit evidence packs with configurable retention windows. Designed for internal governance reviews and regulatory audits.\n\n**Workflow Control Plane (WCP)** — Govern long-running, multi-step AI workflows with step-level gate checks, a durable step ledger, cancellation, and SSE streaming. WCP works with any orchestration framework — your code controls execution, AxonFlow controls governance and visibility.\n\n**Multi-Agent Planning (MAP)** — Define agents in YAML, let AxonFlow turn natural language requests into executable workflows with automatic plan generation and execution tracking.\n\n**SQL Injection Response Scanning** — Detect SQLi payloads in MCP connector responses. Protects against data exfiltration when compromised data is returned from databases.\n\n**Media Governance** — Governance pipeline for image inputs, including content classification and policy enforcement on multimodal requests.\n\n**Code Governance** — Detect LLM-generated code, identify language and security issues (secrets, eval, shell injection). Logged for compliance.\n\n**Decision Records \u0026 Audit Trails** — Every request and governed step is recorded with decision context. Know what was blocked, why it was blocked, and which policy or approval path applied. Token usage tracked for cost analysis.\n\n**Decision \u0026 Execution Replay** — Debug governed workflows with step-by-step state and policy decisions. Timeline view and compliance exports included.\n\n**Cost Controls** — Set budgets at org, team, agent, or user level. Track LLM spend across providers with configurable alerts and enforcement actions.\n\n**Multi-Model Routing** — Route requests across OpenAI, Anthropic, Bedrock, Ollama based on cost, capability, or compliance requirements. Failover included.\n\n**Circuit Breaker** — Emergency kill switch wired into the request pipeline. Instantly halt all LLM traffic when something goes wrong in production.\n\n**Proxy Mode** — Full request lifecycle: policy, planning, routing, audit. Recommended for new projects.\n\n**Gateway Mode** — Request-boundary governance for existing stacks. Pre-check → your call → audit.\n\n\u003e **[Choosing a mode](https://docs.getaxonflow.com/docs/sdk/choosing-a-mode)** · **[Architecture deep-dive](https://docs.getaxonflow.com/docs/architecture/overview)**\n\n## Who This Is For\n\n**Good fit:**\n- Production AI teams needing governance before shipping\n- Platform teams building internal AI infrastructure\n- Regulated industries (healthcare, finance, legal) with compliance requirements\n- Teams wanting audit trails and policy enforcement without building it themselves\n- Teams running multi-step agent workflows that need execution control, retries, and step-level visibility\n\n**Not a good fit:**\n- Single-prompt experiments or notebooks\n- Prototypes where governance isn't a concern yet\n- Projects where adding a service layer is overkill\n\n**[Full Documentation](https://docs.getaxonflow.com)** · **[Getting Started Guide](https://docs.getaxonflow.com/docs/getting-started)** · **[API Reference](./docs/api/)**\n\n**2-minute demo:** See AxonFlow enforcing runtime policies and execution control in a real workflow — [Watch on YouTube](https://youtu.be/BSqU1z0xxCo)\n\n**Architecture deep dive (12 min):** How the control plane works, policy enforcement flow, and multi-agent planning — [Watch on YouTube](https://youtu.be/Q2CZ1qnquhg)\n\n---\n\n## Pick Your First 10-Minute Path\n\nIf you're adding governance to an existing AI stack (LangChain, CrewAI, direct API calls), start with Path A. If you're building new multi-step agent workflows that need execution control, start with Path B.\n\n### Path A: Govern Existing LLM Calls\n\nAdd policy enforcement, PII detection, and audit trails to your current AI stack — without changing your orchestration logic.\n\n```bash\n# Gateway Mode: Pre-check → Your LLM call → Audit\ncurl -X POST http://localhost:8080/api/policy/pre-check \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\"user_token\": \"demo-user\", \"client_id\": \"demo-client\", \"query\": \"Look up customer with SSN 123-45-6789\"}'\n# Returns: {\"approved\": true, \"requires_redaction\": true, \"pii_detected\": [\"ssn\"]}\n```\n\nWorks with LangChain, CrewAI, or any framework — AxonFlow acts as a governance sidecar.\n\n\u003e **[Choosing a mode guide](https://docs.getaxonflow.com/docs/sdk/choosing-a-mode)** — covers Gateway Mode, Proxy Mode, and when to use each.\n\n### Path B: Execution Control for Long-Running Workflows\n\nUse the Workflow Control Plane (WCP) to manage multi-step AI workflows with step-level gates, a durable ledger, cancellation, and SSE streaming.\n\n```bash\n# Run the execution tracking demo (requires Docker services running)\n./examples/execution-tracking/http/example.sh\n```\n\nThis creates a WCP workflow, runs step-level gate checks, records a step ledger, demonstrates cancellation, and shows unified execution status.\n\n\u003e **[Execution tracking guide](https://docs.getaxonflow.com/docs/orchestration/wcp/overview)** — WCP workflow creation, step gates, SSE streaming, and unified execution status.\n\n---\n\n## Quick Start\n\nIf you want to see how this looks before setting it up, here's a short demo: [2-minute walkthrough](https://youtu.be/BSqU1z0xxCo)\n\n**Prerequisites:** [Docker Desktop](https://docs.docker.com/get-docker/) installed and running.\n\n```bash\n# Clone and start\ngit clone https://github.com/getaxonflow/axonflow.git\ncd axonflow\n\n# Set your API key (at least one LLM provider required for AI features)\necho \"OPENAI_API_KEY=sk-your-key-here\" \u003e .env   # or ANTHROPIC_API_KEY\n\n# Start services\ndocker compose up -d\n\n# Wait for services to be healthy (~30 seconds)\ndocker compose ps   # All services should show \"healthy\"\n\n# Verify it's running\ncurl http://localhost:8080/health\ncurl http://localhost:8081/health\n```\n\n**That's it.** Services are now running:\n\n| Service | URL | Purpose |\n|---------|-----|---------|\n| Agent | http://localhost:8080 | Policy enforcement, PII detection |\n| Orchestrator | http://localhost:8081 | LLM routing, WCP, tenant policies |\n| Grafana | http://localhost:3000 | Dashboards (admin / grafana_localdev456) |\n| Prometheus | http://localhost:9090 | Metrics |\n\n\u003e **Note:** All commands in this README assume you're in the repository root directory (`cd axonflow`).\n\n### Execution Control Demo (60 seconds)\n\nWith services running, try the execution control workflow:\n\n```bash\n./examples/execution-tracking/http/example.sh\n```\n\nThis demonstrates:\n1. **MAP plan creation** via `/api/request`\n2. **WCP workflow** with step-level gate checks and completion\n3. **Cancellation** via the unified execution API\n4. **SSE streaming** for real-time execution events\n5. **Workflow listing** and status queries\n\n### Supported LLM Providers\n\n| Provider | Community | Enterprise | Notes |\n|----------|:---------:|:----------:|-------|\n| **OpenAI** | ✅ | ✅ | GPT-5.x, GPT-4o, GPT-4 |\n| **Anthropic** | ✅ | ✅ | Claude Opus 4.6, Claude Sonnet 4.6 |\n| **Azure OpenAI** | ✅ | ✅ | Azure AI Foundry \u0026 Classic endpoints |\n| **Google Gemini** | ✅ | ✅ | Gemini 3.x (Pro, Flash, Flash-Lite) |\n| **Ollama** | ✅ | ✅ | Local/air-gapped deployments |\n| **AWS Bedrock** | ❌ | ✅ | HIPAA-compliant, data residency |\n\n\u003e LLM provider configuration applies to Proxy Mode and MAP, where AxonFlow routes requests to the provider.\n\u003e In Gateway Mode and WCP, your application calls the LLM directly, including via frameworks like LangChain or CrewAI, so any provider works.\n\n\u003e **[Provider configuration guide](https://docs.getaxonflow.com/docs/llm/overview)**\n\n### See Governance in Action (30 seconds)\n\n```bash\n# Example: Send a request containing an SSN — AxonFlow detects and flags it for redaction\ncurl -X POST http://localhost:8080/api/policy/pre-check \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\"user_token\": \"demo-user\", \"client_id\": \"demo-client\", \"query\": \"Look up customer with SSN 123-45-6789\"}'\n```\n\n```json\n{\"approved\": true, \"requires_redaction\": true, \"pii_detected\": [\"ssn\"], \"policies\": [\"pii_ssn_detection\"]}\n```\n\n### Full Interactive Demo (10 min)\n\nExperience the complete governance suite: PII detection, SQL injection blocking,\nproxy and gateway modes, MCP connectors, multi-agent planning, and observability.\n\n**Requires:** Python 3.10+ (for demo scripts)\n\n```bash\n# Ensure your .env has a valid API key\ncat .env   # Should show OPENAI_API_KEY=sk-... or ANTHROPIC_API_KEY=sk-ant-...\n\n# Restart services if you just added the key\ndocker compose up -d --force-recreate\n\n# Run the interactive demo\n./examples/demo/demo.sh\n```\n\nThe demo walks through a realistic customer support scenario with live LLM calls.\nSee [`examples/demo/README.md`](examples/demo/README.md) for options (`--quick`, `--part N`).\n\n---\n\nAxonFlow runs inline with LLM traffic, enforcing policies and routing decisions in single-digit milliseconds — fast enough to prevent failures rather than observe them after the fact.\n\n---\n\n### Integration Options\n\nFor Go, Java, Python, and TypeScript applications, we recommend using the **[AxonFlow SDKs](https://docs.getaxonflow.com/docs/sdk/overview)**. All SDKs are thin wrappers over the same REST APIs, which remain fully supported for custom integrations.\n\n| Integration | Recommended For |\n|-------------|-----------------|\n| **SDKs** | Application code, services, strongly typed environments |\n| **HTTP APIs** | Agents, automation, CLI tools, CI pipelines, languages without SDKs |\n\nAll features—policy enforcement, audit logging, MCP connectors, WCP workflows—are available via both SDKs and HTTP.\n\nFor AI agent runtimes, see:\n- [**OpenClaw**](https://docs.getaxonflow.com/docs/integration/openclaw/) — policy enforcement, approval gates, and audit trails for OpenClaw tool execution\n- [**Anthropic Computer Use**](https://docs.getaxonflow.com/docs/integration/computer-use/) — governed desktop and tool actions\n- [**Claude Agent SDK**](https://docs.getaxonflow.com/docs/integration/claude-agent-sdk/) — MCP tool governance patterns\n\n\u003e **[SDK Documentation](https://docs.getaxonflow.com/docs/sdk/overview)** · **[API Reference](./docs/api/)**\n\n### vs LangChain / LangSmith\n\n| Feature | AxonFlow | LangChain/LangSmith |\n|---------|----------|---------------------|\n| **Governance** | Inline policy enforcement | Post-hoc monitoring |\n| **Architecture** | Active prevention | Passive detection (observability) |\n| **Workflow Execution Control** | Step-level gates, durable ledger, cancellation | Chain sequencing only |\n| **Evidence \u0026 Replay** | Compliance exports, decision replay, audit retention | Trace logging |\n| **Enterprise Focus** | Built for compliance \u0026 security first | Developer-first framework |\n| **Multi-Tenant** | Production-ready isolation | DIY multi-tenancy |\n| **Self-Hosted** | Full core available | Partial (monitoring requires cloud) |\n\n**The Key Difference:** LangChain/LangSmith focus on observability and post-hoc analysis, while AxonFlow enforces policies inline during request execution.\n\n**Best of Both Worlds:** Many teams use LangChain for orchestration logic with AxonFlow as the governance layer on top.\n\n---\n\n## Architecture\n\n```\n┌─────────────┐    ┌──────────────────────────────────────────────────┐\n│  Your App   │───▶│                Agent (:8080)                     │\n│   (SDK)     │    │  ┌──────────┐ ┌──────────┐ ┌──────────────────┐ │\n└─────────────┘    │  │  Policy  │ │   MCP    │ │  Media / Code    │ │\n                   │  │  Engine  │ │Connectors│ │  Governance      │ │\n                   │  │  (60+)   │ │          │ │                  │ │\n                   │  └──────────┘ └──────────┘ └──────────────────┘ │\n                   │  ┌──────────────────┐ ┌─────────────────────┐   │\n                   │  │ PII / SQLi       │ │ Circuit Breaker     │   │\n                   │  │ Detection        │ │ (Kill Switch)       │   │\n                   │  └──────────────────┘ └─────────────────────┘   │\n                   └────────────────────┬─────────────────────────────┘\n                                        │\n                                        ▼\n                   ┌──────────────────────────────────────────────────┐\n                   │             Orchestrator (:8081)                 │\n                   │  ┌──────────┐ ┌──────────┐ ┌──────────────────┐ │\n                   │  │   WCP    │ │  MAP     │ │  Cost Controls   │ │\n                   │  │  Step    │ │  Plan +  │ │  \u0026 Multi-Model   │ │\n                   │  │  Gates   │ │  Execute │ │  Routing         │ │\n                   │  └──────────┘ └──────────┘ └──────────────────┘ │\n                   │  ┌──────────────────┐ ┌─────────────────────┐   │\n                   │  │ HITL Approval    │ │ Evidence Export      │   │\n                   │  │ Gates            │ │ \u0026 Decision Replay    │   │\n                   │  └──────────────────┘ └─────────────────────┘   │\n                   └────────────────────┬─────────────────────────────┘\n                                        │\n                                        ▼\n                   ┌──────────────────────────────────────────────────┐\n                   │                 LLM Providers                    │\n                   │   (OpenAI, Anthropic, Gemini, Bedrock, Ollama)   │\n                   └──────────────────────────────────────────────────┘\n\n           PostgreSQL (policies, audit, evidence) • Redis (cache)\n```\n\n- **Agent** (:8080): Policy enforcement, PII detection, SQLi response scanning, media/code governance, MCP connectors, circuit breaker\n- **Orchestrator** (:8081): WCP step gates, multi-agent planning, LLM routing, cost controls, HITL approval gates, evidence export, decision replay\n\n### Why AxonFlow often becomes the default control plane\n\nTeams typically start by placing AxonFlow in front of a single workflow or agent to evaluate policy enforcement, auditability, and execution control. As usage grows, running some AI workflows through AxonFlow and others outside it tends to create fragmented audit logs, inconsistent policies, and duplicated observability. For this reason, teams that continue using AxonFlow often standardize on it as a single control plane for AI workflows, while retaining their existing orchestration frameworks and execution logic.\n\n### What evaluators are saying\n\n\u003e *\"So many teams are just making the first steps, which work quite easily in a localhost environment, but then when they try to move this to production they find that they have to build so much core infrastructure. I am happy to see this solution looking so much more mature and feature rich.\"*\n\u003e — Principal Technical Product Manager · Global travel marketplace\n\n\u003e *\"Most failures aren't model quality issues. The idea of a lightweight, inline control plane that doesn't replace the orchestrator but governs execution step-by-step feels like a pragmatic way to tackle that.\"*\n\u003e — Staff Software Engineer · Global travel marketplace\n\n\u003e *\"This is massively needed. Large corporations are building their own frameworks but a new business looking to go agentic can't do it without this.\"*\n\u003e — Principal Product Manager, AI/ML · Global payments platform\n\n\u003e *\"Your product gives a no-fluff approach to bolt the security early on and not as an afterthought.\"*\n\u003e — Principal Engineer · Global travel marketplace\n\n*Quotes are anonymized and lightly edited for clarity.*\n\n---\n\n## Three-Tier Licensing\n\nAxonFlow offers three tiers. Community is free with no license key. Evaluation is free with a license key and unlocks governance features designed for teams taking AI to production:\n\n| Feature | Community | Evaluation (Free) | Enterprise |\n|---------|-----------|-------------------|------------|\n| Tenant policies | 20 | 50 | Unlimited |\n| Org-wide policies | 0 | 5 | Unlimited |\n| Audit retention | 3 days | 14 days | 3650 days |\n| Concurrent executions | 5 | 25 | Unlimited |\n| HITL Approval Gates | — | 100 pending, 24h expiry | Unlimited, configurable expiry |\n| Policy Simulation | — | 300/day | Unlimited |\n| Evidence Export | — | 14-day window, 3/day | Unlimited |\n\n[Get a free Evaluation license](https://getaxonflow.com/evaluation-license?utm_source=readme_eval) · [Full feature matrix](https://docs.getaxonflow.com/docs/features/community-vs-enterprise?utm_source=readme_eval)\n\n### Stay on Community if:\n- Single team prototyping AI features\n- Development and local evaluation\n- Your limits fit in Community capacity (20 tenant policies, 5 concurrent executions)\n\n### Upgrade to Evaluation (Free) when:\n- Taking AI to production with a small team\n- Need Human-in-the-Loop approval gates for governed workflows\n- Want to simulate policy changes before deploying them (dry-run)\n- Need evidence exports for compliance proof or audit prep\n- Need organization-wide policies (up to 5) and 14-day audit retention\n\n**Get your free Evaluation license:** https://getaxonflow.com/evaluation-license\n\n### You need Enterprise when:\n\n**Identity \u0026 Organization Controls**\n- SSO + SAML authentication\n- SCIM user lifecycle management\n- Multi-tenant isolation\n\n**Compliance \u0026 Risk**\n- EU AI Act conformity workflows + 10-year retention\n- SEBI/RBI compliance exports + 5-year retention\n- Unlimited HITL approval queues with configurable expiry\n- Emergency circuit breaker (kill switch)\n\n**Platform \u0026 Operations**\n- One-click AWS CloudFormation deployment\n- Usage analytics and cost attribution\n- Priority support with SLA\n- Customer Portal UI for runtime management\n\nSee the full **[Community vs Evaluation vs Enterprise feature matrix](https://docs.getaxonflow.com/docs/features/community-vs-enterprise?utm_source=readme_eval)**\n*(designed for security reviews, procurement, and platform evaluations)*\n\n**Enterprise:** [AWS Marketplace](https://aws.amazon.com/marketplace) or [sales@getaxonflow.com](mailto:sales@getaxonflow.com)\n\n---\n\n## SDKs\n\n```bash\npip install axonflow          # Python\nnpm install @axonflow/sdk     # TypeScript\ngo get github.com/getaxonflow/axonflow-sdk-go/v4  # Go\n```\n\n```xml\n\u003c!-- Java (Maven) --\u003e\n\u003cdependency\u003e\n    \u003cgroupId\u003ecom.getaxonflow\u003c/groupId\u003e\n    \u003cartifactId\u003eaxonflow-sdk\u003c/artifactId\u003e\n    \u003cversion\u003e4.0.0\u003c/version\u003e\n\u003c/dependency\u003e\n```\n\n### Python\n\n```python\nfrom axonflow import AxonFlow\n\nasync with AxonFlow(endpoint=\"http://localhost:8080\") as ax:\n    response = await ax.proxy_llm_call(\n        user_token=\"user-123\",\n        query=\"Analyze customer sentiment\",\n        request_type=\"chat\"\n    )\n```\n\n### TypeScript\n\n```typescript\nimport { AxonFlow } from '@axonflow/sdk';\n\nconst axonflow = new AxonFlow({\n  endpoint: 'http://localhost:8080',\n  clientId: 'my-app',\n  clientSecret: 'my-secret'\n});\n\nconst response = await axonflow.proxyLLMCall({\n  userToken: 'user-123',\n  query: 'Analyze customer sentiment',\n  requestType: 'chat'\n});\n```\n\n### Go\n\n```go\nimport axonflow \"github.com/getaxonflow/axonflow-sdk-go/v4\"\n\nclient := axonflow.NewClient(axonflow.AxonFlowConfig{\n    Endpoint:     \"http://localhost:8080\",\n    ClientID:     \"my-app\",\n    ClientSecret: \"my-secret\",\n})\n\nresponse, err := client.ProxyLLMCall(\n    \"user-123\",                          // userToken\n    \"Analyze customer sentiment\",        // query\n    \"chat\",                              // requestType\n    map[string]interface{}{},            // context\n)\n```\n\n### Java\n\n```java\nimport com.getaxonflow.sdk.AxonFlowClient;\nimport com.getaxonflow.sdk.AxonFlowConfig;\nimport com.getaxonflow.sdk.types.*;\n\nAxonFlowClient client = AxonFlowClient.builder()\n    .endpoint(\"http://localhost:8080\")\n    .clientId(\"my-app\")\n    .clientSecret(\"my-secret\")\n    .build();\n\n// Gateway Mode: Pre-check → Your LLM call → Audit\nPolicyApprovalResult approval = client.getPolicyApprovedContext(\n    PolicyApprovalRequest.builder()\n        .query(\"Analyze customer sentiment\")\n        .clientId(\"my-app\")\n        .userToken(\"user-123\")\n        .build());\n\nif (approval.isApproved()) {\n    // Make your LLM call here...\n    client.auditLLMCall(AuditOptions.builder()\n        .contextId(approval.getContextId())\n        .clientId(\"my-app\")\n        .model(\"gpt-4\")\n        .success(true)\n        .build());\n}\n```\n\n\u003e **[SDK Documentation](https://docs.getaxonflow.com/docs/sdk/overview)**\n\n\u003e **Telemetry:** SDKs send anonymous usage data (SDK version, OS, architecture) on initialization. No prompts, payloads, API keys, or tenant identifiers are collected. Opt out: `export DO_NOT_TRACK=1` or `export AXONFLOW_TELEMETRY=off`. See [Telemetry Documentation](https://docs.getaxonflow.com/docs/telemetry) for full details including SDK-level config options.\n\n---\n\n## Examples\n\n| Example | Description |\n|---------|-------------|\n| **[Execution Tracking](examples/execution-tracking/)** | WCP workflows, step ledger, MAP plans, cancellation |\n| **[Support Demo](examples/support-demo/)** | Customer support with PII redaction and RBAC |\n| **[Code Governance](examples/code-governance/)** | Detect and audit LLM-generated code |\n| **[Hello World](examples/hello-world/)** | Minimal SDK example (30 lines) |\n\n\u003e **[Browse all examples](examples/)**\n\n---\n\n## Development\n\n```bash\ndocker compose up -d              # Start services\ndocker compose logs -f            # View logs\ngo test ./platform/... -cover     # Run tests\n```\n\nFor a full development environment with health checks and automatic waits, use:\n```bash\n./scripts/local-dev/start.sh      # Recommended for development\n```\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for the complete development guide.\n\n| Package | Coverage | Threshold |\n|---------|----------|-----------|\n| Orchestrator | 76.8% | 76% |\n| Agent | 76.6% | 76% |\n| Connectors | 77.1% | 76% |\n| Shared Policy | 82.4% | 80% |\n\n---\n\n## Contributing\n\nWe welcome contributions. See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.\n\n- 76% minimum test coverage required\n- Tests must be fast (\u003c5s), deterministic\n- Security-first: validate inputs, no secrets in logs\n\n---\n\n## Links\n\n- **Docs:** https://docs.getaxonflow.com\n- **License:** [BSL 1.1](LICENSE) (converts to Apache 2.0 after 4 years)\n- **Issues:** https://github.com/getaxonflow/axonflow/issues\n- **Enterprise:** [sales@getaxonflow.com](mailto:sales@getaxonflow.com)\n\n---\n\n\u003e **Evaluating AxonFlow in production?** We're opening limited Design Partner slots.\n\u003e\n\u003e Free 30-minute architecture and incident-readiness review, priority issue triage, roadmap input, and early feature access.\n\u003e\n\u003e [Apply here](https://getaxonflow.com/design-partner?utm_source=readme_platform) or email [design-partners@getaxonflow.com](mailto:design-partners@getaxonflow.com).\n\u003e\n\u003e No commitment required. We reply within 48 hours.\n\n\u003e **Questions or feedback?**\n\u003e\n\u003e Comment in [GitHub Discussions](https://github.com/getaxonflow/axonflow/discussions/239) or email [hello@getaxonflow.com](mailto:hello@getaxonflow.com) for private feedback.\n\n### Public Issues (Technical Questions Welcome)\n\nIf you are evaluating AxonFlow and encounter unclear behavior, edge cases, or questions about guarantees such as policy enforcement, audit semantics, or failure modes, opening a GitHub issue or discussion is welcome. This includes situations where you are unsure whether something is expected behavior, a limitation, or a mismatch with your use case.\n\nFor private or sensitive questions, you can also reach us at hello@getaxonflow.com.\n\n### Evaluating AxonFlow or Exploring Internally?\n\nIf you looked at AxonFlow in any capacity — reading code, cloning SDKs, testing locally, or mapping it to an internal use case — we would value your perspective.\n\nThis includes cases where you:\n\n- paused evaluation\n- decided not to proceed\n- are still exploring\n- are borrowing ideas for internal work\n\n[Anonymous evaluation feedback (30 seconds)](https://getaxonflow.com/feedback)\n\nNo attribution. No tracking. No follow-up unless you explicitly opt in.\n\n---\n\n_Quick Start verified locally: Mar 2026_\n","funding_links":[],"categories":[],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgetaxonflow%2Faxonflow","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fgetaxonflow%2Faxonflow","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgetaxonflow%2Faxonflow/lists"}