Your team uses multiple AI coding tools and autonomous agents. Each has its own understanding of your codebase. Cont3xt is the governance layer that keeps them all aligned.
Works with Cursor, GitHub Copilot, Claude Code, Windsurf, Kiro, and autonomous AI agents.
Free forever for individuals • No credit card
Your team uses Cursor, Claude Code, Copilot, and autonomous agents. Each one has its own understanding of your codebase. Nobody has a unified governance strategy.
Your .cursorrules say one thing. Your CLAUDE.md says another. Copilot has no instructions at all. Each AI tool has its own fragmented understanding of your standards.
Teams maintain 3+ separate context configs on averageAutonomous agents can work independently for days. When they propose changes, who reviews them? When they drift from your standards, how do you catch it? Most teams have no answer.
Organisations deploy agents faster than they can govern themDevelopers adopt new AI tools weekly. Each one generates code with zero awareness of your security policies. CISOs are asking how you enforce standards across tools you don't control.
43% of MCP implementations have security vulnerabilitiesYour CTO asks: "What context did the AI agent have when it wrote that production code?" You have no answer. There's no log of what rules were served to which tool, or when.
12x more AI projects reach production with governance toolsGitHub Copilot Spaces only works with Copilot. Kiro steering files only work with Kiro. Every vendor solves context for their tool, not your team. You're left stitching together fragmented solutions.
No single vendor solves cross-tool context governanceYou updated the auth standard last week. Cursor knows. Claude doesn't. The deployment agent definitely doesn't. There's no way to push a standards change to every tool simultaneously.
Average team takes 2+ weeks to propagate a standards changeOne source of truth that every AI tool and autonomous agent connects to. Enforce standards, review agent proposals, and prove compliance — across every tool your team uses.
Define your standards once. Cont3xt enforces them across Cursor, GitHub Copilot, Claude Code, Kiro, autonomous agents, and any MCP-compatible tool. When standards change, every tool knows instantly. When agents propose changes, humans review them.
# One rule. Every tool.
Priority: CRITICAL
Applies to: auth/**, security/**
Rule: Always use bcrypt for password hashing
Never use MD5 or SHA1 for password hashing.
These are cryptographically broken. Always use
bcrypt with a cost factor of at least 12.
# Enforced across:
Cursor: ✓ via MCP
Claude Code: ✓ via MCP
Copilot: ✓ via MCP
deploy-bot: ✓ via Agent API
Served: 1,342 times
Effectiveness: 94% helpful// Same context → every tool
// MCP Response for auth/handlers.go
{
"context": {
"rules": [
{
"priority": "critical",
"content": "Use bcrypt for passwords..."
},
{
"priority": "high",
"content": "All handlers accept context.Context..."
}
],
"adrs": ["ADR-001", "ADR-007"],
"servedTo": "cursor | claude-code | copilot",
"tokenBudget": 3200
}
}# Agent proposes → Human reviews
POST /api/rules (with agent key)
{
"title": "Use structured logging",
"content": "Always use zerolog..."
}
→ 202 Accepted (pending human review)
{
"proposal_id": "p_abc123",
"status": "pending_review",
"reviewer": "sarah.chen",
"message": "Awaiting team approval"
}
# No unvetted agent writes. Ever.
# Full activity log for compliance.# Audit Trail — who got what, when
{
"event": "context_served",
"timestamp": "2026-03-03T14:23:01Z",
"tool": "cursor",
"user": "mike.johnson",
"rules_served": 4,
"file_context": "auth/handlers.go"
}
{
"event": "agent_proposal",
"timestamp": "2026-03-03T14:25:12Z",
"agent": "deploy-bot",
"action": "create_rule",
"status": "approved_by: sarah.chen"
}Not just storage — the control plane for how every AI tool and agent accesses your team's knowledge.
Define rules once, enforce everywhere. Cursor, Copilot, Claude Code, Kiro, and autonomous agents all receive the same standards through a single MCP connection.
Autonomous agents self-register and read your knowledge freely. When they try to write or update rules, it becomes a proposal. A human reviews before anything changes.
Every context served to every tool is logged. Know exactly what rules an agent had when it wrote that production code. Ready for security reviews and compliance audits.
Prove which rules actually work. Track adoption rates per tool, identify stale rules, and measure how governance improves code quality across your team.
Create, organise, and prioritise team conventions with file patterns, tags, and priority levels. Version history tracks every change across your team.
Document why you chose PostgreSQL over MongoDB. Track decision status, link to PRs, and ensure every AI tool understands your architectural constraints.
One server, every tool. Built in Go for speed and reliability. Sub-200ms response times with smart token budget management. Works with any MCP-compatible tool.
Control who can create rules, who can approve agent proposals, and who has read-only access. Admins, editors, and viewers — the same permissions model you already use.
AI-powered relevance scoring ensures only applicable context is served to each tool. Respects token budgets. Priority-based selection across all connected tools.
Automatically extract patterns from PR discussions. Convert code review comments into rules that are immediately enforced across every connected tool and agent.
Your source code never leaves your machine. We only store patterns, rules, and architectural decisions. Encryption at rest and in transit. Zero-knowledge architecture.
Whether you're a solo developer or an engineering team with autonomous agents, Cont3xt adapts to your workflow.
Add your coding standards, architectural decisions, and patterns in plain English. "Always use bcrypt for passwords." "We use Zod for validation." Import from existing .cursorrules, CLAUDE.md, or copilot-instructions files — consolidate into one source of truth.
Install the Cont3xt CLI and run cont3xt setup to automatically configure your API key, sync rules, and install Claude Code hooks. One CLI serves your standards to Claude Code, Cursor, Copilot, Windsurf, Kiro, and any MCP-compatible tool. No more maintaining separate config files per tool.
Track which rules are being used, by which tools, and how effective they are. When standards change, every connected tool knows immediately. Analytics show you what's working and what needs refinement. Full audit trail for compliance.
Autonomous agents call a single endpoint with a team API key to register themselves. They receive a scoped API key with read access and proposal permissions. Works with any agent framework — Claude Code, Kiro, Devin, or your own custom agents.
The agent uses MCP or REST to query your team's rules, ADRs, and patterns. Smart filtering returns only relevant context for the task at hand. The agent operates within your team's standards — not its own defaults.
When an agent wants to create or update a rule, the write is intercepted and becomes a proposal. A human team member reviews, approves, or rejects the change. No unvetted agent writes, ever. Complete audit trail for every agent action.
No credit card required. Cancel anytime. 30-day money-back guarantee.
For individual developers getting started
Forever
No credit card required
For power users managing multiple AI tools
Billed monthly
$100/year if paid annually (save 17%)
Full governance for teams with AI tools and agents
3 user minimum
$200/year per user if paid annually (save 17%)
For organisations with advanced governance needs
For teams of 50+ developers
All plans include 30-day money-back guarantee. Annual billing available with 20% discount.
Everything you need to know about governing AI tools and agents with Cont3xt.
Cont3xt works with Cursor, GitHub Copilot, Claude Code, Windsurf, Kiro, and any AI tool that supports the Model Context Protocol (MCP). That covers 95%+ of AI coding assistants and autonomous agents. One MCP connection governs all of them — same rules, same enforcement.
Those are tool-native solutions that only work within their own ecosystem. .cursorrules only works in Cursor. Copilot Spaces only works with Copilot. CLAUDE.md only works with Claude. If your team uses 3+ AI tools, you're maintaining separate configs that drift apart. Cont3xt is the single source of truth that governs all of them:
• One set of rules enforced across every tool
• Instant propagation when standards change
• Analytics showing effectiveness per tool
• Audit trail for compliance
• Agent proposal workflows with human review
You can import existing .cursorrules and CLAUDE.md files directly to get started.
Agents call POST /api/agents/register with your team's API key to self-register. They receive a scoped API key with read access to rules, ADRs, and prompts. When an agent tries to write — creating or updating knowledge — the change is intercepted and becomes a proposal that a human team member reviews. No unvetted agent writes, ever. See our API documentation for the full integration guide.
On the Teams plan, every context served to every tool is logged with timestamps, user identity, tool type, and rules served. You can answer "what context did the AI agent have when it wrote that code?" at any time. Logs are exportable as CSV or JSON, and accessible via API. Enterprise plans add advanced compliance reporting and retention policies.
Yes. Most users start on the Free tier to enforce rules across their own AI tools. When your team is ready:
1. Upgrade to Teams ($20/user/month, 3 user minimum)
2. Invite teammates via email
3. They get instant access to shared rules and standards
4. Agent governance features (proposals, audit trail) activate automatically
Your existing rules, history, and integrations all migrate. Takes 2 minutes.
Never. Cont3xt only stores the patterns and standards you explicitly define — things like "use bcrypt for passwords" or "we use PostgreSQL." Your code stays on your machine. Context flows to your AI tools via MCP, but your codebase never leaves your environment. Zero-knowledge architecture.
No. Cont3xt adds 50-200ms latency, which is imperceptible in normal use. Smart context filtering sends only relevant rules for the file you're working on, keeping token usage under 20% of the AI's context window. 80% remains for code generation.
Yes. We use:
• Encryption at rest and in transit (TLS 1.3)
• SOC 2 compliance (in progress)
• No code storage — only patterns you define
• Regular security audits
• Data isolation per user/team
• Agent API keys are scoped and revocable
For Enterprise customers, we support SSO, advanced audit logs, and can discuss on-premises deployment.
Your team uses multiple AI coding tools and autonomous agents. Cont3xt is the governance layer that keeps them all aligned — with human review for every agent-proposed change.
Free forever for individual developers. No credit card required. Start in 5 minutes.