Contexer is the decision control layer between engineering teams and coding agents. It captures decisions while engineers work, requires human approval before a proposed decision becomes trusted context, distributes approved decisions across Claude Code, Cursor, Codex, and Gemini CLI, detects drift in GitHub pull requests, and enforces approved rules through a local commit-time guard.

Which AI coding assistants does Contexer work with?

Claude Code, Cursor, Codex, Gemini CLI, and any AI coding agent that supports the Model Context Protocol (MCP).

What is the difference between Open Source and Teams?

Open Source is for individual developers: local decision capture, human approval, cross-agent context, file-aware retrieval, version history, and the local guard. Teams adds a shared set of approved decisions, team sync and reconciliation, and drift detection on GitHub pull requests.

How does enforcement work?

The shipped local guard runs at commit time. It warns when a change conflicts with an approved decision, and deterministic regex and secret checks can block the commit locally when explicitly armed. Drift detection on GitHub pull requests is advisory today.

Yes. Contexer Open Source is MIT licensed and runs locally with no account. Teams is free during early access; pricing will be announced once our IP process is settled.

Open source and local first. Shared decision governance when your team is ready. See how it works ›

contexer.ai

Teams Install open source

The engineering control layer between your team and your coding agents.

Keep AI-generated code aligned with your engineering decisions.

Give Claude, Cursor, Codex and Gemini the decisions that matter as they work. Contexer independently captures and checks the resulting changes for drift before they land.

Install open source Explore Teams

MIT licensed · Runs locally, no account required · Optional cloud sync

Compatibility

One decision set, compiled into each agent’s native format.

Claude CodeCLAUDE.md
Cursor.cursor/rules
CodexAGENTS.md
Gemini CLIGEMINI.md

The alignment gap

Giving agents the rules isn't the same as making sure they follow them.

Engineering decisions already live in docs, repositories, conversations, and people's heads. AI agents can read that context, but teams still need to know which decisions apply and whether the resulting changes actually follow them.

Decisions are scattered

Architecture decisions, constraints, and conventions live across docs, repos, sessions, and senior engineers.

Agents can miss context

Relevant decisions may exist, but the agent may not retrieve or apply them when they matter.

Decisions and code drift

Code can stop following a valid decision, or the code can evolve while the documented decision becomes stale.

DecisionCodeCodeDecision

Senior engineers become the enforcement layer

The people who know why things are built a certain way end up catching the same context-specific mistakes during review.

Having the right context is necessary. Verifying that the resulting work stays aligned with it is a separate problem. Contexer closes this loop.

How it fits together

One workflow, five stages. Capture, approve, share, detect, enforce.

Four stages run in the open source tool on your own machine. Teams adds a shared approved decision set and drift detection on GitHub pull requests.

01
Capture
Open source
Decisions are captured from real coding sessions and kept as persistent, file-aware context your agents can retrieve on demand.
Install open source
- →Captures architecture, constraint, convention, and pattern decisions
- →File-aware retrieval and full version history
- →When context is missing, Contexer scans the codebase and fills the gap
02
Approve
Open source
Nothing an agent proposes becomes trusted context until a human reviews it. Approval is the gate between suggestion and record.
- →AI-proposed decisions are parked for review, never auto-trusted
- →Approve, edit, skip, or dismiss before anything replays
- →Approved decisions are versioned, with the reasoning attached
03
Share
OSS locally · Teams shared
Approved decisions are compiled into each agent's native format, so Claude Code, Cursor, Codex, and Gemini CLI work from the same record.
- →One decision set, distributed across all four agents
- →Open source keeps it local to your machine
- →Teams adds a shared approved set with sync and reconciliation
04
Detect
Teams
GitHub pull requests are checked against the team's approved decisions, so architecture drift is visible during review.
Explore Teams
- →Runs on GitHub pull requests against your shared decision set
- →Architecture drift detection, advisory today
- →GitHub only. No other source-control providers
05
Enforce
Open source · local guard
The shipped local guard checks your changes at commit time, before anything reaches a pull request.
Loops back to Capture
- →Warns before commit when a change conflicts with an approved decision
- →Deterministic regex and secret checks can block locally when explicitly armed
- →Findings become new captured decisions, closing the loop

The loop closes here. Stage 05 feeds back into stage 01: every guard or drift finding becomes a new decision to review, so your engineering knowledge keeps evolving instead of going stale.

Workflow preview

Capture, proposed decision, human approval. Right inside your terminal.

These mockups show the intended flow. The actual UI may differ. A decision is captured while you work, an agent proposal waits for review, and only an approved decision becomes trusted context.

01Capture

Live

Captured as global convention

Rule

Release sequence: branch → bump → PR → merge → tag

Loaded into 0 sessions · applies to all future repos

State a rule. Contexer stores it.

Directives you state outright become global conventions, loaded into every future session.

02Propose

Idle

Pending decisionAI proposed

Use RabbitMQ for async jobs, not Kafka.

Rationale: lower ops overhead, fits current scale.

1 approve

2 edit

3 ignore

AI surfaces a decision mid-work.

Architecture and constraint decisions are parked for review, never replayed until you approve.

03Approve

Idle

Approved · v1

Decision

Use RabbitMQ for async jobs.

immutable

replayed across agents

One keystroke. Versioned forever.

Approve, edit, or ignore. Approved decisions are immutable and replayed across every agent.

What Contexer actually captures

Engineering decisions. Not chat history.

Architecture decisions

“We standardized on PostgreSQL.”

Constraints

“Never expose S3 buckets publicly.”

Conventions

“Terraform modules live under infrastructure/modules.”

Patterns

“Every API emits OpenTelemetry traces.”

These are engineering decisions. Not chat history, not transcripts, not session logs.

Not another AI memory tool

AI memory tools recall conversations. Contexer is your team’s decision of record.

CLAUDE.md, AGENTS.md, .cursorrules, and per-assistant memory features are single player. One person edits a file, another tool ignores it, nothing is reviewed, nothing is versioned. Contexer is the shared layer above them: typed engineering decisions, human approved, versioned forever, and replayed into every agent your team uses.

AI memory and rules files

Contexer

What it holds

Chat history, per-assistant rules files

Typed engineering decisions with rationale, the team's decision of record

How it gets in

Auto-scraped from sessions, or hand-edited by one person

AI proposes, a human reviews and approves before it is trusted

Versioning

Overwritten in place, no history

Approved decisions are immutable. Changes create a new revision with full provenance

Scope

Per-assistant, per-machine

One decision layer, shared across the team

Vendor coverage

Locked to one tool at a time

Vendor neutral: Claude Code, Cursor, Codex, and Gemini CLI

What you can point to

“The model saw some notes.”

Who approved which decision, when, and why

Pricing

Start open source. Add governance when the team needs it.

Start here

Open Source

FreeMIT licensed

For individual developers. Runs locally, stores decisions on your machine, and wires into the coding agents you already use. No account required.

Local first · No account required

✓Capture architecture, constraint, convention, and pattern decisions
✓Human approval before an AI-proposed decision becomes trusted
✓Cross-agent context for Claude Code, Cursor, Codex, and Gemini CLI
✓File-aware retrieval and full version history
✓Local guard: warns at commit time when a change conflicts with a decision
✓Deterministic regex and secret checks can block locally when armed

Install open source Optional cloud sync

Teams

Freeduring early access

Shared decision governance for engineering teams. One approved decision set of record, distributed to every member and every agent. Pricing will be announced once our IP process is settled.

No credit card required · Pricing coming soon

✓Everything in Open Source
✓Shared repository of reviewed and approved team decisions
✓Team sync and reconciliation across members
✓Role-based approval workflow (lead or member)
✓Drift detection on GitHub pull requests, advisory today

Explore Teams

Quick start

Two commands. Under two minutes.

Requires Python 3.12+ and uv. Auto-detects Claude Code, Cursor, Codex, and Gemini CLI, then wires them in.

bash · install

# Step 1. install
$ uv tool install contexer

# Step 2. wire into your AI assistant
$ contexer install

-> detected ~/.claude (Claude Code)
-> detected ~/.cursor (Cursor)
-> detected ~/.codex (Codex)
-> detected ~/.gemini (Gemini CLI)
-> wired all four. restart your AI assistant.