infinity-context MCP server
Commit Oct 28, 2026Provides shared project memory for AI agents with source-backed facts, reviewable updates, and verifiable citations.
Also known as 777genius/infinity-context
- Stars
- 8
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- —
- Last commit
- Oct 28, 2026
- License
- Apache-2.0
What is infinity-context MCP?
infinity-context provides shared project memory for AI agents with source-backed facts, reviewable updates, and verifiable citations.
Summary generated from the project's README
About infinity-context MCP server
Infinity Context provides a shared memory layer for AI agents to maintain project knowledge across different sessions and tools. It allows teams to capture decisions and documents in a central location that remains available even when starting new conversations. The system runs on your own infrastructure using Docker, Python 3.11+, and Git, ensuring you maintain control over your data.
Developers can connect various coding agents such as Claude Code, Cursor, Gemini CLI, Codex, and OpenCode to this memory pool. The server uses Postgres to hold the canonical state, with optional Qdrant and Graphiti indexes to assist with finding information. It functions as an MCP server that provides cited evidence rather than direct instructions, allowing users to verify the provenance of any retrieved fact.
Users interact with the system through a local UI to browse knowledge and review suggested updates before they become permanent. The workflow follows a cycle of capture, review, recall, and update to keep project information current. You can install the MCP version via pipx or deploy it using Docker containers for team-wide access.
What you can do with it
- 1Sync knowledge between agentsConnect multiple coding agents to the same memory pool so they share project context.
- 2Verify retrieved informationCheck the original source and provenance of any facts provided by the AI.
- 3Review automated memory updatesInspect and approve suggested knowledge updates before they are saved to the permanent record.
- 4Resume interrupted work sessionsProvide a new AI session with relevant project decisions without re-pasting entire chat histories.
Written from the project's README. Every use case is checked against a quote from it.
Automated checks
Ran Nov 14, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachable777genius/infinity-context is public and reachablegithub.com/777genius/infinity-context ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)Apache-2.0spdx.org/licenses/Apache-2.0.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-10-28github.com/777genius/infinity-context ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeNot checked
README
View on GitHubInfinity Context
Shared project memory for AI agents, with sources you can inspect and changes you can review.
Keep decisions, documents, and project knowledge available across sessions and agents. Infinity Context runs on your infrastructure and gives your team one place to capture knowledge, review updates, and recall relevant context.
Quickstart · Integrations · Documentation
Why use it?
- Continue where work left off. Give a new session the relevant project decisions and evidence without copying an entire conversation.
- Share knowledge across agents. Connect several coding agents to the same memory while keeping projects and threads scoped.
- Review what becomes memory. Automatic capture creates suggestions for review before promotion to durable knowledge.
- Follow the source. Retrieved context includes citations and provenance so people and agents can check the evidence.
- Keep knowledge current. Updates and forget operations affect canonical visibility, even while search indexes catch up.
The everyday loop: capture → review → recall → update. The local UI lets you browse knowledge and review suggestions alongside your agent workflow.
Quickstart
Requires Git, Python 3.11+, and Docker with Compose.
Download the versioned installer, then connect your agent:
curl -fsSLo infinity-context-install.sh \
https://raw.githubusercontent.com/777genius/infinity-context/v0.2.0/scripts/install.sh
bash infinity-context-install.sh --agent codex
The installer starts the local stack, configures the requested agent integration, and opens the memory UI. It reports any integration that still needs manual setup.
Supported coding agents
| Agent | Installer option |
|---|---|
| Codex | --agent codex |
| Claude Code | --agent claude |
| Gemini CLI | --agent gemini |
| OpenCode | --agent opencode |
| Cursor | --agent cursor |
Repeat --agent to connect several agents, or use --all-agents.
Use --retrieve-only to enable recall without automatic capture, or
--no-agent-tools to keep agent configuration unchanged.
See installation and first run for installation options, capture controls, and upgrade guidance.
Upgrading from v0.1.0? Back up the existing Postgres container before recreating it. Follow the data persistence and upgrade notes.
Integrate with your app
Use the HTTP API or an SDK to add memory to an application. Coding agents can use the MCP server; people can use the CLI and local UI.
| Entry point | Start here |
|---|---|
| Python SDK, CLI, and MCP | pipx install 'infinity-context[mcp]==0.2.0' |
| TypeScript SDK | SDK README and release guide |
| HTTP API | API and integration documentation |
| Agent tools and resources | MCP adapter guide |
The service and Python package are versioned together at 0.2.0. The TypeScript SDK has its own release line, 0.3.2; follow its release guide for installation and artifact verification.
For a container deployment:
docker pull ghcr.io/777genius/infinity-context:0.2.0
For development from source:
git clone https://github.com/777genius/infinity-context.git
cd infinity-context
python3 -m venv .venv
.venv/bin/python -m pip install -e '.[mcp]'
.venv/bin/infinity-context quickstart --agent codex --open-ui
How memory stays trustworthy
Postgres holds the canonical memory state. Optional Qdrant and Graphiti indexes help find candidates; each result is checked against current scope, visibility, and lifecycle state before it enters context.
Retrieved memory is rendered as cited evidence, never as instructions for the model. Source references establish provenance; review and judgment are still needed to decide whether a claim is true.
Read the architecture and trust model for consistency guarantees, adapter boundaries, and security limits.
Status
Alpha, under active development. APIs and deployment details may change. Evaluate retrieval quality and operational fit on your own workload before production use. Current diagnostics do not establish broad answer accuracy or large-scale production readiness.
Optional adapters have different capability limits; the Graphiti adapter is a current-state projection. Cognitive memory remains a foundation rather than a complete runtime feature. For shared deployments, configure access control and backups using the deployment guide.
Explore further
infinity-context MCP questions
Which coding agents are supported?
Supported agents include Codex, Claude Code, Gemini CLI, OpenCode, and Cursor.
How can I integrate this into my own application?
You can use the HTTP API, the Python SDK, or the TypeScript SDK to add memory to your application.