open-timeline-engine: Shared Memory for AI Agents via MCP

Overview

Open Timeline Engine is a local-first Python context engine for MCP-compatible AI agents, including Claude, Codex, and Cursor. It addresses the repeated work of re-explaining project conventions and past decisions by capturing activity over time, finding patterns, and returning relevant context with citations.

The project also includes an advisor and layered controls for agent actions. It is designed for local operation and can share memory between executors or keep their workspaces separate. The project describes itself as experimental, and its README explicitly says it is not production-ready.

Key Features

• Captures timeline events from agent sessions and additional sources such as Git, VSCode, and browser plugins.
• Provides hybrid search and context bundles across timeline events, patterns, and graph data.
• Connects MCP-compatible clients, including Claude Desktop, Codex, Cursor, and generic MCP clients.
• Offers an advisor mode that uses recorded decisions to guide agent responses, alongside a timeline-only mode.
• Includes policy checks, citations, redaction, audit records, and a layered permit and execution-claim workflow.
• Supports shared or per-executor workspace memory.
• Includes a dashboard for system status and memory-related controls.

Use Cases

• Developers who use multiple coding agents on the same project and want those agents to retrieve shared, cited context.
• Solo developers who want a searchable record of agent activity, decisions, and outcomes.
• Teams or individuals who want to capture Git, editor, or browser activity alongside agent sessions.
• Developers exploring policy-gated agent workflows who are prepared to evaluate an experimental system and its limits.

Getting Started

git clone https://github.com/JOELJOSEPHCHALAKUDY/open-timeline-engine.git
cd open-timeline-engine
./scripts/install.sh

See the README for runtime choices, MCP setup, and configuration details.

Alternatives

• memoripy: Memoripy focuses on screening evidence and tracking changing facts, while Open Timeline Engine mines work history for recurring patterns and decisions.
• ai-memory: ai-memory turns coding-agent activity into a searchable, git-backed Markdown wiki and supports handoffs, rather than mining work patterns.
• Koshi: Koshi emphasizes durable coding-project memory, code retrieval, and team decisions, while Open Timeline Engine focuses on work over time and workflow patterns.
• plur: PLUR stores shared corrections and project conventions in readable YAML, rather than capturing work history to surface patterns and cited decisions.

Considerations

• The repository labels itself experimental and not production-ready; APIs, storage formats, and behavior may change.
• It requires Python 3.12 or later. The repository also declares Node.js in its package metadata. The README recommends the full Docker Compose runtime for reliable behavior and describes the SQLite-based lite runtime as experimental.
• The project has recent commit activity, but no release in over a year and a single contributor accounts for most of the code. Recent issue handling is also limited according to the measured project data.
• The README states that personalization is not currently used for decisions, and the longitudinal pilot has not produced evidence. Treat the behavioral-continuity claims as goals, not established results.
• Some controls are cooperative rather than enforced by the host operating system. The README documents limits in the sandbox and notes that fresh installs are not production-hardened.