ai-memory
NexusAi Summary: AI Memory is an open-source, Rust-based memory layer for coding agents. It captures sanitized lifecycle events, consolidates them into a git-versioned Markdown wiki, and hands context between tools so developers resume work seamlessly across Claude, Codex, Cursor, Kiro, and more.
NexusAi Overview
Install lightweight hooks for your preferred agent. AI Memory records prompts, tool lifecycle events, and session boundaries, then consolidates them into concise briefs and durable notes. The next agent starts with a typed handoff, while the wiki remains searchable and versioned alongside your codebase.
How It Works
AI Memory suits engineers who bounce between coding agents, maintain complex projects, or collaborate across machines. Ideal for teams standardizing decisions and procedures, boutique consultancies juggling many clients, or solo developers needing reliable recall after context switches. It benefits CLI-first users, VS Code/Cursor practitioners, and research engineers who want auditable reasoning without trading away speed, privacy, or tool choice.
- Capture sanitized lifecycle events from popular coding agents with minimal overhead.
- Consolidate sessions into coherent, git-versioned Markdown briefs and durable notes.
- Hand off open questions and next steps to the next agent.
- Search memory with FTS5, entity matches, graph neighbors, and optional embeddings.
- Resume cross-harness workstreams while enforcing capture exclusions and repository policies.
Highlights
Who Uses It
Install via Docker images (linux/amd64, arm64) or native binaries for macOS, Linux, and Windows/WSL2. Start the server, set a bearer token, and run ai-memory install-mcp and install-hooks for your agent. Configure providers only if you want LLM consolidation or embeddings. Use ai-memory run to create or resume a managed workstream; direct agent launches remain unchanged. Browse /web for the read-only wiki and call memory_query from your client to retrieve pages. finalize-session safely closes clients lacking a true session-end hook.
NexusAi: Long-term memory that survives tool switches and delivers coherent, auditable context.
Getting Started
AI Memory turns fleeting agent sessions into durable, searchable project knowledge without running a database cluster. It emphasizes safe capture, coherent consolidation, and smooth cross-agent continuity, backed by Git for provenance. The result is reliable context recovery, simpler onboarding, and fewer re-explanations—no vendor lock‑in, minimal ceremony, and pragmatic deployment options.
Open the tool and review its core product experience.
Create your account or access your existing workspace.
Use your own task to judge speed, quality, and fit.
Check similar AI tools before making a final decision.
ai-memory
AkitaOnRails is an experienced engineering creator and community builder known for practical tooling and deep platform knowledge. The project focuses on developer ergonomics, auditable workflows, and cross-vendor interoperability, emphasizing Rust reliability, simple deployment, and clear documentation for teams adopting agentic coding in real projects.
AI Memory provides long-term, cross-agent project memory for AI coding workflows. It listens through lightweight lifecycle hooks or MCP, stores sanitized observations, and compiles coherent session briefs and durable notes as Markdown in a git repository—no vector database required. When a session ends, it consolidates decisions, open questions, and next steps; the next agent receives a bounded handoff to continue without rehashing context. A built-in MCP server and thin CLI expose tools for querying, writing pages, feedback, and administrative operations. Retrieval blends SQLite FTS5 with entity-aware and graph-neighbor signals, with optional embeddings from OpenAI, Gemini, Voyage, or local-compatible endpoints. Managed workstreams resume native sessions across clients like Claude Code, Codex, Command Code, Kiro, Cursor, Kimi Code, Antigravity, Grok, and others, while enforcing capture exclusions and per-repo policies. Runs locally via Docker or native binaries on Linux, macOS, and Windows (including WSL2).
Visit ai-memory
Comments (0)
No Comments Found