Visit website Free to browse · updated 2026-06-19
What is Kindex?
Kindex functions as a memory system that AI coding agents can use to avoid starting sessions without prior knowledge. It builds intelligence through typed nodes, weighted connections, and decaying edges that reflect provenance and importance over time. The platform includes multiple context tiers that select appropriate detail levels automatically, along with git-tracked project configurations for team environments. Additional features cover task management linked to graph elements, time-based reminders, and policy-based security checks that run with minimal overhead. It supports various workflows from creative writing projects to multi-repository engineering by enabling searches across connected concepts and maintaining durable work items outside of individual sessions.
Key features
Weighted knowledge graph with typed nodes/edges and decay
5 adaptive context tiers (100-4000 tokens)
Git integration with team/org inheritance
Hybrid FTS5 + graph traversal search
Task management, reminders, watches and session tags
Conversation modes and cache-optimized LLM prompts
Security via signet-eval policy enforcement
What you can use Kindex for
Creative Writing
Ingest multi-file novel vaults containing characters, locations, magic systems, and plot outlines in a single pass, with cross-referencing by content mentions and millisecond search results.
Multi-Repo Engineering
Link knowledge across repositories using git-tracked project contracts so service repos inherit platform and org context while maintaining privacy scoping.
Task and Session Management
Track durable work items linked to graph concepts, use session tags for seamless resume across sessions, and apply reminders plus watches that surface automatically at session start.
How to use Kindex
- 1Run pip install kindex[mcp] to install the package
- 2Add the MCP server with claude mcp add --scope user --transport stdio kindex
- 3Initialize a project using the kin init command
- 4Use kin status or kin search to interact with the knowledge graph
- 5Configure policy.yaml for security rules and operational guardrails
Kindex pricing
Pricing model: Open Source. Plan details are indicative — check the site for current prices.
MIT License
Free
- Full access to all features
- pip install kindex[mcp]
- GitHub source available
Editor's verdict
Pros
- +Auto-selects context depth to fit token budget
- +Blazingly fast (192 nodes/11k edges in 142ms)
- +Portable git-tracked artifacts for multi-repo use
Cons
- –Requires MCP-capable AI workflows and CLI setup
- –Opt-in work policies and git hooks for enforcement
- –Zero-config but SQLite-based local storage only
Our take: Kindex is a solid coding & dev choice. It's valued for auto-selects context depth to fit token budget and blazingly fast (192 nodes/11k edges in 142ms). The main trade-off is requires mcp-capable ai workflows and cli setup. A good pick if you want capable AI without a high upfront cost.
Frequently asked questions
Kindex is a persistent weighted knowledge graph that serves as the memory layer for MCP-capable AI coding agents, capturing relationships, provenance, and decay instead of flat session logs.
Summary
Kindex is a solid coding & dev choice. It's valued for auto-selects context depth to fit token budget and blazingly fast (192 nodes/11k edges in 142ms). The main trade-off is requires mcp-capable ai workflows and cli setup. A good pick if you want capable AI without a high upfront cost.
Did you find this helpful?
User reviews
Verified reviews from the community shape this tool's rating.
Loading reviews…
Kindex alternatives
Similar coding & dev tools worth comparing.
