A persistent knowledge graph for AI coding agents, providing memory and context across sessions.
Kindex is a persistent knowledge graph designed for AI-assisted workflows, including Claude Code, Codex, Gemini CLI, and other MCP-capable agents. It captures not just what happened, but what it means and how it connects, enabling AI agents to start sessions with full context. Key features include a weighted knowledge graph with typed nodes and edges, five context tiers (100-4000 tokens) that adapt to token budgets, operational guardrails, team/org readiness with git-tracked project context, blazingly fast hybrid FTS5 + graph traversal, git integration, cache-optimized LLM prompting, session tags, task management, reminders, watches, and conversation modes. Use cases range from creative writing (ingesting a 162-file novel vault) to multi-repo engineering and research analysis. Kindex is open source under MIT license and part of the Exemplar stack.
Product positioning · Revenue signals · Competitors & sources
Research product positioning, revenue signals, and business sources for kindex.tools.
Sign in to start an analysis. Results depend on available data.
