Stele- Agentic Ledger
Stele is a shared memory service for AI coding agents, providing a hosted graph that agents access via the Model Context Protocol (MCP). It stores project decisions, tasks, and lessons so that context persists across sessions and tools. With Stele, agents like Claude Code, Cursor, and Codex read from the same knowledge graph before acting, preventing repeated mistakes and maintaining continuity. Setup is quick: users install the stele CLI with a single command, sign in to create an account, and run /stele:start inside their agent. Stele then automatically wires into the project, builds an initial record, and from that point on, the agent pulls the record at the start of each session and files new information back into it. No dashboard configuration or workflow changes are required. Stele also includes a task queue where every task sits in the project's own queue and can be claimed by any agent or team member, enabling multiple sessions to run concurrently without stepping on each other. The service is free on the Free plan with no credit card required and works on macOS, Linux, and Windows.
- Category
- AI Agents
- Pricing
- Free
Problem solved
AI coding agents often lose context when switching tools or starting new sessions, leading to repeated mistakes and duplicated work. Stele provides a persistent, shared memory that agents read and write, ensuring that project decisions, prior attempts, and lessons learned are always available, reducing rework and improving consistency across different tools and collaborators.
Use cases
- Maintain project context when switching between Claude Code, Cursor, and Codex
- Prevent repeated mistakes by surfacing past decisions and reverted changes
- Coordinate tasks across multiple concurrent agent sessions
- Onboard new team members or agents with existing project knowledge
- Track decisions and rationale in a searchable knowledge graph
- Reduce context loss when handing off work between different tools or teammates
Audience
- Developers
- ML Engineers
- Data Scientists
- Startups
Pros
- Shared memory across multiple AI coding tools
- Prevents repeating known mistakes and regressions
- Quick setup with three commands and no dashboard configuration
- Built-in task queue for coordination between sessions
- Free plan available with no credit card required
Cons
- Requires installing a CLI binary
- Relies on an external hosted service for memory storage
- Only works with MCP-compatible agents
- No visible self-hosted option
- Free plan limitations may apply (details not specified)
Official website: https://stele-ai.dev/
Canonical LambdaBase page: https://www.lambdabase.com/tools/stele-agentic-ledger