agentcairn
ccf/agentcairn
Cross-project Markdown memory for coding agents, with an Obsidian vault and a DuckDB index
agentcairn is a long-term memory tool for AI coding agents that stores durable context as Markdown notes in an Obsidian-compatible vault. It recalls the relevant pieces across projects and across supported agents.
What agentcairn does
agentcairn captures durable context from the coding agents you use, stores it as inspectable Markdown with provenance, and recalls only the most relevant pieces when another agent needs them. The Markdown vault is the canonical memory. DuckDB is a derived retrieval cache that can be deleted or rebuilt without losing the vault. Supported hosts share the same configured vault, so memory is not split per tool.
History is non-lossy: superseded and expired facts stay inspectable and are demoted rather than hidden, and a fact edited by hand is honored on the next reconciled read. The plugin bundles an MCP server, a memory skill and host-specific hooks, and launches through uvx. The README documents Claude Code and Codex installs, with Claude Code getting commands such as /agentcairn:recall and /agentcairn:remember. The default vault is ~/agentcairn, a separate Obsidian companion exposes provenance and links, and the README says it sends no telemetry.
Key features
- Markdown vault as canonical memory, with notes, frontmatter and wikilinks
- DuckDB retrieval cache that can be rebuilt without touching the vault
- One vault shared across supported coding agents
- Non-lossy history: superseded and expired facts are demoted, not deleted
- Bundled MCP server, memory skill and ambient hooks in a plugin
- Claude Code commands for recall, remember, memory, savings and ingest
When to use it
- Carrying decisions and project context from one coding agent session to the next
- Sharing one memory between Claude Code and Codex
- Reading or hand-editing agent memory in Obsidian
Who it is for: Developers who use more than one coding agent and want a local, human-readable memory that they own.
How it fits with Hermes Agent
Supports Hermes among other agents: the repository is tagged hermes, hermes-agent and hermes-plugin on GitHub, while the opening part of the README documents Claude Code and Codex installs.
How to install agentcairn
These commands are copied from the project's README. Check the repository for the latest steps before you run them.
npx skills add ccf/agentcairn --skill agentcairn-setup -gRequirements: uv, because the plugin launches through uvx
FAQ
What is agentcairn?
agentcairn is a long-term, cross-project memory for AI coding agents. It keeps memory as Markdown notes in a vault and uses DuckDB as a replaceable retrieval cache.
Does agentcairn work with Hermes Agent?
It is tagged hermes-agent and hermes-plugin on GitHub, but the opening part of the README documents Claude Code and Codex installs. Check the repository for the current Hermes setup steps.
Is agentcairn free and open source?
Yes. agentcairn is open source under the Apache-2.0 license.
Similar memory for Hermes Agent
All memoryLocal, deterministic memory for coding agents, shipped as one dependency-free Python file
xraysight Hermes Memory UIRead-only Dashboard and Desktop plugin for inspecting Hermes Agent memory and external memory providers
howardpen9 hermes-gbrain-bridgeConverts Hermes, Claude Code, Codex and OpenClaw agent memory into markdown for gBrain
Danialsamadi SynapseLocal-first typed long-term memory for AI agents on SQLite, with an MCP server, CLI and HTTP API
lovevin1314-tech NexSandglassLocal-only memory plugin for Hermes Agent with drift tracking, four-way search and no external dependencies
ZenSystemAI ZengramShared memory MCP server that lets agents on different machines store and recall facts and decisions
Related guides: SOUL.md for Hermes Agent: what it is and how to write one · Run multiple Hermes agents with profiles