Acontext
Open-source skill memory layer: distills agent runs into editable Markdown SKILL.md files that agents fetch with tools on the next run; cloud or self-hosted.
Tags
Description
When a task completes or fails, an LLM distillation pass extracts what worked, what failed and user preferences, and a skill agent writes them into skill files following a schema you define in SKILL.md. Recall is by tool use (get_skill, get_skill_file), not semantic top-k; skills export as ZIP for use with any agent or framework. SDKs: pip install acontext, npm i @acontext/acontext. Hosted at acontext.io (free credits to start) or self-host with Docker and an OpenAI API key. From the makers of Memobase.
Agent access
Related
- memU: Agent-driven personal memory stored as a shared LLM wiki across sessions, agents and devices; distills reusable skills from Codex, Claude Code, Cursor and OpenClaw history.
- Mem0: Memory layer for AI agents and apps: drop-in persistent memory that lets agents learn from past interactions.
- Letta: Platform for stateful agents with memory, identity and the ability to learn over time (formerly MemGPT); terminal UI, App Server, desktop app and Agent SDK.
- Agent Skills Specification: Open format for packaging agent capabilities as folders with a SKILL.md (YAML frontmatter + instructions) plus optional scripts and resources.
- Honcho: Open-source memory infrastructure for stateful agents that models people, agents, groups and projects over time; managed or self-hosted.
Sources
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)