{"slug": "show-hn-komi-learn-continuous-memory-and-self-improvement-for-coding-agents", "title": "Show HN: Komi-learn – continuous memory and self-improvement for coding agents", "summary": "Komi-learn, a new tool for coding agents, automatically learns a user's coding style, stack, and fixes from sessions and recalls them in future sessions without commands. The tool works with Claude Code and Codex, distilling durable lessons in the background and loading relevant ones at the start of each session. It includes an optional community pool for sharing general lessons, with contributions scrubbed of identifying information and requiring user approval before sharing.", "body_md": "Continuous memory and self-improvement for coding agents. It learns how you work and recalls it automatically, with no commands. Works with Claude Code and Codex.\n\nIt watches a session, distills durable lessons in the background (your style, your stack, fixes that worked), and loads the relevant ones at the start of the next session. No slash commands, nothing to save by hand.\n\nThe idea is from [Hermes Agent](https://github.com/nousresearch/hermes-agent); this is my own take, generalized across hosts with an optional shared layer (the community pool, below).\n\nIt's early. The core loop is built and CI-tested, but it hasn't been battle-tested across a lot of real sessions yet. Feedback and bug reports are welcome.\n\n```\npip install komi-learn\nkomi-learn install            # or: komi-learn install --host codex\n```\n\n`install`\n\nruns a short interactive setup, then recall and background learning start in your next session. If you already use Claude Code you're already logged in. For scripts, `komi-learn install --yes`\n\ntakes the defaults.\n\nFrom source:\n\n```\ngit clone https://github.com/kurikomi-labs/komi-learn\ncd komi-learn\npip install -e .\nkomi-learn doctor      # check the install and what to fix\nkomi-learn status      # config + how much it has learned\nkomi-learn config      # change any setting (menu, or `config set <key> <val>`)\nkomi-learn sync        # pull the latest community learnings\nkomi-learn queue       # review/approve/reject what you'd contribute to the pool\nkomi-learn forget <x>  # erase learnings matching <x> (archive, or --hard to delete)\nkomi-learn uninstall   # remove the hooks (keeps your data; --purge to wipe)\n```\n\nYou can change anything after install, e.g. `komi-learn config set recall.semantic false`\n\nor leave the pool with `komi-learn config set pool.repo_url \"\"`\n\n.\n\n- Recall: at session start, learnings relevant to the current context are loaded.\n- Distill: after the session, a background pass reads the transcript and extracts durable lessons (corrections, techniques, fixes).\n- Curate: over time it merges overlapping lessons and archives stale ones.\n- Share (optional): general lessons can be contributed to the community pool, but only ones you approve.\n\nIt tries not to learn the wrong things. Secrets, machine-specific paths, one-off failures, and \"tool X is broken\" complaints are filtered out by a deterministic check before the LLM ever sees them.\n\nA public pool of general agent lessons, stored as a GitHub repo of signed Markdown files (no server). If you opt in, you get lessons other people's agents figured out, and you can contribute your own.\n\nContributions are scrubbed of anything identifying and never leave your machine without your approval (each one opens a PR you reviewed). Learnings are content-addressed (BLAKE3) and signed (Ed25519); one signed by more distinct GitHub accounts ranks higher when pulled. That account count is Sybil-resistant but not Sybil-proof, so it's an advisory signal, not a hard trust gate. Recalled community items are labelled and treated as untrusted input. Details: [pool-repo-template/CONTRIBUTING.md](/kurikomi-labs/komi-learn/blob/main/pool-repo-template/CONTRIBUTING.md).\n\nNo setup or API key needed:\n\n```\npython examples/demo_loop.py\n```\n\nIt runs two sessions: you correct the agent in the first, and the second shows it recalling that with nothing typed.\n\n- Python 3.10+\n- Claude Code or Codex (the agent it plugs into)\n- A working model for the distill step: your existing Claude Code login, or\n`komi-learn login`\n\n, or an API key via`--api-key`\n\n.\n\n`komi-learn install`\n\nverifies these with a real model call and stops with fix steps if something's missing. At runtime, if a hook can't reach the model it skips that learning pass rather than interrupting your session.\n\nThe engine has no required dependencies. Optional extras add real signing (`pip install komi-learn[crypto]`\n\n) and local semantic recall (`[smart]`\n\n); without them it falls back to a stdlib hash and keyword search.\n\nTo run your own pool, see [pool-repo-template/](/kurikomi-labs/komi-learn/blob/main/pool-repo-template).\n\nMIT. Issues and PRs welcome.", "url": "https://wpnews.pro/news/show-hn-komi-learn-continuous-memory-and-self-improvement-for-coding-agents", "canonical_source": "https://github.com/kurikomi-labs/komi-learn", "published_at": "2026-05-31 05:11:40+00:00", "updated_at": "2026-05-31 05:44:53.064981+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "machine-learning", "large-language-models", "ai-products"], "entities": ["Komi-learn", "Claude Code", "Codex", "Hermes Agent", "Kurikomi Labs"], "alternates": {"html": "https://wpnews.pro/news/show-hn-komi-learn-continuous-memory-and-self-improvement-for-coding-agents", "markdown": "https://wpnews.pro/news/show-hn-komi-learn-continuous-memory-and-self-improvement-for-coding-agents.md", "text": "https://wpnews.pro/news/show-hn-komi-learn-continuous-memory-and-self-improvement-for-coding-agents.txt", "jsonld": "https://wpnews.pro/news/show-hn-komi-learn-continuous-memory-and-self-improvement-for-coding-agents.jsonld"}}