lore
Skill0 starsChanged 34 days agoMITView on GitHub
The user's memory across projects and hosts: how they want coding agents to work, hard rules, and gotchas per stack. Its index is already in your global instructions. Use this skill when the user says remember, always, never, or drop that; when they correct you in a way that should hold in other projects; when the index says a memory sync is due; when you need a category file the index points to; or when the user asks to sync memory, learn from taste, or see what lore knows. What is true in one project only belongs in that project's own memory (KIS), never here.
13 other projects use a version of this.
Adopted by
Most recently taken up by
- thesatellite-ai/loretheir own version, first seen Nov 11, 2026
- DevNewbie1826/skillstheir own version, first seen Nov 8, 2026
- opum-ai/proof-skillstheir own version, first seen Nov 8, 2026
- ryanmaclean/skillstheir own version, first seen Nov 1, 2026
- simota/agent-skillstheir own version, first seen Oct 23, 2026
- 0xHoneyJar/construct-gygaxtheir own version, first seen Oct 11, 2026
Other versions13 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 13
- carapace-sh/carapace-bin★ 2ksame name
- specstoryai/getspecstory★ 1.4ksame name
- simota/agent-skills★ 99same name
- opum-ai/proof-skills★ 27same name
- TheaDust/lore★ 23same name
- sh008/claude-lore★ 6same name
- DevNewbie1826/skills★ 2same name
- prantikmedhi/lore★ 2same name
- 0xHoneyJar/construct-gygax★ 1same name
- karloows/orchraft★ 1same name
3 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
Files
lore
One memory for every host (Claude Code, Codex, Command Code, agy) and every project. It learns on its own: hooks hand finished sessions to a background harvester, a cheap model turns your corrections into candidate rules, and a candidate becomes a rule once it is seen on two different days or the user states it outright. A stronger model reviews new memory once a day. You rarely run anything yourself.
L means python3 ~/.agents/skills/lore/scripts/lore.py. Never edit the memory files by hand;
they are rewritten by L.
What you already have
The index (~/.agents/memory/INDEX.md) is in your global instructions: the strongest rules
inline, then one "read when" line per category file. Follow the rules. Read a category file
(~/.agents/memory/<category>.md) only when its read-when line matches the task. Do not
summarise memory back to the user.
When the user speaks to memory
| The user says | You run |
| "remember: X", or "always / never X" about how to work | L add --cat <category> --text "<X as one imperative rule>" --src "<their words>" --explicit |
| "drop that" about a rule shown under New | L drop <id shown in brackets> |
| "what does lore know about X" | L list --cat <category> (or --candidates) |
| "pin that" | L pin <id> |
A rule is one imperative sentence that holds in any project. A fact about this project only (its stack, paths, deploy steps) goes to the project's KIS instead; say so. Never store secrets, credentials or personal data.
Commands
/lore:sync(commands/sync.md): run it when the index says a sync is due./lore:learn-taste(commands/learn-taste.md): harvest Command Code's taste files./lore:retro(commands/retro.md): record this session's lessons by hand, when the user asks.
Hosts without slash commands: "run the lore sync step" and follow the command file.
Checks and repairs
L deliver --check (is every host's copy of the index current), L hooks check (are the
capture hooks installed), L index (index size against its budget), L review --force (run the
daily review now). Logs: ~/.agents/memory/harvest.log, review.log, notices (project facts
that belong in KIS).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

