memory
Skill131 starsChanged 4 months agoMITView on GitHub
Retrieve and maintain the memory. Use when you need to remember something. personal history, preferences, people, projects, decisions, routines, places, or prior events or when the user asks to remember something.
99 other projects use a version of this.
Adopted by
Most recently taken up by
- jaydubya818/ClaudeClaw-Mission-Controltheir own version, first seen Nov 8, 2026
- naman10parikh/pennytheir own version, first seen Nov 8, 2026
- NaturNestAI-Labs/beings-protocoltheir own version, first seen Nov 8, 2026
- maxtechera/memorytheir own version, first seen Nov 8, 2026
- wubinKKK/private-ai-workbenchtheir own version, first seen Nov 8, 2026
- shetengteng/skillix-hubtheir own version, first seen Nov 8, 2026
Used with
Projects with it also use ecosystem-reference, fractal-delegation and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
naman10parikhSkillecosystem-reference
Quick-access directory of Claude Code ecosystem repos, skills, plugins, MCPs, and patterns. Use when improving the harness.
Also in 5 projects that use this
naman10parikhSkillfractal-delegation
Also in 5 projects that use this
Maliot100XSkillsend-user-message
Send a proactive message to a paired user via their channel (Discord, Telegram, or WhatsApp). Use when the agent needs to notify, alert, or send an unsolicited message to a user — especially during heartbeat invocations, scheduled tasks, or when the agent wants to reach out without a prior user message in the current conversation. Triggers: 'send message to user', 'notify user', 'alert user', 'message the user on discord/telegram/whatsapp', or any need to proactively communicate with a paired sender.
Also in 4 projects that use this
jh941213Skilldocs-ci
문서 CI(링크·OpenAPI lint·breaking 게이트·mermaid 검증)나 문서 드리프트 감지(docs.yaml 신선도 검사)를 설치하거나 돌릴 때 사용. 문서 내용 작성은 docs-architecture·docs-interfaces·docs-manuals.
Also in 3 projects that use this
jh941213Skilldocs-interfaces
API·이벤트 인터페이스 문서(OpenAPI·swagger·AsyncAPI 명세, 엔드포인트 레퍼런스, API 구성도·시퀀스, API CHANGELOG)를 만들거나 코드와 맞출 때 사용. 시스템 구성도는 docs-architecture, API 설계 원칙은 api-design-principles.
Also in 3 projects that use this
- +6 more from naman10parikh
- +1 more from jh941213
Other versions99 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 99
- google-gemini/gemini-cli★ 109ksame name
- jlia0/tinyclaw★ 3.6ksame name
- TinyAGI/tinyagi★ 3.6ksame name
- TinyAGI/tinyclaw★ 3.6ksame name
- Chachamaru127/claude-code-harness★ 3.2ksame name
- codeaholicguy/ai-devkit★ 1.7ksame name
- Muuuun/luxas★ 1.2ksame name
- shobcoder/shob★ 575same name
- boshu2/agentops★ 453same name
- makushingu/harness-maxxing★ 161same name
89 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.
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- browser-targetSkill
- connect-deviceSkill
- contactsSkill
- event-sourcesSkill
- gsv-manualSkill
- image-readingSkill
- process-orchestrationSkill
- skill-authoringSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 5 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 82 edited versions we read keep these.
5
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 19
Do not search memory for self-contained questions whose answer cannot depend on the user.
- Distinctive: line 21
Search narrowly before opening broad pages:
- Very specific: line 27
Use wiki info personal when the page location is unknown.
- Distinctive: line 31
Write immediately when the user explicitly says to remember an unambiguous fact or corrects an existing fact.
- Distinctive: line 63
Search narrowly by likely Process or date when possible.
- Very specific: line 67
Search before adding duplicate information.
- Distinctive: line 77
When a name is ambiguous, make the short-name page a disambiguation page that points to specific pages.
- Very specific: line 79
Use man wiki for exact wiki syntax and general wiki workflows.
“Common” is measured: kept by at least 6 in 10 of the 82 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Manage Personal Memory
Memory belongs to the human, not to an individual agent. Every agent working for the same user reads and writes the same two layers:
- The
personalwiki contains durable information that is searched and loaded when relevant. - The owner's
context.d/10-personal.mdcontains a very small set of stable facts and preferences that should affect nearly every interaction. It appears under the editable<user>context root in the prompt; do not confuse it with the current agent's~/context.d/.
The Kernel responsibility ledger is working state, not personal memory. Do not copy unresolved responsibilities into the wiki merely because they exist. Record an outcome later only when it is useful history.
Decide When to Read
Retrieve memory before asking, recommending, or acting when the correct interpretation or outcome could depend on personal history that is not already in the current context. Examples include an ambiguous person or project name, the user's usual grocery order, travel details, prior decisions, recurring preferences, and where the user normally keeps something.
Do not search memory for self-contained questions whose answer cannot depend on the user. Do not search merely to prove that memory exists.
Search narrowly before opening broad pages:
wiki search <query> --prefix personalUse wiki info personal when the page location is unknown. Once a relevant page is known, use normal filesystem tools to read its Markdown under the path reported by wiki info.
Decide When to Write
Write immediately when the user explicitly says to remember an unambiguous fact or corrects an existing fact. Also write a concrete, stable fact or preference that the user states explicitly when it will improve future help.
When work reveals durable context about a person or relationship, preserve it promptly even if the user did not explicitly ask you to remember it. This includes identity, alternate names or handles, how the person relates to the user, ongoing shared projects, and meaningful interaction outcomes. Do not wait for compaction or a later request to make the information durable. Record inferred context as an inference rather than omitting it, and include a date or source when it will help resolve future ambiguity.
Search and merge before writing when the fact may already exist, refers to an ambiguous person or project, supersedes older information, or belongs on more than one existing page. Ship can perform short lookups and edits directly; delegate when the work benefits from parallel execution, separate context, or lengthy investigation.
Append a journal entry for a meaningful event or outcome whose chronology may matter later. Use ISO dates:
pages/journal/YYYY/MM/YYYY-MM-DD.mdRecover Missing Memory from Process History
Process history is lossless recovery evidence, not the primary memory system. Search the Personal wiki first. If relevant memory is missing and earlier Process work likely contains the answer, inspect that history and then repair the wiki with the durable facts you recover.
For compacted history belonging to a visible Process, prefer the structured Process interface:
proc segments --pid <pid>
proc segment <segment-id> --pid <pid> --jsonReset and terminated Process transcripts are stored as gzipped JSONL under the run-as agent's home. Locate candidate archives and stream them through zcat; do not try to read compressed bytes as text or modify the immutable archive:
find "$HOME/processes" -name '*.jsonl.gz'
zcat "$HOME/processes/<pid>/history/<archive>.jsonl.gz" \
| jq -r '[.ts, .role, .content] | @tsv' \
| rg -i '<term>'Search narrowly by likely Process or date when possible. Once the missing context is recovered, merge the useful facts into the appropriate Personal wiki page. Do not copy raw transcripts, reasoning, or tool noise into memory.
Organize the Wiki
Search before adding duplicate information. Read a page before editing it. Keep index.md as an orientation page and use inbox/ only for information that genuinely cannot yet be placed. Promote durable information into topical pages such as:
pages/people/pages/projects/pages/preferences/pages/decisions/pages/routines/pages/places/pages/concepts/
When a name is ambiguous, make the short-name page a disambiguation page that points to specific pages. Replace superseded facts rather than accumulating contradictions. Prefer concise facts and useful context over prose about the act of remembering.
Use man wiki for exact wiki syntax and general wiki workflows.
Maintain Standing Memory
Edit the owner's context.d/10-personal.md only for explicit, stable facts or preferences that should be present in almost every interaction. Keep the file small, preserve its purpose statement, and replace corrected facts.
Unresolved work belongs in the r12y responsibility ledger. Do not duplicate it in standing context or the wiki.
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.

