memory-init
Skill38 starsChanged 3 months agoMITView on GitHub
Scaffold a project-local memory directory and AGENTS.md guidance for reusable agent knowledge. Use when a user wants to initialize `.agents/memories/` or another project memory path so future agents can read and maintain durable project context.
8 other projects use a version of this.
Adopted by
Most recently taken up by
- dmzoneill/skillstheir own version, first seen Oct 24, 2026
- osipovn/agents-skillstheir own version, first seen Oct 17, 2026
- Rishabh-Choudhry/anchor-memorytheir own version, first seen Oct 15, 2026
- haraldalder-vibemogger/project-memorytheir own version, first seen Sep 26, 2026
- yunshu0909/yunshu_skillshubtheir own version, first seen Aug 10, 2026
- clearstackhq/clearstack-litetheir own version, first seen Aug 10, 2026
Other versions8 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 8
- yunshu0909/yunshu_skillshub★ 760same name
- d2a8k3u/claude-code-memory★ 8same name
- osipovn/agents-skills★ 4same name
- dmzoneill/skills★ 2same name
- haraldalder-vibemogger/project-memory★ 2same name
- spudy-vibing/Claude-Code-Prompts★ 2same name
- clearstackhq/clearstack-lite★ 1same name
- Rishabh-Choudhry/anchor-memory★ 0same name
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 repository31 other files it gives its agents.
- AGENTS.mdAGENTS.md
- asyncSkill
- code-reviewSkill
- create-planSkill
- design-workflow-graphsSkill
- eli5Skill
- engineering-backend-architectSkill
- expertsSkill
- getnoteSkill
- github-create-prSkill
- github-release-notes-writerSkill
- github-release-pr-writerSkill
- github-renovate-prsSkill
- gitlab-create-mrSkill
- gitlab-mr-inline-commentSkill
- glab-cliSkill
- google-fonts-curatorSkill
- jevSkill
- mac-keychainSkill
- multi-codex-orchestratorSkill
- opentelemetry-go-reviewerSkill
- otel-changelog-pr-verifierSkill
- otel-go-reviewerSkill
- programmer-motivatorSkill
- short-fiction-scaffolderSkill
- skill-converterSkill
- skill-metadata-maintainerSkill
- studySkill
- subagent-orchestratorSkill
- volcengine-ark-image-generatorSkill
- yuqueSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 3 are distinctive, and 5 are very specific to this project.
0
Common
Most of the 7 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 15
Use this skill to initialize durable project knowledge, not to store one-off task notes.
- Very specific: line 21
Use .agents/memories/ unless the user specifies another memory directory.
- Very specific: line 66
read index.md before non-trivial work.
- Distinctive: line 68
update memories when the user requests memory maintenance or when the current task explicitly includes maintaining reusable project …
- Distinctive: line 70
avoid one-off task notes, temporary debugging details, secrets, credentials, and conversation-specific context.
- Very specific: line 71
keep index.md aligned when memory files are added, removed, renamed, or materially changed.
- Very specific: line 73
If an existing managed block is present, update it instead of appending a duplicate.
- Distinctive: line 77
Treat memory as durable project knowledge.
“Common” is measured: kept by at least 6 in 10 of the 7 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
Memory Init
Scaffold lightweight, project-local memory files that future agents can read and maintain.
Use this skill to initialize durable project knowledge, not to store one-off task notes. The default target is .agents/memories/, with AGENTS.md guidance for Codex-oriented future use.
Workflow
- Confirm the target project root.
- Inspect whether the target memory directory and agent instruction file already exist.
- Use
.agents/memories/unless the user specifies another memory directory. - Run
scripts/init_project_memory.py. - Review the generated or changed files.
- Report the memory directory and whether the agent instruction file was created or updated.
Command
From any location:
python3 <skill-dir>/scripts/init_project_memory.py /path/to/projectWith a custom memory directory:
python3 <skill-dir>/scripts/init_project_memory.py /path/to/project --memory-dir docs/agent-memoryTo create memory files without touching the agent instruction file:
python3 <skill-dir>/scripts/init_project_memory.py /path/to/project --skip-agentWith an alternate agent instruction file:
python3 <skill-dir>/scripts/init_project_memory.py /path/to/project --agent-file CLAUDE.mdGenerated Files
The scaffold creates missing files only. Existing memory files are preserved.
README.md: explains the memory system, what belongs there, and what must not be stored.index.md: routes future agents to the relevant memory files.project.md: stores stable project conventions and durable context.workflows.md: stores repeatable commands, release steps, CI workflows, and maintenance procedures.lessons.md: stores reusable lessons, pitfalls, and verified fixes.
AGENTS.md Guidance
By default, the script creates or updates AGENTS.md in the project root. It inserts a concise ## Project Memory section that tells future agents to:
- read
index.mdbefore non-trivial work; - open only task-relevant memory files;
- update memories when the user requests memory maintenance or when the current task explicitly includes maintaining reusable project knowledge;
- recommend a memory update instead of editing memory files when memory maintenance is outside the current task scope;
- avoid one-off task notes, temporary debugging details, secrets, credentials, and conversation-specific context;
- keep
index.mdaligned when memory files are added, removed, renamed, or materially changed.
If an existing managed block is present, update it instead of appending a duplicate. If only an unmarked ## Project Memory / ## Project Memories section exists, preserve the user-authored section and append a managed block.
Memory Standard
Treat memory as durable project knowledge. Add or update entries when the user asks for memory maintenance or the task explicitly includes maintaining reusable project context.
Good memory candidates:
- stable repository conventions;
- commands or workflows that were verified locally;
- recurring environment constraints;
- cross-task architectural or maintenance decisions;
- reusable debugging lessons.
Do not store:
- secrets, credentials, tokens, or private keys;
- raw conversation logs;
- single-task progress notes;
- speculative conclusions that have not been verified;
- noisy command output that is not reusable.
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.

