mrvladd-d/memobank/mb-garden
Claude Code and agent skill by mrvladd-d/memobank
- community
Checking that the SKILL.md file is still in the repository.
Clean up and maintain Memory Bank docs so they stay accurate and easy to load.
Install
npx vskill@latest install mrvladd-d/memobank/mb-gardenSKILL.md
mb-garden — Memory Bank maintenance (doc gardening)
- What it does: maintains Memory Bank docs, fixes hygiene issues, and keeps navigation coherent.
- Use it when:
.memory-bank/exists and you want to reduce drift, stale docs, or broken routing. - Input: an existing
.memory-bank/. - Output: a cleaner Memory Bank, resolved hygiene issues, and optional CI support for ongoing maintenance.
Goal
Keep .memory-bank/ accurate, navigable, and cheap-to-prime. mb-garden owns maintenance and cleanup guidance. The current package also ships deterministic tool assets, but mb-lint and mb-doctor remain separate gates with separate roles.
Preconditions
.memory-bank/exists.
Process
1) Install deterministic maintenance tools
If the repo doesn’t have the tools yet: 1) Create scripts/mb-lint.mjs using assets/mb-lint.mjs. 2) Create scripts/mb-doctor.mjs using assets/mb-doctor.mjs. 3) Make them executable (optional). 4) Add package scripts (optional): "mb:lint": "node scripts/mb-lint.mjs" and "mb:doctor": "node scripts/mb-doctor.mjs".
Current package location includes both tool assets under mb-garden/assets; this is a packaging detail, not conceptual ownership of the doctor readiness role.
Then run:
Fix all ERROR findings:
- missing YAML frontmatter (description + status)
- broken links
- missing folder index.md
- invalid or missing schema-backed task records, including missing/invalid mandatory
tier - obsolete
.memory-bank/tasks/backlog.mdor markdown task cards used as workflow state - old
risk/risk.leveltask fields; task routing is only bytask.tier
mb-lint is the structural/mechanical hygiene gate. mb-doctor is the workflow/autonomous readiness gate over mb-lint. Default mb-doctor is suitable for normal health checks and fresh skeletons. Run strict mode only after the JSON task queue exists: after /prd-to-tasks, before scheduler execution inside /autonomous, or before /autopilot when the queue is already prepared:
Do not treat strict mode as required for a bare generated skeleton with an empty task registry.
2) Refactor for ergonomics
- Split docs > ~700 lines into atomic docs + router index.
- Ensure key concepts are covered either by classic duo docs (
architectureWHAT/WHY +guidesHOW) or by equivalent spec-driven support docs. - Replace copy-pasted code/config with links to source.
3) Archive stale or superseded docs
- Move outdated docs to
.memory-bank/archive/. - Leave a short tombstone stub in the original location with a link to the archive.
4) Optional: add CI gate
If the repo uses GitHub Actions:
- Create
.github/workflows/memory-bank-lint.ymlfromassets/memory-bank-lint.yml.
5) Produce a short report
Write a summary (what changed + what remains) to:
.tasks/TASK-MB-GARDEN/TASK-MB-GARDEN-S-01-final-report-docs-01.md
6) Weekly maintenance checklist
Run this checklist weekly (or every 5–10 meaningful changes):
- Frontmatter audit: every
.memory-bank/**/*.mdhasdescription+status. - Stale docs: scan for docs not updated in >2 weeks — archive or refresh.
- Concept coverage check: every key concept is covered either by a matching
architecture/<X>.md+guides/<X>.mdpair or by clearly routed equivalent spec-driven support docs. - RTM sync:
requirements.mdRTM matches actual feature/test status. - Task record hygiene: completed tasks marked done in
.task.json, every task hastier, and no orphaned tasks exist without a feature link. - Changelog:
.memory-bank/changelog.mdhas entries for recent changes. - Index links: all links in
index.mdand router-indexes resolve correctly. - Archive tombstones: every doc in
archive/has a tombstone stub at original location.
Definition of done
node scripts/mb-lint.mjspasses (0 errors).node scripts/mb-doctor.mjsreadiness gate passes in default mode;--strictpasses post-queue before scheduler/autopilot execution.- Index navigation is coherent.
- No obvious duplication or stale docs without archive.
- Weekly checklist items are addressed.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/mb-garden/SKILL.md
Popularity
- Repository stars
- 32
- Repository forks
- 8
- 7-day trend score
- 17
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 17 Mar 2026, 7 months ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/mrvladd-d/memobank/mb-garden)