memory-garden
Skill0 starsChanged 9 months agoMITView on GitHub
Write structured memory entries following the 4-Tier Context Iceberg for efficient context management. Use when recording daily learnings, documenting decisions, or extracting seeds. Trigger phrases: "write a memory entry", "record this learning", "save this to memory", "create a context artifact", "document this insight".
In this repository47 other files it gives its agents.
- agent-teachingSkill
- cf-pages-trespies-deploySkill
- compression-ritualSkill
- context-ingestionSkill
- decision-propagationSkill
- documentation-auditSkill
- era-architectureSkill
- file-managementSkill
- frontend-from-backendSkill
- handoff-protocolSkill
- health-auditSkill
- house-poc-builderSkill
- implementation-promptSkill
- iterative-scoutingSkill
- mcp-cloudflare-builderSkill
- mcp-server-builderSkill
- multi-surface-strategySkill
- parallel-tracksSkill
- patient-learning-protocolSkill
- planning-with-filesSkill
- poc-tool-forgeSkill
- pre-implementation-checklistSkill
- process-extractionSkill
- product-positioningSkill
- project-explorationSkill
- release-specificationSkill
- repo-context-syncSkill
- repo-statusSkill
- research-modesSkill
- research-synthesisSkill
- retrospectiveSkill
- seed-extractionSkill
- seed-librarySkill
- seed-to-skill-converterSkill
- semantic-clustersSkill
- skill-audit-upgradeSkill
- skill-creationSkill
- skill-maintenanceSkill
- spec-constellation-to-prompt-suiteSkill
- static-tool-verifySkill
- status-templateSkill
- status-writingSkill
- strategic-scoutSkill
- strategic-to-tactical-workflowSkill
- web-research-externalSkill
- workflow-routerSkill
- workspace-navigationSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
13 rules. About 0 are common, 13 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈13
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 16
Purpose: Write structured, semantically rich memory entries for efficient context management
- Distinctive: line 308
Rule: If it wouldn't matter in 3 months → compress or discard.
- Distinctive: line 364
Add new seeds, decisions, patterns
- Distinctive: line 365
Update existing entries if needed
- Distinctive: line 374
Delete raw daily notes older than 7 days (after compression)
- Distinctive: line 375
Keep only what matters
- Distinctive: line 381
Before finalizing a memory entry, verify:
- Distinctive: line 435
Read this skill: before writing memory entries
- Distinctive: line 438
Apply the 3-month rule: when compressing
- Distinctive: line 439
Maintain the cycle: every 3-7 days
- Distinctive: line 440
Run the quality checklist: before finalizing
- Distinctive: line 448
Maintenance: Update when new memory patterns emerge
- Distinctive: line 464
When running inside the Dojo Genesis plugin:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
OpenClaw Integration: This skill is invoked by the Dojo Genesis plugin via
/dojo run memory-garden. The agent receives project context automatically via thebefore_agent_starthook. Usedojo_get_contextfor full state,dojo_save_artifactto persist outputs, anddojo_update_stateto record phase transitions and decisions.
Memory Garden Writer Skill
Version: 1.0
Created: 2026-02-02
Author: Manus
Purpose: Write structured, semantically rich memory entries for efficient context management
Overview
This skill encodes the pattern for writing memory entries that follow the 4-Tier Context Iceberg and Hierarchical Memory principles from Dojo Genesis v0.0.19. Use this skill to create memory entries that are easy to search, retrieve, and compress.
Philosophy: Memory should be a garden, not a landfill. Cultivate what matters, compost what doesn't.
When to Use This Skill
- Creating daily memory notes after a session
- Writing compressed summaries of conversations
- Extracting "seeds" (reusable insights) from experiences
- Documenting decisions and their rationale
- Maintaining the memory hierarchy (Tier A → Tier B → Tier C)
Memory Hierarchy (From v0.0.19)
Tier A: Raw Daily Notes
- Location:
memory/YYYY-MM-DD.md - Purpose: Capture everything from today's session
- Lifespan: 1-3 days before compression
- Format: Timestamped entries with context
Tier B: Curated Wisdom
- Location:
MEMORY.md(root level) - Purpose: Distilled insights, decisions, patterns
- Lifespan: Permanent, but evolves
- Format: Structured sections with triggers
Tier C: Compressed Archive
- Location:
memory/archive/YYYY-MM.md - Purpose: Historical record, rarely accessed
- Lifespan: Permanent, read-only
- Format: Semantic summaries
Daily Memory Entry Template (Tier A)
# Memory: YYYY-MM-DD
**Session:** [Morning Planning | Deep Work | Creative Session | Review]
**Context:** [What we were working on]
**Duration:** [Approximate time spent]
---
## Key Activities
### [HH:MM] [Activity Name]
**What:** [Brief description of what happened]
**Why:** [The goal or motivation]
**Outcome:** [What was produced or decided]
**Insights:**
- [Specific insight or learning]
- [Pattern or principle discovered]
**Related:**
- Links to: [file, artifact, or previous memory]
- Builds on: [previous work or decision]
---
## Decisions Made
### Decision: [Short title]
**Context:** [What led to this decision]
**Options Considered:**
1. [Option A] - [Pros/Cons]
2. [Option B] - [Pros/Cons]
**Chosen:** [Selected option]
**Rationale:** [Why this was the best choice]
**Trigger:** [When to revisit this decision]
---
## Seeds Extracted
### Seed: [Name]
**Pattern:** [The reusable insight or principle]
**Why It Matters:** [The value or application]
**Trigger:** [When to apply this seed]
- [Context or situation]
- [Keywords or signals]
**Example:** [Concrete example from this session]
**Related Seeds:** [Other seeds this connects to]
---
## Open Questions
- [ ] [Question that needs resolution]
- [ ] [Uncertainty or ambiguity to clarify]
---
## Next Steps
- [ ] [Actionable task]
- [ ] [Follow-up or continuation]
---
## Metadata
**Tags:** #[category] #[topic] #[type]
**Compression Status:** Raw (not yet compressed)
**Importance:** High | Medium | Low
**Retention:** [How long to keep in Tier A before compression]Curated Memory Template (Tier B)
# Memory (Curated Wisdom)
**Last Updated:** YYYY-MM-DD
**Maintenance Cycle:** Every 3-7 days
**Purpose:** Distilled insights, decisions, and patterns that matter beyond a single session
---
## Core Principles
### [Principle Name]
**Statement:** [Clear, concise principle]
**Origin:** [Where this came from - date, context, or experience]
**Application:** [When and how to apply this]
**Examples:**
- [Concrete example 1]
- [Concrete example 2]
**Trigger:** [Keywords or contexts that should surface this principle]
---
## Key Decisions
### [Decision Title]
**Date:** YYYY-MM-DD
**Context:** [What led to this decision]
**Decision:** [What was decided]
**Rationale:** [Why this was chosen]
**Status:** Active | Revisit on [date] | Deprecated
**Trigger:** [When to recall this decision]
---
## Patterns & Insights
### [Pattern Name]
**Observation:** [What we've noticed repeatedly]
**Evidence:**
- [Instance 1: date, context]
- [Instance 2: date, context]
- [Instance 3: date, context]
**Implication:** [What this means for future work]
**Trigger:** [When to apply this pattern]
---
## Seeds (Reusable Knowledge)
### Seed: [Name]
**Pattern:** [The reusable insight]
**Why It Matters:** [The value]
**Trigger:** [When to apply]
**Origin:** [Where this came from]
**Last Used:** YYYY-MM-DD
**Usage Count:** [Number of times applied]
---
## Compression History
| Date | Compressed From | Summary | Retained |
|------|-----------------|---------|----------|
| YYYY-MM-DD | memory/YYYY-MM-DD.md | [Brief summary] | [What was kept] |
---
## Metadata
**Total Seeds:** [Number]
**Total Decisions:** [Number]
**Total Patterns:** [Number]
**Last Maintenance:** YYYY-MM-DD
**Next Maintenance:** YYYY-MM-DDCompressed Archive Template (Tier C)
# Memory Archive: YYYY-MM
**Compressed:** YYYY-MM-DD
**Source:** [List of daily files compressed]
**Compression Ratio:** [X]%
**Method:** Semantic compression (3-month rule)
---
## Summary
[2-3 paragraph summary of the month's activities, focusing on decisions, lessons, and patterns]
---
## Significant Events
### [Event Name]
**Date:** YYYY-MM-DD
**What:** [Brief description]
**Impact:** [Why this mattered]
**Outcome:** [Result or consequence]
---
## Lessons Learned
1. **[Lesson Title]:** [What we learned and why it matters]
2. **[Lesson Title]:** [What we learned and why it matters]
---
## Decisions Made
| Date | Decision | Rationale | Status |
|------|----------|-----------|--------|
| YYYY-MM-DD | [Brief decision] | [Why] | [Active/Deprecated] |
---
## Seeds Extracted
| Seed Name | Pattern | Trigger |
|-----------|---------|---------|
| [Name] | [Brief pattern] | [When to apply] |
---
## Metadata
**Compression Method:** Semantic (3-month rule)
**Original Size:** [X] lines
**Compressed Size:** [Y] lines
**Compression Ratio:** [Z]%
**Retention:** Permanent (read-only)The "3-Month Rule" (From Cipher)
Rule: If it wouldn't matter in 3 months → compress or discard.
Keep:
- Decisions and their rationale
- Lessons learned and patterns discovered
- Seeds (reusable insights)
- Significant events and outcomes
Compress:
- Routine activities ("worked on X")
- Pleasantries and confirmations
- Detailed step-by-step logs (keep summary only)
Discard:
- Duplicate information
- Temporary notes that were resolved
- Irrelevant tangents
Semantic Compression Guidelines
What to Keep (Verbatim)
- Decisions: The choice, rationale, and context
- Insights: Novel patterns or principles
- Seeds: Reusable knowledge with triggers
- Failures: What didn't work and why
- Breakthroughs: Moments of clarity or innovation
What to Summarize
- Activities: "Worked on X, Y, Z" → "Implemented feature X"
- Discussions: Long back-and-forth → Key points and outcome
- Research: Detailed findings → Summary and conclusion
- Iterations: Multiple attempts → Final approach and why
What to Discard
- Pleasantries: "Great work!" "Thank you!" (unless significant)
- Confirmations: "Got it" "Understood" "Proceeding"
- Redundant logs: Repeated information
- Resolved questions: Questions that were answered and no longer relevant
Memory Maintenance Cycle
Every 3-7 Days:
- Review Tier A (Daily Notes):
- Identify seeds to extract
- Identify decisions to document
- Identify patterns to record
- Update Tier B (Curated Wisdom):
- Add new seeds, decisions, patterns
- Update existing entries if needed
- Remove deprecated information
- Compress to Tier C (Archive):
- Apply 3-month rule
- Create semantic summary
- Move to archive folder
- Prune:
- Delete raw daily notes older than 7 days (after compression)
- Keep only what matters
Quality Checklist
Before finalizing a memory entry, verify:
Daily Notes (Tier A)
- Timestamped entries with context
- Clear "What, Why, Outcome" structure
- Insights and learnings captured
- Decisions documented with rationale
- Seeds extracted with triggers
- Tags and metadata included
Curated Memory (Tier B)
- Principles are clear and actionable
- Decisions include context and rationale
- Patterns have evidence (3+ instances)
- Seeds have clear triggers
- Compression history is updated
- Maintenance date is set
Compressed Archive (Tier C)
- Summary captures key events
- Decisions and lessons are preserved
- Seeds are extracted
- Compression ratio is calculated
- Original files are deleted after compression
Examples
From Dojo Genesis:
- Daily notes from v0.0.17-v0.0.23 development
- Curated wisdom from Cipher collaboration
- Compressed archives from backend migration
Study these for:
- How to extract seeds from experiences
- How to document decisions with context
- How to apply the 3-month rule
- How to maintain the memory hierarchy
Common Pitfalls to Avoid
❌ Hoarding Everything: Keeping every detail → ✅ Compress ruthlessly
❌ Vague Insights: "This was useful" → ✅ "This pattern applies when X"
❌ Missing Triggers: Seed without context → ✅ Seed with clear "when to apply"
❌ No Maintenance: Let Tier A grow forever → ✅ Compress every 3-7 days
❌ Duplicate Information: Same thing in multiple places → ✅ Single source of truth
Usage Instructions
- Read this skill before writing memory entries
- Choose the right template (Tier A, B, or C)
- Fill in each section with specific, structured content
- Apply the 3-month rule when compressing
- Maintain the cycle every 3-7 days
- Run the quality checklist before finalizing
Skill Metadata
Token Savings: ~5,000-8,000 tokens per session (structured format enables efficient retrieval)
Quality Impact: Ensures consistent memory format across sessions
Maintenance: Update when new memory patterns emerge
Related Skills:
specification-writer- For documenting technical decisionsseed-extraction- For extracting reusable insightsworkspace-navigation- For managing memory files efficiently
Last Updated: 2026-02-02
Maintained By: Manus
Status: Active
OpenClaw Tool Integration
When running inside the Dojo Genesis plugin:
- Start by calling
dojo_get_contextto retrieve full project state, history, and artifacts - During the skill, follow the workflow steps documented above
- Save outputs using
dojo_save_artifactwith theartifactsoutput directory - Update project state by calling
dojo_update_stateto record skill completion and any phase transitions
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
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.

