{"skill":{"slug":"dory-memory","displayName":"Dory-Proof Memory System","summary":"File-based memory system for AI agents that forget between sessions. Implements the \"Dory-Proof\" pattern for continuity across context resets. Use when setting up agent memory, building workspace structure, implementing task tracking, or preventing context-loss errors. Triggers on \"memory system\", \"remember between sessions\", \"Dory pattern\", \"agent continuity\", or \"workspace setup\".","description":"---\nname: dory-memory\ndescription: File-based memory system for AI agents that forget between sessions. Implements the \"Dory-Proof\" pattern for continuity across context resets. Use when setting up agent memory, building workspace structure, implementing task tracking, or preventing context-loss errors. Triggers on \"memory system\", \"remember between sessions\", \"Dory pattern\", \"agent continuity\", or \"workspace setup\".\n---\n\n# Dory-Proof Memory System\n\nAI agents forget everything between sessions. This skill implements a file-based memory system that survives context resets.\n\n## Core Principle\n\n**Text > Brain.** Write everything down. Files are memory. The agent only \"remembers\" what's on disk.\n\n## The Dory-Proof Pattern (Critical)\n\nWhen the user gives a task:\n1. **IMMEDIATELY** write their EXACT WORDS to `state/ACTIVE.md`\n2. Then interpret what it means\n3. Then do the work\n4. Mark complete when done\n\n**Why:** Paraphrasing introduces drift. Exact words preserve intent across context flushes.\n\n## Workspace Structure\n\n```\nworkspace/\n├── AGENTS.md        # Operating rules (system file, don't rename)\n├── SOUL.md          # Identity + personality\n├── USER.md          # About the human\n├── MEMORY.md        # Curated long-term memory (<10KB)\n├── LESSONS.md       # \"Never again\" safety rules\n├── TOOLS.md         # Tool-specific notes\n│\n├── state/           # Active state (check every session)\n│   ├── ACTIVE.md    # Current task (exact user words)\n│   ├── HOLD.md      # Blocked items (check before acting!)\n│   ├── STAGING.md   # Drafts awaiting approval\n│   └── DECISIONS.md # Recent choices with timestamps\n│\n├── memory/          # Historical\n│   ├── YYYY-MM-DD.md\n│   ├── recent-work.md\n│   └── archive/\n│\n└── ops/             # Operational\n    └── WORKSPACE-INDEX.md\n```\n\n## Boot Sequence (Every Session)\n\n1. Read `state/HOLD.md` — what's BLOCKED\n2. Read `state/ACTIVE.md` — current task\n3. Read `state/DECISIONS.md` — recent choices\n4. Read `memory/recent-work.md` — last 48 hours\n5. Read `MEMORY.md` — long-term (main session only)\n\nOutput status line after boot:\n```\n📋 Boot: ACTIVE=[task] | HOLD=[n] items | STAGING=[n] drafts\n```\n\n## State File Formats\n\n### state/ACTIVE.md\n```markdown\n## Current Instruction\n**User said:** \"[exact quote]\"\n**Interpretation:** [what you think it means]\n**Status:**\n- [ ] Step 1\n- [ ] Step 2\n```\n\n### state/HOLD.md\n```markdown\n[YYYY-MM-DD HH:MM | session] Item — reason blocked\n```\n**ALL agents must check before acting on anything that looks ready.**\n\n### state/DECISIONS.md\n```markdown\n[YYYY-MM-DD HH:MM | session] Decision made\n```\n\n## Conflict Resolution\n\nWhen files conflict, priority (highest first):\n1. **state/HOLD.md** — blocks override all\n2. **state/ACTIVE.md** — current instruction\n3. **state/DECISIONS.md** — recent choices\n4. **AGENTS.md** — general rules\n\n## Memory Scoring (Before Saving to MEMORY.md)\n\nScore on 4 axes (0–3 each):\n\n| Axis | 0 | 1 | 2 | 3 |\n|------|---|---|---|---|\n| Longevity | Gone tomorrow | Weeks | Months | Years+ |\n| Reuse | One-off | Occasional | Frequent | Every session |\n| Impact | Trivial | Nice to know | Changes outputs | Changes decisions |\n| Uniqueness | Obvious | Slightly helpful | Hard to rederive | Impossible without |\n\n**Save if:** Total ≥ 8, OR any axis = 3 AND total ≥ 6.\n\n## Quick Setup\n\nCopy template files from `assets/templates/` to your workspace:\n```bash\ncp -r skills/dory-memory/assets/templates/* ~/.openclaw/workspace/\n```\n\nThen customize SOUL.md and USER.md for your agent.\n\n## References\n\n- `references/IMPLEMENTATION-GUIDE.md` — Full setup walkthrough\n- `references/ANTI-PATTERNS.md` — Common mistakes to avoid\n","tags":{"latest":"1.0.0"},"stats":{"comments":0,"downloads":2780,"installsAllTime":12,"installsCurrent":12,"stars":4,"versions":1},"createdAt":1770515052867,"updatedAt":1778486077727},"latestVersion":{"version":"1.0.0","createdAt":1770515052867,"changelog":"Initial release: file-based memory for AI agents that forget","license":null},"metadata":null,"owner":{"handle":"justinhartbiz","userId":"s17b6mcexg4qqgbmmaq4xw7ksh884v78","displayName":"justinhartbiz","image":"https://avatars.githubusercontent.com/u/27245787?v=4"},"moderation":null}