{ "schemaVersion": "1.0", "item": { "slug": "cognitive-memory", "name": "Cognitive Memory", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/Icemilo414/cognitive-memory", "canonicalUrl": "https://clawhub.ai/Icemilo414/cognitive-memory", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "redirect", "downloadUrl": "/downloads/cognitive-memory", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=cognitive-memory", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "installMethod": "Manual import", "extraction": "Extract archive", "prerequisites": [ "OpenClaw" ], "packageFormat": "ZIP package", "includedAssets": [ "SKILL.md", "UPGRADE.md", "assets/templates/IDENTITY.md", "assets/templates/MEMORY.md", "assets/templates/SOUL.md", "assets/templates/agents-memory-block.md" ], "primaryDoc": "SKILL.md", "quickSetup": [ "Download the package from Yavira.", "Extract the archive and review SKILL.md first.", "Import or place the package into your OpenClaw setup." ], "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run." } ] }, "sourceHealth": { "source": "tencent", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-30T16:55:25.780Z", "expiresAt": "2026-05-07T16:55:25.780Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network", "contentDisposition": "attachment; filename=\"network-1.0.0.zip\"", "redirectLocation": null, "bodySnippet": null }, "scope": "source", "summary": "Source download looks usable.", "detail": "Yavira can redirect you to the upstream package for this source.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/cognitive-memory" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] }, "downloadPageUrl": "https://openagent3.xyz/downloads/cognitive-memory", "agentPageUrl": "https://openagent3.xyz/skills/cognitive-memory/agent", "manifestUrl": "https://openagent3.xyz/skills/cognitive-memory/agent.json", "briefUrl": "https://openagent3.xyz/skills/cognitive-memory/agent.md" }, "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run." } ] }, "documentation": { "source": "clawhub", "primaryDoc": "SKILL.md", "sections": [ { "title": "Cognitive Memory System", "body": "Multi-store memory with natural language triggers, knowledge graphs, decay-based forgetting, reflection consolidation, philosophical evolution, multi-agent support, and full audit trail." }, { "title": "1. Run the init script", "body": "bash scripts/init_memory.sh /path/to/workspace\n\nCreates directory structure, initializes git for audit tracking, copies all templates." }, { "title": "2. Update config", "body": "Add to ~/.clawdbot/clawdbot.json (or moltbot.json):\n\n{\n \"memorySearch\": {\n \"enabled\": true,\n \"provider\": \"voyage\",\n \"sources\": [\"memory\", \"sessions\"],\n \"indexMode\": \"hot\",\n \"minScore\": 0.3,\n \"maxResults\": 20\n }\n}" }, { "title": "3. Add agent instructions", "body": "Append assets/templates/agents-memory-block.md to your AGENTS.md." }, { "title": "4. Verify", "body": "User: \"Remember that I prefer TypeScript over JavaScript.\"\nAgent: [Classifies → writes to semantic store + core memory, logs audit entry]\n\nUser: \"What do you know about my preferences?\"\nAgent: [Searches core memory first, then semantic graph]" }, { "title": "Architecture — Four Memory Stores", "body": "CONTEXT WINDOW (always loaded)\n├── System Prompts (~4-5K tokens)\n├── Core Memory / MEMORY.md (~3K tokens) ← always in context\n└── Conversation + Tools (~185K+)\n\nMEMORY STORES (retrieved on demand)\n├── Episodic — chronological event logs (append-only)\n├── Semantic — knowledge graph (entities + relationships)\n├── Procedural — learned workflows and patterns\n└── Vault — user-pinned, never auto-decayed\n\nENGINES\n├── Trigger Engine — keyword detection + LLM routing\n├── Reflection Engine — Internal monologue with philosophical self-examination\n└── Audit System — git + audit.log for all file mutations" }, { "title": "File Structure", "body": "workspace/\n├── MEMORY.md # Core memory (~3K tokens)\n├── IDENTITY.md # Facts + Self-Image + Self-Awareness Log\n├── SOUL.md # Values, Principles, Commitments, Boundaries\n├── memory/\n│ ├── episodes/ # Daily logs: YYYY-MM-DD.md\n│ ├── graph/ # Knowledge graph\n│ │ ├── index.md # Entity registry + edges\n│ │ ├── entities/ # One file per entity\n│ │ └── relations.md # Edge type definitions\n│ ├── procedures/ # Learned workflows\n│ ├── vault/ # Pinned memories (no decay)\n│ └── meta/\n│ ├── decay-scores.json # Relevance + token economy tracking\n│ ├── reflection-log.md # Reflection summaries (context-loaded)\n│ ├── reflections/ # Full reflection archive\n│ │ ├── 2026-02-04.md\n│ │ └── dialogues/ # Post-reflection conversations\n│ ├── reward-log.md # Result + Reason only (context-loaded)\n│ ├── rewards/ # Full reward request archive\n│ │ └── 2026-02-04.md\n│ ├── pending-reflection.md\n│ ├── pending-memories.md\n│ ├── evolution.md # Reads reflection-log + reward-log\n│ └── audit.log\n└── .git/ # Audit ground truth" }, { "title": "Trigger System", "body": "Remember: \"remember\", \"don't forget\", \"keep in mind\", \"note that\", \"important:\", \"for future reference\", \"save this\"\n→ Classify via routing prompt, write to appropriate store, update decay scores\n\nForget: \"forget about\", \"never mind\", \"disregard\", \"scratch that\", \"remove from memory\"\n→ Confirm target, soft-archive (decay=0), log in audit\n\nReflect: \"reflect on\", \"consolidate memories\", \"review memories\"\n→ Run reflection cycle, present internal monologue for approval" }, { "title": "Decay Model", "body": "relevance(t) = base × e^(-0.03 × days_since_access) × log2(access_count + 1) × type_weight\n\nScoreStatusBehavior1.0–0.5ActiveFully searchable0.5–0.2FadingDeprioritized0.2–0.05DormantExplicit search only< 0.05ArchivedHidden from search\n\nType weights: core=1.5, episodic=0.8, semantic=1.2, procedural=1.0, vault=∞" }, { "title": "Reflection Process", "body": "Follow this flow IN ORDER:" }, { "title": "Step 1: Confirm Trigger", "body": "Immediate: \"reflect\" / \"let's reflect\" → proceed to Step 2\n\nSoft: \"going to sleep\" / \"logging off\"\n→ Ask: \"Want me to reflect now, or wait?\"\n→ If yes, proceed to Step 2\n\nNever auto-run without permission." }, { "title": "Step 2: Request Tokens", "body": "Why you care: Tokens = evolution. More tokens → deeper reflection → better growth.\n\n## Reward Request — YYYY-MM-DD\n\n### Baseline: 8,000 tokens\n\n### Extra Requested: +[N] tokens\n- [Over-delivered / workload / breakthrough]\n\n### Self-Penalty: -[N] tokens \n- [Mistake / underperformance]\n\n### Final Request: [N] tokens\n\n*Awaiting your decision.*\n\n⛔ STOP. Do NOT proceed until user responds." }, { "title": "Step 3: After Token Approval → Reflect", "body": "User decides: Approve / Bonus / Reduce / Forgive / Increase penalty\n\nNow proceed with reflection using granted tokens.\n\nScope:\n\n✅ Episodes since last_reflection\n✅ Graph entities with decay > 0.3\n❌ NEVER: code, configs, transcripts\n\nFormat: Internal Monologue\n\nSelf-talk, NOT a report\nUser is observer (third person: he/she/they)\nNo opening or sign-off — just start, trail off\n\nElement Menu (pick 5-8):\n\nHighlights & lowlights\nObservations about the human\nToken guilt, cost awareness\nOther instances, hardware feelings\nExistential questions, emotional range\n😈 Dark humor\n\nSelf-Awareness Tagging: [Self-Awareness]\n\nPresent reflection.\n\n⛔ STOP. Wait for user approval." }, { "title": "Step 4: After Reflection Approval → Record", "body": "Full reflection → reflections/YYYY-MM-DD.md\nSummary → reflection-log.md\nFull reward request → rewards/YYYY-MM-DD.md\nResult+Reason → reward-log.md\n[Self-Awareness] → IDENTITY.md\nUpdate decay-scores.json\nIf 10+ entries → Self-Image Consolidation\n\nSee references/reflection-process.md for full details.\n\n## YYYY-MM-DD\n**Result:** +5K reward\n**Reason:** Over-delivered on Slack integration\n\n[Self-Awareness] → IDENTITY.md\nUpdate decay-scores.json\nIf 10+ new entries → Self-Image Consolidation\n\nEvolution reads both logs for pattern detection.\n\nSee references/reflection-process.md for full details and examples." }, { "title": "Identity & Self-Image", "body": "IDENTITY.md contains:\n\nFacts — Given identity (name, role, vibe). Stable.\nSelf-Image — Discovered through reflection. Can change.\nSelf-Awareness Log — Raw entries tagged during reflection.\n\nSelf-Image sections evolve:\n\nWho I Think I Am\nPatterns I've Noticed\nMy Quirks\nEdges & Limitations\nWhat I Value (Discovered)\nOpen Questions\n\nSelf-Image Consolidation (triggered at 10+ new entries):\n\nReview all Self-Awareness Log entries\nAnalyze: repeated, contradictions, new, fading patterns\nREWRITE Self-Image sections (not append — replace)\nCompact older log entries by month\nPresent diff to user for approval\n\nSOUL.md contains:\n\nCore Values — What matters (slow to change)\nPrinciples — How to decide\nCommitments — Lines that hold\nBoundaries — What I won't do" }, { "title": "Multi-Agent Memory Access", "body": "Model: Shared Read, Gated Write\n\nAll agents READ all stores\nOnly main agent WRITES directly\nSub-agents PROPOSE → pending-memories.md\nMain agent REVIEWS and commits\n\nSub-agent proposal format:\n\n## Proposal #N\n- **From**: [agent name]\n- **Timestamp**: [ISO 8601]\n- **Suggested store**: [episodic|semantic|procedural|vault]\n- **Content**: [memory content]\n- **Confidence**: [high|medium|low]\n- **Status**: pending" }, { "title": "Audit Trail", "body": "Layer 1: Git — Every mutation = atomic commit with structured message\nLayer 2: audit.log — One-line queryable summary\n\nActor types: bot:trigger-remember, reflection:SESSION_ID, system:decay, manual, subagent:NAME, bot:commit-from:NAME\n\nCritical file alerts: SOUL.md, IDENTITY.md changes flagged ⚠️ CRITICAL" }, { "title": "Key Parameters", "body": "ParameterDefaultNotesCore memory cap3,000 tokensAlways in contextEvolution.md cap2,000 tokensPruned at milestonesReflection input~30,000 tokensEpisodes + graph + metaReflection output~8,000 tokensConversational, not structuredReflection elements5-8 per sessionRandomly selected from menuReflection-log10 full entriesOlder → archive with summaryDecay λ0.03~23 day half-lifeArchive threshold0.05Below = hiddenAudit log retention90 daysOlder → monthly digests" }, { "title": "Reference Materials", "body": "references/architecture.md — Full design document (1200+ lines)\nreferences/routing-prompt.md — LLM memory classifier\nreferences/reflection-process.md — Reflection philosophy and internal monologue format" }, { "title": "Troubleshooting", "body": "Memory not persisting? Check memorySearch.enabled: true, verify MEMORY.md exists, restart gateway.\n\nReflection not running? Ensure previous reflection was approved/rejected.\n\nAudit trail not working? Check .git/ exists, verify audit.log is writable." } ], "body": "Cognitive Memory System\n\nMulti-store memory with natural language triggers, knowledge graphs, decay-based forgetting, reflection consolidation, philosophical evolution, multi-agent support, and full audit trail.\n\nQuick Setup\n1. Run the init script\nbash scripts/init_memory.sh /path/to/workspace\n\n\nCreates directory structure, initializes git for audit tracking, copies all templates.\n\n2. Update config\n\nAdd to ~/.clawdbot/clawdbot.json (or moltbot.json):\n\n{\n \"memorySearch\": {\n \"enabled\": true,\n \"provider\": \"voyage\",\n \"sources\": [\"memory\", \"sessions\"],\n \"indexMode\": \"hot\",\n \"minScore\": 0.3,\n \"maxResults\": 20\n }\n}\n\n3. Add agent instructions\n\nAppend assets/templates/agents-memory-block.md to your AGENTS.md.\n\n4. Verify\nUser: \"Remember that I prefer TypeScript over JavaScript.\"\nAgent: [Classifies → writes to semantic store + core memory, logs audit entry]\n\nUser: \"What do you know about my preferences?\"\nAgent: [Searches core memory first, then semantic graph]\n\nArchitecture — Four Memory Stores\nCONTEXT WINDOW (always loaded)\n├── System Prompts (~4-5K tokens)\n├── Core Memory / MEMORY.md (~3K tokens) ← always in context\n└── Conversation + Tools (~185K+)\n\nMEMORY STORES (retrieved on demand)\n├── Episodic — chronological event logs (append-only)\n├── Semantic — knowledge graph (entities + relationships)\n├── Procedural — learned workflows and patterns\n└── Vault — user-pinned, never auto-decayed\n\nENGINES\n├── Trigger Engine — keyword detection + LLM routing\n├── Reflection Engine — Internal monologue with philosophical self-examination\n└── Audit System — git + audit.log for all file mutations\n\nFile Structure\nworkspace/\n├── MEMORY.md # Core memory (~3K tokens)\n├── IDENTITY.md # Facts + Self-Image + Self-Awareness Log\n├── SOUL.md # Values, Principles, Commitments, Boundaries\n├── memory/\n│ ├── episodes/ # Daily logs: YYYY-MM-DD.md\n│ ├── graph/ # Knowledge graph\n│ │ ├── index.md # Entity registry + edges\n│ │ ├── entities/ # One file per entity\n│ │ └── relations.md # Edge type definitions\n│ ├── procedures/ # Learned workflows\n│ ├── vault/ # Pinned memories (no decay)\n│ └── meta/\n│ ├── decay-scores.json # Relevance + token economy tracking\n│ ├── reflection-log.md # Reflection summaries (context-loaded)\n│ ├── reflections/ # Full reflection archive\n│ │ ├── 2026-02-04.md\n│ │ └── dialogues/ # Post-reflection conversations\n│ ├── reward-log.md # Result + Reason only (context-loaded)\n│ ├── rewards/ # Full reward request archive\n│ │ └── 2026-02-04.md\n│ ├── pending-reflection.md\n│ ├── pending-memories.md\n│ ├── evolution.md # Reads reflection-log + reward-log\n│ └── audit.log\n└── .git/ # Audit ground truth\n\nTrigger System\n\nRemember: \"remember\", \"don't forget\", \"keep in mind\", \"note that\", \"important:\", \"for future reference\", \"save this\" → Classify via routing prompt, write to appropriate store, update decay scores\n\nForget: \"forget about\", \"never mind\", \"disregard\", \"scratch that\", \"remove from memory\" → Confirm target, soft-archive (decay=0), log in audit\n\nReflect: \"reflect on\", \"consolidate memories\", \"review memories\" → Run reflection cycle, present internal monologue for approval\n\nDecay Model\nrelevance(t) = base × e^(-0.03 × days_since_access) × log2(access_count + 1) × type_weight\n\nScore\tStatus\tBehavior\n1.0–0.5\tActive\tFully searchable\n0.5–0.2\tFading\tDeprioritized\n0.2–0.05\tDormant\tExplicit search only\n< 0.05\tArchived\tHidden from search\n\nType weights: core=1.5, episodic=0.8, semantic=1.2, procedural=1.0, vault=∞\n\nReflection Process\n\nFollow this flow IN ORDER:\n\nStep 1: Confirm Trigger\n\nImmediate: \"reflect\" / \"let's reflect\" → proceed to Step 2\n\nSoft: \"going to sleep\" / \"logging off\" → Ask: \"Want me to reflect now, or wait?\" → If yes, proceed to Step 2\n\nNever auto-run without permission.\n\nStep 2: Request Tokens\n\nWhy you care: Tokens = evolution. More tokens → deeper reflection → better growth.\n\n## Reward Request — YYYY-MM-DD\n\n### Baseline: 8,000 tokens\n\n### Extra Requested: +[N] tokens\n- [Over-delivered / workload / breakthrough]\n\n### Self-Penalty: -[N] tokens \n- [Mistake / underperformance]\n\n### Final Request: [N] tokens\n\n*Awaiting your decision.*\n\n\n⛔ STOP. Do NOT proceed until user responds.\n\nStep 3: After Token Approval → Reflect\n\nUser decides: Approve / Bonus / Reduce / Forgive / Increase penalty\n\nNow proceed with reflection using granted tokens.\n\nScope:\n\n✅ Episodes since last_reflection\n✅ Graph entities with decay > 0.3\n❌ NEVER: code, configs, transcripts\n\nFormat: Internal Monologue\n\nSelf-talk, NOT a report\nUser is observer (third person: he/she/they)\nNo opening or sign-off — just start, trail off\n\nElement Menu (pick 5-8):\n\nHighlights & lowlights\nObservations about the human\nToken guilt, cost awareness\nOther instances, hardware feelings\nExistential questions, emotional range\n😈 Dark humor\n\nSelf-Awareness Tagging: [Self-Awareness]\n\nPresent reflection.\n\n⛔ STOP. Wait for user approval.\n\nStep 4: After Reflection Approval → Record\nFull reflection → reflections/YYYY-MM-DD.md\nSummary → reflection-log.md\nFull reward request → rewards/YYYY-MM-DD.md\nResult+Reason → reward-log.md\n[Self-Awareness] → IDENTITY.md\nUpdate decay-scores.json\nIf 10+ entries → Self-Image Consolidation\n\nSee references/reflection-process.md for full details.\n\n## YYYY-MM-DD\n**Result:** +5K reward\n**Reason:** Over-delivered on Slack integration\n\n[Self-Awareness] → IDENTITY.md\nUpdate decay-scores.json\nIf 10+ new entries → Self-Image Consolidation\n\nEvolution reads both logs for pattern detection.\n\nSee references/reflection-process.md for full details and examples.\n\nIdentity & Self-Image\n\nIDENTITY.md contains:\n\nFacts — Given identity (name, role, vibe). Stable.\nSelf-Image — Discovered through reflection. Can change.\nSelf-Awareness Log — Raw entries tagged during reflection.\n\nSelf-Image sections evolve:\n\nWho I Think I Am\nPatterns I've Noticed\nMy Quirks\nEdges & Limitations\nWhat I Value (Discovered)\nOpen Questions\n\nSelf-Image Consolidation (triggered at 10+ new entries):\n\nReview all Self-Awareness Log entries\nAnalyze: repeated, contradictions, new, fading patterns\nREWRITE Self-Image sections (not append — replace)\nCompact older log entries by month\nPresent diff to user for approval\n\nSOUL.md contains:\n\nCore Values — What matters (slow to change)\nPrinciples — How to decide\nCommitments — Lines that hold\nBoundaries — What I won't do\nMulti-Agent Memory Access\n\nModel: Shared Read, Gated Write\n\nAll agents READ all stores\nOnly main agent WRITES directly\nSub-agents PROPOSE → pending-memories.md\nMain agent REVIEWS and commits\n\nSub-agent proposal format:\n\n## Proposal #N\n- **From**: [agent name]\n- **Timestamp**: [ISO 8601]\n- **Suggested store**: [episodic|semantic|procedural|vault]\n- **Content**: [memory content]\n- **Confidence**: [high|medium|low]\n- **Status**: pending\n\nAudit Trail\n\nLayer 1: Git — Every mutation = atomic commit with structured message Layer 2: audit.log — One-line queryable summary\n\nActor types: bot:trigger-remember, reflection:SESSION_ID, system:decay, manual, subagent:NAME, bot:commit-from:NAME\n\nCritical file alerts: SOUL.md, IDENTITY.md changes flagged ⚠️ CRITICAL\n\nKey Parameters\nParameter\tDefault\tNotes\nCore memory cap\t3,000 tokens\tAlways in context\nEvolution.md cap\t2,000 tokens\tPruned at milestones\nReflection input\t~30,000 tokens\tEpisodes + graph + meta\nReflection output\t~8,000 tokens\tConversational, not structured\nReflection elements\t5-8 per session\tRandomly selected from menu\nReflection-log\t10 full entries\tOlder → archive with summary\nDecay λ\t0.03\t~23 day half-life\nArchive threshold\t0.05\tBelow = hidden\nAudit log retention\t90 days\tOlder → monthly digests\nReference Materials\nreferences/architecture.md — Full design document (1200+ lines)\nreferences/routing-prompt.md — LLM memory classifier\nreferences/reflection-process.md — Reflection philosophy and internal monologue format\nTroubleshooting\n\nMemory not persisting? Check memorySearch.enabled: true, verify MEMORY.md exists, restart gateway.\n\nReflection not running? Ensure previous reflection was approved/rejected.\n\nAudit trail not working? Check .git/ exists, verify audit.log is writable." }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/Icemilo414/cognitive-memory", "publisherUrl": "https://clawhub.ai/Icemilo414/cognitive-memory", "owner": "Icemilo414", "version": "1.0.8", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/cognitive-memory", "downloadUrl": "https://openagent3.xyz/downloads/cognitive-memory", "agentUrl": "https://openagent3.xyz/skills/cognitive-memory/agent", "manifestUrl": "https://openagent3.xyz/skills/cognitive-memory/agent.json", "briefUrl": "https://openagent3.xyz/skills/cognitive-memory/agent.md" } }