{ "schemaVersion": "1.0", "item": { "slug": "total-recall", "name": "Total Recall", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/gavdalf/total-recall", "canonicalUrl": "https://clawhub.ai/gavdalf/total-recall", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "redirect", "downloadUrl": "/downloads/total-recall", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=total-recall", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "installMethod": "Manual import", "extraction": "Extract archive", "prerequisites": [ "OpenClaw" ], "packageFormat": "ZIP package", "includedAssets": [ "CHANGELOG.md", "CODE_OF_CONDUCT.md", "CONTRIBUTING.md", "INSTALL-AGENT.md", "README.md", "SKILL.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. Then review README.md for any prerequisites, environment setup, or post-install checks. 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. Then review README.md for any prerequisites, environment setup, or post-install checks. 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/total-recall" }, "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/total-recall", "agentPageUrl": "https://openagent3.xyz/skills/total-recall/agent", "manifestUrl": "https://openagent3.xyz/skills/total-recall/agent.json", "briefUrl": "https://openagent3.xyz/skills/total-recall/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. Then review README.md for any prerequisites, environment setup, or post-install checks. 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. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run." } ] }, "documentation": { "source": "clawhub", "primaryDoc": "SKILL.md", "sections": [ { "title": "Total Recall — Autonomous Agent Memory", "body": "The only memory skill that watches on its own.\n\nNo database. No vectors. No manual saves. Just an LLM observer that compresses your conversations into prioritised notes, consolidates when they grow, and recovers anything missed. Five layers of redundancy, zero maintenance. ~$0.10/month.\n\nWhile other memory skills ask you to remember to remember, this one just pays attention." }, { "title": "Architecture", "body": "Layer 1: Observer (cron, every 15-30 min)\n ↓ compresses recent messages → observations.md\nLayer 2: Reflector (auto-triggered when observations > 8000 words)\n ↓ consolidates, removes superseded info → 40-60% reduction\nLayer 3: Session Recovery (runs on every /new or /reset)\n ↓ catches any session the Observer missed\nLayer 4: Reactive Watcher (inotify daemon, Linux only)\n ↓ triggers Observer after 40+ new JSONL writes, 5-min cooldown\nLayer 5: Pre-compaction hook (memoryFlush)\n ↓ emergency capture before OpenClaw compacts context" }, { "title": "What It Does", "body": "Observer reads recent session transcripts (JSONL), sends them to an LLM, and appends compressed observations to observations.md with priority levels (high, medium, low)\nReflector kicks in when observations grow too large, consolidating related items and dropping stale low-priority entries\nSession Recovery runs at session start, checks if the previous session was captured, and does an emergency observation if not\nReactive Watcher watches the session directory with inotify so high-activity periods get captured faster than the cron interval\nPre-compaction hook fires when OpenClaw is about to compact context, ensuring nothing is lost" }, { "title": "1. Install the skill", "body": "clawdhub install total-recall" }, { "title": "2. Set your API key", "body": "Add to your .env or OpenClaw config:\n\nOPENROUTER_API_KEY=sk-or-v1-xxxxx" }, { "title": "3. Run the setup script", "body": "bash skills/total-recall/scripts/setup.sh\n\nThis will:\n\nCreate the memory directory structure (memory/, logs/, backups)\nOn Linux with inotify + systemd: install the reactive watcher service\nPrint cron job and agent configuration instructions for you to add manually" }, { "title": "4. Configure your agent to load observations", "body": "Add to your agent's workspace context (e.g., MEMORY.md or system prompt):\n\nAt session startup, read `memory/observations.md` for cross-session context.\n\nOr use OpenClaw's memoryFlush.systemPrompt to inject a startup instruction." }, { "title": "Platform Support", "body": "PlatformObserver + Reflector + RecoveryReactive WatcherLinux (Debian/Ubuntu/etc.)Full supportWith inotify-toolsmacOSFull supportNot available (cron-only)\n\nAll core scripts use portable bash. stat, date, and md5 commands are handled cross-platform via _compat.sh." }, { "title": "Configuration", "body": "All scripts read from environment variables with sensible defaults:\n\nVariableDefaultDescriptionOPENROUTER_API_KEY(required)OpenRouter API key for LLM callsMEMORY_DIR$OPENCLAW_WORKSPACE/memoryWhere observations.md livesSESSIONS_DIR~/.openclaw/agents/main/sessionsOpenClaw session transcriptsOBSERVER_MODELdeepseek/deepseek-v3.2Primary model for compressionOBSERVER_FALLBACK_MODELgoogle/gemini-2.5-flashFallback if primary failsOBSERVER_LOOKBACK_MIN15Minutes to look back (daytime)OBSERVER_MORNING_LOOKBACK_MIN480Minutes to look back (before 8am)OBSERVER_LINE_THRESHOLD40Lines before reactive trigger (Linux)OBSERVER_COOLDOWN_SECS300Cooldown between reactive triggers (Linux)REFLECTOR_WORD_THRESHOLD8000Words before reflector runsOPENCLAW_WORKSPACE~/your-workspaceWorkspace root" }, { "title": "LLM Provider Configuration", "body": "Total Recall uses any OpenAI-compatible chat completion API. Switch providers by setting environment variables:\n\nVariableDefaultDescriptionLLM_BASE_URLhttps://openrouter.ai/api/v1API endpointLLM_API_KEYfalls back to OPENROUTER_API_KEYAPI keyLLM_MODELdeepseek/deepseek-v3.2Model to use" }, { "title": "Provider examples", "body": "# OpenRouter (default)\nexport OPENROUTER_API_KEY=\"your-key\"\n\n# Ollama (local)\nexport LLM_BASE_URL=\"http://localhost:11434/v1\"\nexport LLM_API_KEY=\"ollama\"\nexport LLM_MODEL=\"llama3.1:8b\"\n\n# Groq\nexport LLM_BASE_URL=\"https://api.groq.com/openai/v1\"\nexport LLM_API_KEY=\"your-groq-key\"\nexport LLM_MODEL=\"llama-3.3-70b-versatile\"" }, { "title": "Files Created", "body": "memory/\n observations.md # The main observation log (loaded at startup)\n observation-backups/ # Reflector backups (last 10 kept)\n .observer-last-run # Timestamp of last observer run\n .observer-last-hash # Dedup hash of last processed messages\nlogs/\n observer.log\n reflector.log\n session-recovery.log\n observer-watcher.log" }, { "title": "Cron Jobs", "body": "The setup script creates these OpenClaw cron jobs:\n\nJobScheduleDescriptionmemory-observerEvery 15 minCompress recent conversationmemory-reflectorHourlyConsolidate if observations are large" }, { "title": "Reactive Watcher (Linux only)", "body": "The reactive watcher uses inotifywait to detect session activity and trigger the observer faster than cron alone. Requires Linux with inotify-tools installed.\n\n# Install inotify-tools (Debian/Ubuntu)\nsudo apt install inotify-tools\n\n# Check watcher status\nsystemctl --user status total-recall-watcher\n\n# View logs\njournalctl --user -u total-recall-watcher -f" }, { "title": "Cost", "body": "Using DeepSeek v3.2 via OpenRouter:\n\n~$0.03-0.10/month for typical usage (observer + reflector)\n~15-30 cron runs/day, each processing a few hundred tokens" }, { "title": "Observer", "body": "Finds recently modified session JSONL files\nFilters out subagent/cron sessions\nExtracts user + assistant messages from the lookback window\nDeduplicates using MD5 hash comparison\nSends to LLM with the observer prompt (priority-based compression)\nAppends result to observations.md\nIf observations exceed the word threshold, triggers reflector" }, { "title": "Reflector", "body": "Backs up current observations\nSends entire log to LLM with consolidation instructions\nValidates output is shorter than input (sanity check)\nReplaces observations with consolidated version\nCleans old backups (keeps last 10)" }, { "title": "Session Recovery", "body": "Runs at every /new or /reset\nHashes recent lines of the last session file\nCompares against stored hash from last observer run\nIf mismatch: runs observer in recovery mode (4-hour lookback)\nFallback: raw message extraction if observer fails" }, { "title": "Reactive Watcher", "body": "Uses inotifywait to monitor session directory\nCounts JSONL writes to main session files only\nAfter 40+ lines: triggers observer (with 5-min cooldown)\nResets counter when cron/external observer runs are detected" }, { "title": "Customizing the Prompts", "body": "The observer and reflector system prompts are in prompts/:\n\nprompts/observer-system.txt — controls how conversations are compressed\nprompts/reflector-system.txt — controls how observations are consolidated\n\nEdit these to match your agent's personality and priorities." }, { "title": "Dream Cycle", "body": "The Dream Cycle is an optional nightly agent that runs after hours to consolidate observations.md. It archives stale items and adds semantic hooks so nothing useful is actually lost. Context stays lean; everything remains findable." }, { "title": "What It Does", "body": "Classifies every observation by impact (critical / high / medium / low / minimal) and age\nArchives items that have passed their relevance threshold\nAdds a semantic hook for each archived item (specific keywords + archive reference)\nValidates the result and rolls back automatically if something goes wrong" }, { "title": "Features", "body": "Multi-Hook Retrieval — 4-5 alternative search phrasings per archived item. Searches using different words than the original still find the memory.\n\nConfidence Scoring — every observation gets a confidence score (0.0-1.0) and source type (explicit, implicit, inference, weak, uncertain). High-confidence items are preserved longer; low-confidence items are archived sooner.\n\nMemory Type System — 7 types with per-type TTLs: event (14d), fact (90d), preference (180d), goal (365d), habit (365d), rule (never), context (30d). Embedded as invisible HTML metadata comments in observations.md.\n\nObservation Chunking — clusters of 3+ related observations are compressed into single summary entries. Source observations are archived; a chunk hook replaces them. Achieves up to 75% token reduction.\n\nImportance Decay — per-type daily decay applied to importance scores before each archival decision. Items that decay below the archive threshold are queued for removal. Rates: event (-0.5/day), fact (-0.1/day), preference (-0.02/day), rule/habit/goal (no decay).\n\nPattern Promotion — scans recent dream logs for recurring themes (3+ occurrences across 3+ separate days). Writes promotion proposals to memory/dream-staging/ for human review. Use staging-review.sh to list, show, approve, or reject proposals. The context type is never promoted automatically." }, { "title": "Setup", "body": "Run bash skills/total-recall/scripts/setup.sh — creates Dream Cycle directories automatically.\n\n\nAdd the nightly cron job:\n# Dream Cycle — nightly at 3am\n0 3 * * * OPENCLAW_WORKSPACE=~/your-workspace bash ~/your-workspace/skills/total-recall/scripts/dream-cycle.sh preflight\n\n\n\nConfigure your cron agent using prompts/dream-cycle-prompt.md as the system prompt. Recommended models: Claude Sonnet for the Dreamer (analysis + decisions), DeepSeek v3.2 for the Observer (cheap, fast).\n\n\nStart with READ_ONLY_MODE=true for the first few nights. Check memory/dream-logs/ after each run to verify what it would have archived.\n\n\nSwitch to READ_ONLY_MODE=false once satisfied." }, { "title": "Configuration", "body": "VariableDefaultDescriptionDREAM_TOKEN_TARGET8000Token target for observations.md after consolidationREAD_ONLY_MODEfalseSet true for dry-run analysis without writes" }, { "title": "Files", "body": "FileDescriptionscripts/dream-cycle.shShell helper: preflight, archive, update-observations, write-log, write-metrics, validate, rollbackprompts/dream-cycle-prompt.mdAgent prompt for the nightly Dream Cycle rundream-cycle/README.mdDream Cycle quick referenceschemas/observation-format.mdExtended observation metadata format" }, { "title": "Directories Created", "body": "memory/\n archive/\n observations/ # Archived items (one .md file per night)\n chunks/ # Chunked observation groups\n dream-logs/ # Nightly run reports\n dream-staging/ # Pattern promotion proposals awaiting human review\n .dream-backups/ # Pre-run safety backups\nresearch/\n dream-cycle-metrics/\n daily/ # JSON metrics per night" }, { "title": "Troubleshooting", "body": "Observer not running?\n\nCheck logs/observer.log for errors\nVerify OPENROUTER_API_KEY is set and valid\nConfirm cron is active: crontab -l\n\nObservations not being loaded at session start?\n\nEnsure your agent's startup instructions include reading memory/observations.md\nCheck MEMORY_DIR points to the right location\n\nReactive watcher not triggering (Linux)?\n\nRun systemctl --user status total-recall-watcher\nCheck inotify-tools is installed: which inotifywait\nView watcher logs: journalctl --user -u total-recall-watcher -f\n\nDream Cycle archiving too aggressively?\n\nEnable READ_ONLY_MODE=true and review dream logs before going live\nAdjust DREAM_TOKEN_TARGET upward to archive less per run\n\nDream Cycle not archiving enough?\n\nLower DREAM_TOKEN_TARGET to trigger more aggressive consolidation" }, { "title": "Inspired By", "body": "This system is inspired by how human memory works during sleep — the hippocampus (observer) captures experiences, and during sleep consolidation (reflector), important memories are strengthened while noise is discarded.\n\nRead more: Your AI Has an Attention Problem\n\n\"Get your ass to Mars.\" — Well, get your agent's memory to work." } ], "body": "Total Recall — Autonomous Agent Memory\n\nThe only memory skill that watches on its own.\n\nNo database. No vectors. No manual saves. Just an LLM observer that compresses your conversations into prioritised notes, consolidates when they grow, and recovers anything missed. Five layers of redundancy, zero maintenance. ~$0.10/month.\n\nWhile other memory skills ask you to remember to remember, this one just pays attention.\n\nArchitecture\nLayer 1: Observer (cron, every 15-30 min)\n ↓ compresses recent messages → observations.md\nLayer 2: Reflector (auto-triggered when observations > 8000 words)\n ↓ consolidates, removes superseded info → 40-60% reduction\nLayer 3: Session Recovery (runs on every /new or /reset)\n ↓ catches any session the Observer missed\nLayer 4: Reactive Watcher (inotify daemon, Linux only)\n ↓ triggers Observer after 40+ new JSONL writes, 5-min cooldown\nLayer 5: Pre-compaction hook (memoryFlush)\n ↓ emergency capture before OpenClaw compacts context\n\nWhat It Does\nObserver reads recent session transcripts (JSONL), sends them to an LLM, and appends compressed observations to observations.md with priority levels (high, medium, low)\nReflector kicks in when observations grow too large, consolidating related items and dropping stale low-priority entries\nSession Recovery runs at session start, checks if the previous session was captured, and does an emergency observation if not\nReactive Watcher watches the session directory with inotify so high-activity periods get captured faster than the cron interval\nPre-compaction hook fires when OpenClaw is about to compact context, ensuring nothing is lost\nQuick Start\n1. Install the skill\nclawdhub install total-recall\n\n2. Set your API key\n\nAdd to your .env or OpenClaw config:\n\nOPENROUTER_API_KEY=sk-or-v1-xxxxx\n\n3. Run the setup script\nbash skills/total-recall/scripts/setup.sh\n\n\nThis will:\n\nCreate the memory directory structure (memory/, logs/, backups)\nOn Linux with inotify + systemd: install the reactive watcher service\nPrint cron job and agent configuration instructions for you to add manually\n4. Configure your agent to load observations\n\nAdd to your agent's workspace context (e.g., MEMORY.md or system prompt):\n\nAt session startup, read `memory/observations.md` for cross-session context.\n\n\nOr use OpenClaw's memoryFlush.systemPrompt to inject a startup instruction.\n\nPlatform Support\nPlatform\tObserver + Reflector + Recovery\tReactive Watcher\nLinux (Debian/Ubuntu/etc.)\tFull support\tWith inotify-tools\nmacOS\tFull support\tNot available (cron-only)\n\nAll core scripts use portable bash. stat, date, and md5 commands are handled cross-platform via _compat.sh.\n\nConfiguration\n\nAll scripts read from environment variables with sensible defaults:\n\nVariable\tDefault\tDescription\nOPENROUTER_API_KEY\t(required)\tOpenRouter API key for LLM calls\nMEMORY_DIR\t$OPENCLAW_WORKSPACE/memory\tWhere observations.md lives\nSESSIONS_DIR\t~/.openclaw/agents/main/sessions\tOpenClaw session transcripts\nOBSERVER_MODEL\tdeepseek/deepseek-v3.2\tPrimary model for compression\nOBSERVER_FALLBACK_MODEL\tgoogle/gemini-2.5-flash\tFallback if primary fails\nOBSERVER_LOOKBACK_MIN\t15\tMinutes to look back (daytime)\nOBSERVER_MORNING_LOOKBACK_MIN\t480\tMinutes to look back (before 8am)\nOBSERVER_LINE_THRESHOLD\t40\tLines before reactive trigger (Linux)\nOBSERVER_COOLDOWN_SECS\t300\tCooldown between reactive triggers (Linux)\nREFLECTOR_WORD_THRESHOLD\t8000\tWords before reflector runs\nOPENCLAW_WORKSPACE\t~/your-workspace\tWorkspace root\nLLM Provider Configuration\n\nTotal Recall uses any OpenAI-compatible chat completion API. Switch providers by setting environment variables:\n\nVariable\tDefault\tDescription\nLLM_BASE_URL\thttps://openrouter.ai/api/v1\tAPI endpoint\nLLM_API_KEY\tfalls back to OPENROUTER_API_KEY\tAPI key\nLLM_MODEL\tdeepseek/deepseek-v3.2\tModel to use\nProvider examples\n# OpenRouter (default)\nexport OPENROUTER_API_KEY=\"your-key\"\n\n# Ollama (local)\nexport LLM_BASE_URL=\"http://localhost:11434/v1\"\nexport LLM_API_KEY=\"ollama\"\nexport LLM_MODEL=\"llama3.1:8b\"\n\n# Groq\nexport LLM_BASE_URL=\"https://api.groq.com/openai/v1\"\nexport LLM_API_KEY=\"your-groq-key\"\nexport LLM_MODEL=\"llama-3.3-70b-versatile\"\n\nFiles Created\nmemory/\n observations.md # The main observation log (loaded at startup)\n observation-backups/ # Reflector backups (last 10 kept)\n .observer-last-run # Timestamp of last observer run\n .observer-last-hash # Dedup hash of last processed messages\nlogs/\n observer.log\n reflector.log\n session-recovery.log\n observer-watcher.log\n\nCron Jobs\n\nThe setup script creates these OpenClaw cron jobs:\n\nJob\tSchedule\tDescription\nmemory-observer\tEvery 15 min\tCompress recent conversation\nmemory-reflector\tHourly\tConsolidate if observations are large\nReactive Watcher (Linux only)\n\nThe reactive watcher uses inotifywait to detect session activity and trigger the observer faster than cron alone. Requires Linux with inotify-tools installed.\n\n# Install inotify-tools (Debian/Ubuntu)\nsudo apt install inotify-tools\n\n# Check watcher status\nsystemctl --user status total-recall-watcher\n\n# View logs\njournalctl --user -u total-recall-watcher -f\n\nCost\n\nUsing DeepSeek v3.2 via OpenRouter:\n\n~$0.03-0.10/month for typical usage (observer + reflector)\n~15-30 cron runs/day, each processing a few hundred tokens\nHow It Works (Technical)\nObserver\nFinds recently modified session JSONL files\nFilters out subagent/cron sessions\nExtracts user + assistant messages from the lookback window\nDeduplicates using MD5 hash comparison\nSends to LLM with the observer prompt (priority-based compression)\nAppends result to observations.md\nIf observations exceed the word threshold, triggers reflector\nReflector\nBacks up current observations\nSends entire log to LLM with consolidation instructions\nValidates output is shorter than input (sanity check)\nReplaces observations with consolidated version\nCleans old backups (keeps last 10)\nSession Recovery\nRuns at every /new or /reset\nHashes recent lines of the last session file\nCompares against stored hash from last observer run\nIf mismatch: runs observer in recovery mode (4-hour lookback)\nFallback: raw message extraction if observer fails\nReactive Watcher\nUses inotifywait to monitor session directory\nCounts JSONL writes to main session files only\nAfter 40+ lines: triggers observer (with 5-min cooldown)\nResets counter when cron/external observer runs are detected\nCustomizing the Prompts\n\nThe observer and reflector system prompts are in prompts/:\n\nprompts/observer-system.txt — controls how conversations are compressed\nprompts/reflector-system.txt — controls how observations are consolidated\n\nEdit these to match your agent's personality and priorities.\n\nDream Cycle\n\nThe Dream Cycle is an optional nightly agent that runs after hours to consolidate observations.md. It archives stale items and adds semantic hooks so nothing useful is actually lost. Context stays lean; everything remains findable.\n\nWhat It Does\nClassifies every observation by impact (critical / high / medium / low / minimal) and age\nArchives items that have passed their relevance threshold\nAdds a semantic hook for each archived item (specific keywords + archive reference)\nValidates the result and rolls back automatically if something goes wrong\nFeatures\n\nMulti-Hook Retrieval — 4-5 alternative search phrasings per archived item. Searches using different words than the original still find the memory.\n\nConfidence Scoring — every observation gets a confidence score (0.0-1.0) and source type (explicit, implicit, inference, weak, uncertain). High-confidence items are preserved longer; low-confidence items are archived sooner.\n\nMemory Type System — 7 types with per-type TTLs: event (14d), fact (90d), preference (180d), goal (365d), habit (365d), rule (never), context (30d). Embedded as invisible HTML metadata comments in observations.md.\n\nObservation Chunking — clusters of 3+ related observations are compressed into single summary entries. Source observations are archived; a chunk hook replaces them. Achieves up to 75% token reduction.\n\nImportance Decay — per-type daily decay applied to importance scores before each archival decision. Items that decay below the archive threshold are queued for removal. Rates: event (-0.5/day), fact (-0.1/day), preference (-0.02/day), rule/habit/goal (no decay).\n\nPattern Promotion — scans recent dream logs for recurring themes (3+ occurrences across 3+ separate days). Writes promotion proposals to memory/dream-staging/ for human review. Use staging-review.sh to list, show, approve, or reject proposals. The context type is never promoted automatically.\n\nSetup\n\nRun bash skills/total-recall/scripts/setup.sh — creates Dream Cycle directories automatically.\n\nAdd the nightly cron job:\n\n# Dream Cycle — nightly at 3am\n0 3 * * * OPENCLAW_WORKSPACE=~/your-workspace bash ~/your-workspace/skills/total-recall/scripts/dream-cycle.sh preflight\n\n\nConfigure your cron agent using prompts/dream-cycle-prompt.md as the system prompt. Recommended models: Claude Sonnet for the Dreamer (analysis + decisions), DeepSeek v3.2 for the Observer (cheap, fast).\n\nStart with READ_ONLY_MODE=true for the first few nights. Check memory/dream-logs/ after each run to verify what it would have archived.\n\nSwitch to READ_ONLY_MODE=false once satisfied.\n\nConfiguration\nVariable\tDefault\tDescription\nDREAM_TOKEN_TARGET\t8000\tToken target for observations.md after consolidation\nREAD_ONLY_MODE\tfalse\tSet true for dry-run analysis without writes\nFiles\nFile\tDescription\nscripts/dream-cycle.sh\tShell helper: preflight, archive, update-observations, write-log, write-metrics, validate, rollback\nprompts/dream-cycle-prompt.md\tAgent prompt for the nightly Dream Cycle run\ndream-cycle/README.md\tDream Cycle quick reference\nschemas/observation-format.md\tExtended observation metadata format\nDirectories Created\nmemory/\n archive/\n observations/ # Archived items (one .md file per night)\n chunks/ # Chunked observation groups\n dream-logs/ # Nightly run reports\n dream-staging/ # Pattern promotion proposals awaiting human review\n .dream-backups/ # Pre-run safety backups\nresearch/\n dream-cycle-metrics/\n daily/ # JSON metrics per night\n\nTroubleshooting\n\nObserver not running?\n\nCheck logs/observer.log for errors\nVerify OPENROUTER_API_KEY is set and valid\nConfirm cron is active: crontab -l\n\nObservations not being loaded at session start?\n\nEnsure your agent's startup instructions include reading memory/observations.md\nCheck MEMORY_DIR points to the right location\n\nReactive watcher not triggering (Linux)?\n\nRun systemctl --user status total-recall-watcher\nCheck inotify-tools is installed: which inotifywait\nView watcher logs: journalctl --user -u total-recall-watcher -f\n\nDream Cycle archiving too aggressively?\n\nEnable READ_ONLY_MODE=true and review dream logs before going live\nAdjust DREAM_TOKEN_TARGET upward to archive less per run\n\nDream Cycle not archiving enough?\n\nLower DREAM_TOKEN_TARGET to trigger more aggressive consolidation\nInspired By\n\nThis system is inspired by how human memory works during sleep — the hippocampus (observer) captures experiences, and during sleep consolidation (reflector), important memories are strengthened while noise is discarded.\n\nRead more: Your AI Has an Attention Problem\n\n\"Get your ass to Mars.\" — Well, get your agent's memory to work." }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/gavdalf/total-recall", "publisherUrl": "https://clawhub.ai/gavdalf/total-recall", "owner": "gavdalf", "version": "1.5.1", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/total-recall", "downloadUrl": "https://openagent3.xyz/downloads/total-recall", "agentUrl": "https://openagent3.xyz/skills/total-recall/agent", "manifestUrl": "https://openagent3.xyz/skills/total-recall/agent.json", "briefUrl": "https://openagent3.xyz/skills/total-recall/agent.md" } }