# Send MemSync Dual Memory System to your agent Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually. ## Fast path - 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. ## Suggested prompts ### New install ```text 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. ``` ### Upgrade existing ```text 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. ``` ## Machine-readable fields ```json { "schemaVersion": "1.0", "item": { "slug": "openclaw-memvid-logger", "name": "MemSync Dual Memory System", "source": "tencent", "type": "skill", "category": "开发工具", "sourceUrl": "https://clawhub.ai/stackBlock/openclaw-memvid-logger", "canonicalUrl": "https://clawhub.ai/stackBlock/openclaw-memvid-logger", "targetPlatform": "OpenClaw" }, "install": { "downloadUrl": "/downloads/openclaw-memvid-logger", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=openclaw-memvid-logger", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "packageFormat": "ZIP package", "primaryDoc": "SKILL.md", "includedAssets": [ "license.txt", "TEMPLATE.md", "README-clawhub.md", "instructions.md", "README.md", "SKILL.md" ], "downloadMode": "redirect", "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/openclaw-memvid-logger" }, "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." ] } }, "links": { "detailUrl": "https://openagent3.xyz/skills/openclaw-memvid-logger", "downloadUrl": "https://openagent3.xyz/downloads/openclaw-memvid-logger", "agentUrl": "https://openagent3.xyz/skills/openclaw-memvid-logger/agent", "manifestUrl": "https://openagent3.xyz/skills/openclaw-memvid-logger/agent.json", "briefUrl": "https://openagent3.xyz/skills/openclaw-memvid-logger/agent.md" } } ``` ## Documentation ### Unified Conversation Logger v1.2.5 Version: 1.2.5 (Critical Fixes Edition) Author: stackBlock License: MIT OpenClaw: >= 2026.2.12 A dual-output conversation logger for OpenClaw that captures everything - user messages, assistant responses, sub-agent conversations, tool calls, and system events - to both JSONL (backup) and Memvid (semantic search) formats. Memvid: A single-file memory layer for AI agents with instant retrieval and long-term memory. Persistent, versioned, and portable memory, without databases. "Replace complex RAG pipelines with a single portable file you own, and give your agent instant retrieval and long-term memory." ### ⚠️ Security & Privacy Notice Before installing, please understand: This skill captures everything - by design. It logs all user messages, assistant responses, sub-agent conversations, tool outputs, and system events to local files. This enables powerful long-term memory but requires trust. What you should know: Broad capture scope: This is intentional - the skill's purpose is complete conversation logging Sensitive data risk: Tool outputs (commands, API responses, file contents) are logged. Review what tools expose. Continuous logging: Once installed, it runs automatically on every assistant response until removed Optional cloud mode: API mode with MEMVID_API_KEY sends data to memvid.com (third-party service). Free/local modes keep data on your machine only. Your responsibility: Secure the JSONL/.mv2 files, rotate logs regularly, and audit what gets captured. Mitigations available: Use Free/Sharding mode to keep data local (no API key needed) Change default paths to encrypted locations Review tools/log.py before installing to understand exactly what gets logged File permissions: restrict access to log files (chmod 600) This skill is for users who want complete conversation memory and accept the privacy trade-offs. ### ✨ What Makes This Different 📝 Dual Storage - Every message saved to JSONL + Memvid simultaneously 🔍 Semantic Search - Ask "What did the researcher agent find about Tesla?" not just keyword search 🤖 Full Context - Captures user input, assistant output, agent chatter, tool results 💾 Three Modes - API (unlimited), Free (50MB), or Sharding (multi-file) 🚀 Always On - Hooks into OpenClaw automatically ### Critical Fixes Memvid Tag Format Fixed: Updated to KEY=VALUE format for Memvid 2.0+ compatibility Old (broken): --tag "user,telegram" New (fixed): --tag "role=user" --tag "source=telegram" Environment Variable Documentation: Added /etc/environment instructions (.bashrc doesn't work for background services) Hook Handler Format: Documented JavaScript (.js) requirement for OpenClaw 2026.2.12+ Comprehensive Troubleshooting: Added detailed troubleshooting section for common setup issues ### Compatibility Verified with OpenClaw 2026.2.12 Verified with Memvid CLI 2.0+ ### v1.2.4 Neural Search Default: Updated search guidance to use --mode neural as default for maximum accuracy Performance Documentation: Clarified latency trade-offs (~200ms for neural vs ~8ms for lexical) Search Mode Policy: Recommends neural for semantic understanding, lexical only when speed is critical ### v1.2.3 Version Cohesion: All files synchronized to v1.2.3 Documentation Consistency: README and SKILL.md now have matching content Security Improvements: Generic paths (no hardcoded user directories), install script asks permission Registry Compliance: Complete metadata (env vars, credentials, warnings) for ClawHub transparency Privacy Documentation: Comprehensive Security & Privacy Notice explaining data capture scope Role Tagging: Distinguishes user, assistant, agent:*, system, and tool messages Full Context: Captures sub-agent chatter, tool results, background processes Three Storage Modes: API mode (single file), Free mode (50MB), Sharding mode (monthly rotation) Semantic Search: Ask "What did the researcher agent find?" or "What did I say about X?" ### Option 1: API Mode (Recommended) - Near Limitless Memory Best for: Heavy users, unified search across everything Cost: $59-299/month via memvid.com # 1. Get API key from memvid.com ($59/month for 1GB, $299 for 25GB) export MEMVID_API_KEY="your_api_key_here" export MEMVID_MODE="single" # 2. Install npm install -g memvid git clone https://github.com/stackBlock/openclaw-memvid-logger.git cp -r openclaw-memvid-logger ~/.openclaw/workspace/skills/ # 3. Create unified memory file memvid create ~/memory.mv2 # 4. Start OpenClaw - everything logs to one searchable file Search everything at once: memvid ask memory.mv2 "What did we discuss about BadjAI?" memvid ask memory.mv2 "What did the researcher agent find about Tesla?" memvid ask memory.mv2 "Show me all the Python scripts I asked for" ### Option 2: Free Mode (50MB Limit) - Complete Memory in One Place Best for: Testing, light usage, single searchable file Cost: FREE # 1. Install (no API key needed) npm install -g memvid git clone https://github.com/stackBlock/openclaw-memvid-logger.git cp -r openclaw-memvid-logger ~/.openclaw/workspace/skills/ export MEMVID_MODE="single" # 2. Create memory file memvid create ~/memory.mv2 # 3. Start OpenClaw ⚠️ Limit: 50MB (~5,000 conversation turns). When you hit it: Archive and start fresh, OR Upgrade to API mode ($59-299/month), OR Switch to Sharding mode ### Option 3: Sharding Mode - More Than 50MB, Free Forever Best for: Long-term use, staying under free tier Cost: FREE Trade-off: Multi-file search # 1. Install (no API key needed) npm install -g memvid git clone https://github.com/stackBlock/openclaw-memvid-logger.git cp -r openclaw-memvid-logger ~/.openclaw/workspace/skills/ export MEMVID_MODE="monthly" # This is the default # 2. Start OpenClaw - auto-creates monthly files How it works: memory_2026-02.mv2 (February) memory_2026-03.mv2 (March - auto-created) Each file stays under 50MB ⚠️ Sharding Search Differences: Single-file search (API/Free modes): # One search gets everything memvid ask memory.mv2 "What car did I decide to buy?" # Returns: Results from ALL conversations across ALL time Sharding search (requires multiple queries): # Must search each month separately memvid ask memory_2026-02.mv2 "car decision" # Recent memvid ask memory_2026-01.mv2 "car decision" # January # Or use a wrapper script to search all files for file in memory_*.mv2; do echo "=== $file ===" memvid ask "$file" "car decision" 2>/dev/null | head -5 done # You must know which month the conversation happened # No cross-month context - "compare this month to last month" won't work Why sharding is harder: Can't ask "what did we discuss in the past 3 months?" in one query No unified timeline across months Must remember which month you talked about what No cross-file semantic comparison ### Role Tags (Automatic) RoleTagExample SearchUser[user]"What did I say about Mercedes?"Assistant[assistant]"What did you recommend?"Sub-agents[agent:researcher], [agent:coder]"What did the researcher find?"System[system]"When did the cron job run?"Tools[tool:exec], [tool:browser]"What commands were run?" ### Everything Captured ✅ User messages (what you type) ✅ Assistant responses (what I say back) ✅ Sub-agent conversations (researcher, coder, vision, math, etc.) ✅ Tool executions (bash commands, browser actions, file edits) ✅ Background processes (cron jobs, heartbeats, scheduled tasks) ✅ System events (config changes, restarts, errors) ### Architecture ┌─────────────────────────────────────────┐ │ OpenClaw Ecosystem │ │ ┌─────────┐ ┌─────────┐ ┌─────────┐ │ │ │ User │ │Assistant│ │ Agents │ │ │ │ Messages│ │Responses│ │Research │ │ │ └────┬────┘ └────┬────┘ └────┬────┘ │ │ └─────────────┴─────────────┘ │ │ │ │ │ ┌──────▼──────┐ │ │ │ log.py │ │ │ │ (this skill)│ │ │ └──────┬──────┘ │ └─────────────────────┼───────────────────┘ │ ┌─────────────────┼─────────────────┐ ↓ ↓ ↓ ┌───────┐ ┌─────────────┐ ┌──────────┐ │ JSONL │ │ Memvid │ │ Search │ │ File │ │ Files │ │ Query │ └───────┘ └─────────────┘ └──────────┘ │ │ ↓ ↓ grep/jq memvid ask/find ### Natural Language Search # What did you say about...? memvid ask memory_2026-02.mv2 "What was your recommendation about the Mercedes vs Tesla?" # What did I ask for...? memvid ask memory_2026-02.mv2 "What Python scripts did I request last week?" # What did agents do...? memvid ask memory_2026-02.mv2 "What did the researcher agent find about options trading?" # System events...? memvid ask memory_2026-02.mv2 "When did the PowerSchool grades cron job run?" ### Keyword Search # Find specific terms memvid find memory_2026-02.mv2 --query "Mercedes" # With filters memvid find memory_2026-02.mv2 --query "script" --tag agent:coder ### Temporal Queries memvid when memory_2026-02.mv2 "yesterday" memvid when memory_2026-02.mv2 "last Tuesday" memvid when memory_2026-02.mv2 "3 days ago" ### ⚡ Search Performance Guide Memvid has three search modes. This skill uses --mode neural by default for maximum accuracy: ### Default: Neural Search (Recommended) # Always use neural for semantic understanding and context memvid ask memory.mv2 "What supplements did Dr. Sinclair recommend?" --mode neural memvid ask memory.mv2 "What did we discuss about BadjAI?" --mode neural memvid ask memory.mv2 "Show me the Python scripts I requested" --mode neural Speed: ~200ms | Best for: Semantic understanding, context, synonyms, conceptual relationships ### Alternative Modes (Use When Explicitly Requested) Mode 1: Lexical Search (Fastest) # Use only for exact keyword matching when speed is critical memvid find memory.mv2 --mode lex --query "metformin" Speed: ~8ms | Use when: Exact word matching needed, latency is critical Mode 2: Hybrid Search (Balanced) # Combines lexical + neural memvid find memory.mv2 --mode hybrid --query "diabetes medications" Speed: ~300-500ms | Use when: You want both exact matches and semantic similarity ### Why Neural as Default? ModeSpeedAccuracyUse Caseneural~200msHighestDefault - semantic understandinglex~8msKeyword onlySpeed-critical exact matcheshybrid~300-500msHighBalanced approach The ~200ms trade-off is worth it: Neural mode understands context, handles paraphrases, and finds conceptually related information that lexical search misses entirely. ### JSONL Backup # Quick grep grep "Mercedes" conversation_log.jsonl # Complex queries with jq jq 'select(.role_tag == "user" and .content | contains("Python"))' conversation_log.jsonl # Time range jq 'select(.timestamp >= "2026-02-01" and .timestamp < "2026-03-01")' conversation_log.jsonl ### Environment Variables VariableDefaultModeDescriptionMEMVID_API_KEY(none)APIYour memvid.com API keyMEMVID_MODEmonthlyAllsingle or monthlyJSONL_LOG_PATH~/workspace/conversation_log.jsonlAllBackup log fileMEMVID_PATH~/workspace/memory.mv2AllBase path for memory filesMEMVID_BIN~/.npm-global/bin/memvidAllPath to memvid CLI ### OpenClaw Hooks (Advanced) Add to openclaw.json: { "hooks": { "internal": { "enabled": true, "entries": { "conversation-logger": { "enabled": true, "command": "python3 ~/.openclaw/workspace/skills/unified-logger/tools/log.py" } } } } } ### Mode 1: Single File (API or Free Mode) memory.mv2 ├── [user] messages ├── [assistant] responses ├── [agent:researcher] findings ├── [agent:coder] code ├── [tool:exec] commands └── [system] events ### Mode 2: Sharding (Monthly Rotation) memory_2026-01.mv2 (January conversations) memory_2026-02.mv2 (February conversations) ← Current memory_2026-03.mv2 (March, auto-created on March 1) ### "Free tier limit exceeded" (Free Mode) # Option 1: Archive and start fresh mv memory.mv2 memory_archive.mv2 memvid create memory.mv2 # Option 2: Switch to monthly sharding export MEMVID_MODE="monthly" # Option 3: Get API key export MEMVID_API_KEY="your_key" # $59-299/month at memvid.com ### "Cannot find memory file" (Sharding Mode) Current month's file auto-creates. If missing: memvid create memory_$(date +%Y-%m).mv2 ### Missing agent conversations Agents log to their own sessions. Ensure skill is installed in main agent workspace and sub-agents inherit it. ### Search returns wrong speaker Memvid uses semantic search. Be specific: ❌ "Mercedes" → Returns all mentions ✅ "What did I say about Mercedes" → Targets [user] frames ✅ "Your recommendation about Mercedes" → Targets [assistant] frames ### Comparing the Three Modes FeatureAPI ModeFree ModeSharding ModeCost$59-299/moFREEFREECapacity1-25GB+50MBUnlimited (files)Files11Multiple (monthly)Unified Search✅ Yes✅ Yes❌ Per-file onlyCross-Context Search✅ Full history✅ Full history❌ Month isolatedBest ForPower usersTestingLong-term free useComplexitySimpleSimpleMust track files ### 💸 The Pricing Gap (AKA Why Sharding Exists) The situation: Memvid's pricing goes from $0 (50MB) straight to $59/month (25GB). The problem: That's like buying a Ferrari when you just need a Honda Civic for your commute. What we're doing about it: I reached out. While they consider it, Sharding Mode exists so you don't have to pay Ferrari prices for Honda Civic usage. You can help: If you also think $0 → $59 is a bit much, reach out to Memvid at memvid.com and tell them stackBlock sent you. The more voices, the faster we get that $10-20 middle tier for the rest of us. Until then: Sharding Mode. Because startups shouldn't have to choose between ramen and memory. 🍜 ### Future Enhancements Auto-archive old months to cold storage Web UI for browsing conversations Cross-file search wrapper script Export to other formats (Markdown, PDF) Conversation threading visualization ### Support GitHub Issues: github.com/stackBlock/openclaw-memvid-logger OpenClaw Discord: discord.com/invite/clawd Memvid Support: memvid.com/docs ### License MIT - See LICENSE About Memvid: Memvid is a single-file memory layer for AI agents with instant retrieval and long-term memory. Persistent, versioned, and portable memory, without databases. Replace complex RAG pipelines with a single portable file you own, and give your agent instant retrieval and long-term memory. ## Trust - Source: tencent - Verification: Indexed source record - Publisher: stackBlock - Version: 1.2.6 ## Source health - Status: healthy - Source download looks usable. - Yavira can redirect you to the upstream package for this source. - Health scope: source - Reason: direct_download_ok - Checked at: 2026-04-30T16:55:25.780Z - Expires at: 2026-05-07T16:55:25.780Z - Recommended action: Download for OpenClaw ## Links - [Detail page](https://openagent3.xyz/skills/openclaw-memvid-logger) - [Send to Agent page](https://openagent3.xyz/skills/openclaw-memvid-logger/agent) - [JSON manifest](https://openagent3.xyz/skills/openclaw-memvid-logger/agent.json) - [Markdown brief](https://openagent3.xyz/skills/openclaw-memvid-logger/agent.md) - [Download page](https://openagent3.xyz/downloads/openclaw-memvid-logger)