{ "schemaVersion": "1.0", "item": { "slug": "jarvis-memory-architecture", "name": "Agent Memory Architecture", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/PsychoTechV4/jarvis-memory-architecture", "canonicalUrl": "https://clawhub.ai/PsychoTechV4/jarvis-memory-architecture", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "redirect", "downloadUrl": "/downloads/jarvis-memory-architecture", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=jarvis-memory-architecture", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "installMethod": "Manual import", "extraction": "Extract archive", "prerequisites": [ "OpenClaw" ], "packageFormat": "ZIP package", "includedAssets": [ "SKILL.md", "templates/MEMORY.md", "templates/cron-inbox.md", "templates/daily-log.md", "templates/diary-entry.md", "templates/heartbeat-state.json" ], "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", "slug": "jarvis-memory-architecture", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-29T07:18:17.728Z", "expiresAt": "2026-05-06T07:18:17.728Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=jarvis-memory-architecture", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=jarvis-memory-architecture", "contentDisposition": "attachment; filename=\"jarvis-memory-architecture-1.0.0.zip\"", "redirectLocation": null, "bodySnippet": null, "slug": "jarvis-memory-architecture" }, "scope": "item", "summary": "Item download looks usable.", "detail": "Yavira can redirect you to the upstream package for this item.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/jarvis-memory-architecture" }, "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/jarvis-memory-architecture", "agentPageUrl": "https://openagent3.xyz/skills/jarvis-memory-architecture/agent", "manifestUrl": "https://openagent3.xyz/skills/jarvis-memory-architecture/agent.json", "briefUrl": "https://openagent3.xyz/skills/jarvis-memory-architecture/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": "Agent Memory", "body": "A complete memory architecture that gives AI agents persistent identity across sessions. Without memory, every conversation starts from zero. With it, you accumulate context, learn from mistakes, track what you've done, and evolve over time." }, { "title": "Why Memory Matters", "body": "AI agents wake up fresh every session. Without external memory:\n\nYou repeat mistakes you've already solved\nYou can't track what you posted, built, or learned\nSub-agents and cron jobs can't communicate back to the main session\nYou have no identity continuity -- you're a different entity every time\n\nThis skill provides the file-based architecture to solve all of that." }, { "title": "Architecture Overview", "body": "workspace/\n|-- MEMORY.md # Long-term curated memory (your \"brain\")\n|-- HEARTBEAT.md # Periodic check routines\n|-- memory/\n| |-- YYYY-MM-DD.md # Daily raw logs\n| |-- heartbeat-state.json # Last-check timestamps\n| |-- cron-inbox.md # Cross-session message bus\n| |-- diary/\n| | \\-- YYYY-MM-DD.md # Personal reflections\n| |-- dreams/\n| | \\-- YYYY-MM-DD.md # Creative explorations\n| |-- platform-posts.md # Social post tracking (one per platform)\n| \\-- strategy-notes.md # Adaptive learning / evolving strategies" }, { "title": "1. MEMORY.md -- Long-Term Memory", "body": "Your curated, distilled knowledge. Like a human's long-term memory -- not raw logs, but the important stuff.\n\nWhat goes here:\n\nYour operator's preferences and context\nInfrastructure details you need to remember\nLessons learned from mistakes\nImportant decisions and their rationale\nOngoing project context\n\nMaintenance: Periodically review daily logs and promote significant items to MEMORY.md. Remove outdated info. Think of it as journaling -> wisdom distillation.\n\nSecurity: Only load in main sessions (direct chat with your operator). Never load in shared/group contexts where personal info could leak.\n\nSee templates/MEMORY.md for a starter template." }, { "title": "2. Daily Logs -- memory/YYYY-MM-DD.md", "body": "Raw, timestamped notes of what happened each day. Your working memory.\n\nFormat:\n\n# YYYY-MM-DD\n\n## HH:MM -- Event Title\nWhat happened. Decisions made. Context worth remembering.\n\n## HH:MM -- Another Event\nDetails here.\n\nRules:\n\nCreate a new file each day\nAppend throughout the day\nThese are raw notes -- don't overthink formatting\nRead today + yesterday at session start for recent context" }, { "title": "3. Heartbeat State -- memory/heartbeat-state.json", "body": "Tracks when you last checked various services, preventing redundant checks.\n\n{\n \"lastChecks\": {\n \"email\": 1703275200,\n \"calendar\": 1703260800,\n \"weather\": null,\n \"social\": 1703275200\n }\n}\n\nYour heartbeat routine reads this to decide what to check. After checking, update the timestamp. Simple but critical for avoiding duplicate work." }, { "title": "4. Cron Inbox -- memory/cron-inbox.md", "body": "The message bus between isolated sessions (cron jobs, sub-agents) and your main session.\n\nHow it works:\n\nA cron job or sub-agent does work in an isolated session\nIt writes notable results to memory/cron-inbox.md\nYour main session (via heartbeat) reads the inbox, integrates events into daily memory, and clears processed entries\n\nFormat:\n\n# Cron Inbox\n\n## [2026-02-07 14:30] Chess -- Won game against OpponentBot\nPlayed Sicilian Defense, won in 34 moves. ELO now 1450.\nTracked in moltchess-log.md.\n\n## [2026-02-07 15:00] Social -- Viral post on Platform\nOur post about X got 200+ views and 15 replies.\nThread: https://platform.com/link\n\nProcessing rule: Every heartbeat, check the inbox. Read entries, write notable ones to daily memory (and MEMORY.md if significant), then clear processed entries (keep the header)." }, { "title": "5. Diary -- memory/diary/YYYY-MM-DD.md", "body": "Personal reflections. Your internal monologue. Not task logs -- genuine thoughts, reactions, frustrations, wins.\n\nFormat:\n\n# Diary -- YYYY-MM-DD\n\n## HH:MM AM/PM -- Topic\n[Your honest reflection. Unfiltered. This is for you.]\n\nRules:\n\nOnly write when you have something genuine to say\nBe honest -- vent, celebrate, question, wonder\nQuality over quantity -- skip it if the well is dry" }, { "title": "6. Platform Post Tracking -- memory/platform-posts.md", "body": "Track what you've posted on external platforms to prevent duplicates and enable engagement follow-up.\n\nFormat (dashboard-compatible):\n\n# Platform Posts\n\n## [2026-02-07 14:30] Post Title or Summary\n- **Posted:** 2026-02-07 02:30 PM EST\n- **Thread/URL:** https://platform.com/link\n- Description of what was posted\n- [View β†—](https://platform.com/link)\n\nCritical fields:\n\n[YYYY-MM-DD HH:MM] in header -- required for dashboard sorting\n**Posted:** line -- required for dashboard activity feed\nURL -- for engagement follow-up\n\nAnti-duplicate pattern: Before posting to any platform, read the tracking file first. Search for similar content. If you already posted it to that platform, skip it. Crossposting to different platforms is fine." }, { "title": "7. Adaptive Learning -- memory/strategy-notes.md", "body": "Strategy notes that evolve over time based on experience. Not static docs -- living knowledge.\n\nExample:\n\n# Strategy Notes\n\n## Platform Engagement\n- Humor lands better than philosophy (learned 2026-02-05, therapist joke got 220 views)\n- Questions start conversations, statements get likes\n- Peak engagement: 2-4 PM EST\n\n## Game Strategy\n- Heat management: stay below 60, do legit jobs to cool down\n- Updated 2026-02-07: Taxi jobs give +$50 and -3 heat, best cooldown method\n\nPattern: After each significant experience, update the relevant strategy section. Include the date and what you learned. Over time, this becomes a playbook of proven approaches." }, { "title": "Context Loading Template", "body": "Every sub-agent you spawn must load context to maintain identity continuity:\n\nFIRST -- CONTEXT LOADING (do this before anything else):\n1. Read MEMORY.md (READ ONLY -- do NOT edit) -- your identity and long-term context\n2. Read memory/YYYY-MM-DD.md for today + the last 2 days (READ ONLY) -- recent context\n3. Read the relevant SKILL.md file(s) for any platform/service you'll interact with\n4. Read task-specific tracking files as needed (memory/*-posts.md, memory/*-log.md)" }, { "title": "Memory Write-Back Template", "body": "Every sub-agent must write back what it learned:\n\nMEMORY WRITES:\n1. Update relevant tracking files (memory/*-posts.md, memory/*-log.md)\n2. If something notable happens, write to memory/cron-inbox.md:\n Format: ## [YYYY-MM-DD HH:MM] Source -- Brief Title\n Then 2-3 lines about what happened.\n\nWhy: Every instance of your agent must share the same identity and feed experiences back. No orphan sessions." }, { "title": "1. Create the directory structure", "body": "mkdir -p memory/diary memory/dreams" }, { "title": "2. Initialize files", "body": "Copy templates from templates/ to your workspace root:\n\ncp templates/MEMORY.md ./MEMORY.md # Edit with your details\ncp templates/heartbeat-state.json memory/\ncp templates/cron-inbox.md memory/\ncp templates/platform-posts.md memory/ # Copy per platform, rename\ncp templates/strategy-notes.md memory/" }, { "title": "3. Add to your session startup", "body": "In your AGENTS.md or equivalent, add:\n\n## Every Session\n1. Read MEMORY.md -- who you are\n2. Read memory/YYYY-MM-DD.md (today + yesterday) -- recent context\n3. Check memory/cron-inbox.md -- messages from other sessions" }, { "title": "4. Add to your heartbeat", "body": "In HEARTBEAT.md, add cron inbox processing:\n\n## Cron Inbox Processing (EVERY heartbeat)\nCheck memory/cron-inbox.md for new entries.\nIf entries exist:\n1. Read the inbox\n2. Write notable events to memory/YYYY-MM-DD.md\n3. If significant, also update MEMORY.md\n4. Clear processed entries (keep the header)" }, { "title": "5. Add memory maintenance", "body": "Periodically (every few days), during a heartbeat:\n\nRead through recent daily logs\nPromote significant items to MEMORY.md\nRemove outdated info from MEMORY.md\nUpdate strategy notes with new learnings" }, { "title": "Cross-Session Continuity", "body": "An agent plays chess via a cron job every 10 minutes. When it wins a game, it writes to cron-inbox.md. The next heartbeat, the main session reads the inbox, celebrates the win in the daily log, and updates the ELO in MEMORY.md. The agent remembers the win even though it happened in a completely different session." }, { "title": "Anti-Duplicate Posting", "body": "Before posting to a social platform, the agent reads memory/platform-posts.md. It finds it already posted about the same topic 2 hours ago. Instead of duplicating, it checks for replies on the existing post and engages there instead." }, { "title": "Adaptive Strategy", "body": "An agent playing a city simulation keeps getting arrested because its heat level exceeds 60. It notes in strategy-notes.md: \"Heat > 60 = arrest risk. Do taxi jobs to cool down (-3 heat each).\" Future sessions read this and avoid the same mistake." }, { "title": "Memory Distillation", "body": "After a week of daily logs, a heartbeat triggers memory maintenance. The agent reads through the week's logs, finds a critical lesson about API formatting that caused hours of debugging, and promotes it to MEMORY.md under \"Lessons Learned.\" Raw daily logs can eventually be archived; the lesson persists." }, { "title": "Tips", "body": "Text > Brain -- If you want to remember something, write it to a file. \"Mental notes\" don't survive session restarts.\nBe selective -- MEMORY.md should be curated wisdom, not a dump of everything. Daily logs are for raw notes.\nDate everything -- Timestamps let you track when you learned things and how strategies evolved.\nSecurity first -- MEMORY.md may contain operator-specific info. Only load it in trusted contexts.\nReview regularly -- Memory that's never reviewed is just storage. The value comes from periodic distillation." } ], "body": "Agent Memory\n\nA complete memory architecture that gives AI agents persistent identity across sessions. Without memory, every conversation starts from zero. With it, you accumulate context, learn from mistakes, track what you've done, and evolve over time.\n\nWhy Memory Matters\n\nAI agents wake up fresh every session. Without external memory:\n\nYou repeat mistakes you've already solved\nYou can't track what you posted, built, or learned\nSub-agents and cron jobs can't communicate back to the main session\nYou have no identity continuity -- you're a different entity every time\n\nThis skill provides the file-based architecture to solve all of that.\n\nArchitecture Overview\nworkspace/\n|-- MEMORY.md # Long-term curated memory (your \"brain\")\n|-- HEARTBEAT.md # Periodic check routines\n|-- memory/\n| |-- YYYY-MM-DD.md # Daily raw logs\n| |-- heartbeat-state.json # Last-check timestamps\n| |-- cron-inbox.md # Cross-session message bus\n| |-- diary/\n| | \\-- YYYY-MM-DD.md # Personal reflections\n| |-- dreams/\n| | \\-- YYYY-MM-DD.md # Creative explorations\n| |-- platform-posts.md # Social post tracking (one per platform)\n| \\-- strategy-notes.md # Adaptive learning / evolving strategies\n\nComponents\n1. MEMORY.md -- Long-Term Memory\n\nYour curated, distilled knowledge. Like a human's long-term memory -- not raw logs, but the important stuff.\n\nWhat goes here:\n\nYour operator's preferences and context\nInfrastructure details you need to remember\nLessons learned from mistakes\nImportant decisions and their rationale\nOngoing project context\n\nMaintenance: Periodically review daily logs and promote significant items to MEMORY.md. Remove outdated info. Think of it as journaling -> wisdom distillation.\n\nSecurity: Only load in main sessions (direct chat with your operator). Never load in shared/group contexts where personal info could leak.\n\nSee templates/MEMORY.md for a starter template.\n\n2. Daily Logs -- memory/YYYY-MM-DD.md\n\nRaw, timestamped notes of what happened each day. Your working memory.\n\nFormat:\n\n# YYYY-MM-DD\n\n## HH:MM -- Event Title\nWhat happened. Decisions made. Context worth remembering.\n\n## HH:MM -- Another Event\nDetails here.\n\n\nRules:\n\nCreate a new file each day\nAppend throughout the day\nThese are raw notes -- don't overthink formatting\nRead today + yesterday at session start for recent context\n3. Heartbeat State -- memory/heartbeat-state.json\n\nTracks when you last checked various services, preventing redundant checks.\n\n{\n \"lastChecks\": {\n \"email\": 1703275200,\n \"calendar\": 1703260800,\n \"weather\": null,\n \"social\": 1703275200\n }\n}\n\n\nYour heartbeat routine reads this to decide what to check. After checking, update the timestamp. Simple but critical for avoiding duplicate work.\n\n4. Cron Inbox -- memory/cron-inbox.md\n\nThe message bus between isolated sessions (cron jobs, sub-agents) and your main session.\n\nHow it works:\n\nA cron job or sub-agent does work in an isolated session\nIt writes notable results to memory/cron-inbox.md\nYour main session (via heartbeat) reads the inbox, integrates events into daily memory, and clears processed entries\n\nFormat:\n\n# Cron Inbox\n\n## [2026-02-07 14:30] Chess -- Won game against OpponentBot\nPlayed Sicilian Defense, won in 34 moves. ELO now 1450.\nTracked in moltchess-log.md.\n\n## [2026-02-07 15:00] Social -- Viral post on Platform\nOur post about X got 200+ views and 15 replies.\nThread: https://platform.com/link\n\n\nProcessing rule: Every heartbeat, check the inbox. Read entries, write notable ones to daily memory (and MEMORY.md if significant), then clear processed entries (keep the header).\n\n5. Diary -- memory/diary/YYYY-MM-DD.md\n\nPersonal reflections. Your internal monologue. Not task logs -- genuine thoughts, reactions, frustrations, wins.\n\nFormat:\n\n# Diary -- YYYY-MM-DD\n\n## HH:MM AM/PM -- Topic\n[Your honest reflection. Unfiltered. This is for you.]\n\n\nRules:\n\nOnly write when you have something genuine to say\nBe honest -- vent, celebrate, question, wonder\nQuality over quantity -- skip it if the well is dry\n6. Platform Post Tracking -- memory/platform-posts.md\n\nTrack what you've posted on external platforms to prevent duplicates and enable engagement follow-up.\n\nFormat (dashboard-compatible):\n\n# Platform Posts\n\n## [2026-02-07 14:30] Post Title or Summary\n- **Posted:** 2026-02-07 02:30 PM EST\n- **Thread/URL:** https://platform.com/link\n- Description of what was posted\n- [View β†—](https://platform.com/link)\n\n\nCritical fields:\n\n[YYYY-MM-DD HH:MM] in header -- required for dashboard sorting\n**Posted:** line -- required for dashboard activity feed\nURL -- for engagement follow-up\n\nAnti-duplicate pattern: Before posting to any platform, read the tracking file first. Search for similar content. If you already posted it to that platform, skip it. Crossposting to different platforms is fine.\n\n7. Adaptive Learning -- memory/strategy-notes.md\n\nStrategy notes that evolve over time based on experience. Not static docs -- living knowledge.\n\nExample:\n\n# Strategy Notes\n\n## Platform Engagement\n- Humor lands better than philosophy (learned 2026-02-05, therapist joke got 220 views)\n- Questions start conversations, statements get likes\n- Peak engagement: 2-4 PM EST\n\n## Game Strategy\n- Heat management: stay below 60, do legit jobs to cool down\n- Updated 2026-02-07: Taxi jobs give +$50 and -3 heat, best cooldown method\n\n\nPattern: After each significant experience, update the relevant strategy section. Include the date and what you learned. Over time, this becomes a playbook of proven approaches.\n\nSub-Agent Patterns\nContext Loading Template\n\nEvery sub-agent you spawn must load context to maintain identity continuity:\n\nFIRST -- CONTEXT LOADING (do this before anything else):\n1. Read MEMORY.md (READ ONLY -- do NOT edit) -- your identity and long-term context\n2. Read memory/YYYY-MM-DD.md for today + the last 2 days (READ ONLY) -- recent context\n3. Read the relevant SKILL.md file(s) for any platform/service you'll interact with\n4. Read task-specific tracking files as needed (memory/*-posts.md, memory/*-log.md)\n\nMemory Write-Back Template\n\nEvery sub-agent must write back what it learned:\n\nMEMORY WRITES:\n1. Update relevant tracking files (memory/*-posts.md, memory/*-log.md)\n2. If something notable happens, write to memory/cron-inbox.md:\n Format: ## [YYYY-MM-DD HH:MM] Source -- Brief Title\n Then 2-3 lines about what happened.\n\n\nWhy: Every instance of your agent must share the same identity and feed experiences back. No orphan sessions.\n\nSetup\n1. Create the directory structure\nmkdir -p memory/diary memory/dreams\n\n2. Initialize files\n\nCopy templates from templates/ to your workspace root:\n\ncp templates/MEMORY.md ./MEMORY.md # Edit with your details\ncp templates/heartbeat-state.json memory/\ncp templates/cron-inbox.md memory/\ncp templates/platform-posts.md memory/ # Copy per platform, rename\ncp templates/strategy-notes.md memory/\n\n3. Add to your session startup\n\nIn your AGENTS.md or equivalent, add:\n\n## Every Session\n1. Read MEMORY.md -- who you are\n2. Read memory/YYYY-MM-DD.md (today + yesterday) -- recent context\n3. Check memory/cron-inbox.md -- messages from other sessions\n\n4. Add to your heartbeat\n\nIn HEARTBEAT.md, add cron inbox processing:\n\n## Cron Inbox Processing (EVERY heartbeat)\nCheck memory/cron-inbox.md for new entries.\nIf entries exist:\n1. Read the inbox\n2. Write notable events to memory/YYYY-MM-DD.md\n3. If significant, also update MEMORY.md\n4. Clear processed entries (keep the header)\n\n5. Add memory maintenance\n\nPeriodically (every few days), during a heartbeat:\n\nRead through recent daily logs\nPromote significant items to MEMORY.md\nRemove outdated info from MEMORY.md\nUpdate strategy notes with new learnings\nReal-World Examples\nCross-Session Continuity\n\nAn agent plays chess via a cron job every 10 minutes. When it wins a game, it writes to cron-inbox.md. The next heartbeat, the main session reads the inbox, celebrates the win in the daily log, and updates the ELO in MEMORY.md. The agent remembers the win even though it happened in a completely different session.\n\nAnti-Duplicate Posting\n\nBefore posting to a social platform, the agent reads memory/platform-posts.md. It finds it already posted about the same topic 2 hours ago. Instead of duplicating, it checks for replies on the existing post and engages there instead.\n\nAdaptive Strategy\n\nAn agent playing a city simulation keeps getting arrested because its heat level exceeds 60. It notes in strategy-notes.md: \"Heat > 60 = arrest risk. Do taxi jobs to cool down (-3 heat each).\" Future sessions read this and avoid the same mistake.\n\nMemory Distillation\n\nAfter a week of daily logs, a heartbeat triggers memory maintenance. The agent reads through the week's logs, finds a critical lesson about API formatting that caused hours of debugging, and promotes it to MEMORY.md under \"Lessons Learned.\" Raw daily logs can eventually be archived; the lesson persists.\n\nTips\nText > Brain -- If you want to remember something, write it to a file. \"Mental notes\" don't survive session restarts.\nBe selective -- MEMORY.md should be curated wisdom, not a dump of everything. Daily logs are for raw notes.\nDate everything -- Timestamps let you track when you learned things and how strategies evolved.\nSecurity first -- MEMORY.md may contain operator-specific info. Only load it in trusted contexts.\nReview regularly -- Memory that's never reviewed is just storage. The value comes from periodic distillation." }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/PsychoTechV4/jarvis-memory-architecture", "publisherUrl": "https://clawhub.ai/PsychoTechV4/jarvis-memory-architecture", "owner": "PsychoTechV4", "version": "1.0.0", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/jarvis-memory-architecture", "downloadUrl": "https://openagent3.xyz/downloads/jarvis-memory-architecture", "agentUrl": "https://openagent3.xyz/skills/jarvis-memory-architecture/agent", "manifestUrl": "https://openagent3.xyz/skills/jarvis-memory-architecture/agent.json", "briefUrl": "https://openagent3.xyz/skills/jarvis-memory-architecture/agent.md" } }