{ "schemaVersion": "1.0", "item": { "slug": "openclaw-continuous-learning", "name": "OpenClaw Continuous Learning", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/adelpro/openclaw-continuous-learning", "canonicalUrl": "https://clawhub.ai/adelpro/openclaw-continuous-learning", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "manual_only", "downloadUrl": "/downloads/openclaw-continuous-learning", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=openclaw-continuous-learning", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "installMethod": "Manual import", "extraction": "Extract archive", "prerequisites": [ "OpenClaw" ], "packageFormat": "ZIP package", "includedAssets": [ "SKILL.md", "scripts/analyze.mjs" ], "primaryDoc": "SKILL.md", "quickSetup": [ "Open the source page and confirm the package flow manually.", "Review SKILL.md if you can obtain the files.", "Treat this source as manual setup until the download is verified." ], "agentAssist": { "summary": "Use the source page and any available docs to guide the install because the item currently does not return a direct package file.", "steps": [ "Open the source page via Open source listing.", "If you can obtain the package, extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the source page and extracted files." ], "prompts": [ { "label": "New install", "body": "I tried to install a skill package from Yavira, but the item currently does not return a direct package file. Inspect the source page and any extracted docs, then tell me what you can confirm and any manual steps still required." }, { "label": "Upgrade existing", "body": "I tried to upgrade a skill package from Yavira, but the item currently does not return a direct package file. Compare the source page and any extracted docs with my current installation, then summarize what changed and what manual follow-up I still need." } ] }, "sourceHealth": { "source": "tencent", "slug": "openclaw-continuous-learning", "status": "source_issue", "reason": "not_found", "recommendedAction": "review_source", "checkedAt": "2026-05-02T20:54:29.414Z", "expiresAt": "2026-05-03T20:54:29.414Z", "httpStatus": 404, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=openclaw-continuous-learning", "contentType": "text/plain", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=openclaw-continuous-learning", "contentDisposition": null, "redirectLocation": null, "bodySnippet": null, "slug": "openclaw-continuous-learning" }, "scope": "item", "summary": "Known item issue.", "detail": "This item's current download entry is known to bounce back to a listing or homepage instead of returning a package file.", "primaryActionLabel": "Open source listing", "primaryActionHref": "https://clawhub.ai/adelpro/openclaw-continuous-learning" }, "validation": { "installChecklist": [ "Open the source listing and confirm there is a real package or setup artifact available.", "Review SKILL.md before asking your agent to continue.", "Treat this source as manual setup until the upstream download flow is fixed." ], "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/openclaw-continuous-learning", "agentPageUrl": "https://openagent3.xyz/skills/openclaw-continuous-learning/agent", "manifestUrl": "https://openagent3.xyz/skills/openclaw-continuous-learning/agent.json", "briefUrl": "https://openagent3.xyz/skills/openclaw-continuous-learning/agent.md" }, "agentAssist": { "summary": "Use the source page and any available docs to guide the install because the item currently does not return a direct package file.", "steps": [ "Open the source page via Open source listing.", "If you can obtain the package, extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the source page and extracted files." ], "prompts": [ { "label": "New install", "body": "I tried to install a skill package from Yavira, but the item currently does not return a direct package file. Inspect the source page and any extracted docs, then tell me what you can confirm and any manual steps still required." }, { "label": "Upgrade existing", "body": "I tried to upgrade a skill package from Yavira, but the item currently does not return a direct package file. Compare the source page and any extracted docs with my current installation, then summarize what changed and what manual follow-up I still need." } ] }, "documentation": { "source": "clawhub", "primaryDoc": "SKILL.md", "sections": [ { "title": "Continuous Learning for AI Agents", "body": "An instinct-based learning system that helps AI agents improve themselves through observation and pattern detection." }, { "title": "What This Skill Does", "body": "Analyzes session history - Reviews agent interactions and outputs\nDetects patterns - Identifies recurring behaviors, preferences, workflows\nCreates instincts - Atomic learnings with confidence scores\nSuggests optimizations - Based on observed behavior patterns\nEnables self-evolution - Converts insights into improvements" }, { "title": "When to Use", "body": "Use when:\n\nBuilding self-improving AI agents\nWant agent to learn from interactions\nDiscovering optimization opportunities\nCreating adaptive automation\nTracking behavioral patterns\n\nSkip when:\n\nStatic, unchanging behavior preferred\nNo session history available\nSimple, deterministic workflows only" }, { "title": "Architecture", "body": "~/.openclaw/agents/ (session .jsonl files)\n │\n ▼\n┌───────────────────────────────────────────┐\n│ analyze.mjs │\n│ • Reads session history │\n│ • Extracts tool calls & errors │\n│ • Detects patterns │\n└───────────────────────────────────────────┘\n │\n ▼\n┌───────────────────────────────────────────┐\n│ memory/learning/ │\n│ • instincts.jsonl (atomic learnings) │\n│ • patterns.json (aggregated) │\n│ • optimizations.json (suggestions) │\n└───────────────────────────────────────────┘" }, { "title": "External Feedback (Sub-Skill)", "body": "This skill works with agent-self-improvement (ClawHub) for external user feedback capture:\n\nInternal Learning: Session analysis (this skill)\nExternal Learning: User feedback via SKILL:agent-self-improvement" }, { "title": "Combined Usage", "body": "# Nightly: Internal analysis\nSKILL:openclaw-continuous-learning --analyze\n\n# After any output: Capture feedback\nSKILL:agent-self-improvement --job --feedback \"\"\n\n# Daily: Generate combined improvements\nSKILL:agent-self-improvement --improve all" }, { "title": "Feedback Flow", "body": "User Response → agent-self-improvement → Directive Hints\n ↓\nSession Analysis → openclaw-continuous-learning → Internal Patterns\n ↓\nCombined Insights → Agent Optimization\n\nBoth skills store learnings in memory/learning/ and can reference each other's data." }, { "title": "Confidence Scoring", "body": "ScoreMeaningBehavior0.3TentativeSuggested but not enforced0.5ModerateApplied when relevant0.7StrongAuto-approved0.9Core behaviorAlways apply\n\nConfidence increases when:\n\nPattern observed repeatedly\nUser doesn't correct behavior\nMultiple observations agree\n\nConfidence decreases when:\n\nUser explicitly corrects\nPattern not observed recently\nContradicting evidence appears" }, { "title": "Instincts", "body": "An instinct is a small learned behavior:\n\nid: prefer-simplicity\ntrigger: \"when solving problems\"\nconfidence: 0.7\ndomain: problem_solving\n---\n# Prefer Simple Solutions\n\n## Action\nAlways choose the simplest solution that meets requirements.\n\n## Evidence\n- Observed preference for minimal code\n- User corrected over-engineered approaches" }, { "title": "Patterns", "body": "Aggregated observations grouped by category:\n\ncode_style\ntesting\ngit\ndebugging\nworkflow\ncommunication" }, { "title": "Optimizations", "body": "Actionable improvements derived from patterns." }, { "title": "1. Agent Self-Improvement", "body": "Agent observes its own sessions:\n- What works consistently?\n- What gets corrected?\n- What patterns emerge?\n\nCreates instincts → Applies high-confidence patterns" }, { "title": "2. User Preference Learning", "body": "Learn user preferences from interactions:\n- Coding style preferences\n- Communication preferences\n- Workflow preferences\n\nAdapt behavior accordingly" }, { "title": "3. Performance Optimization", "body": "Detect performance patterns:\n- Slow operations\n- Bottlenecks\n- Optimization opportunities\n\nSuggest improvements" }, { "title": "4. Error Pattern Detection", "body": "Track error patterns:\n- Common failures\n- Resolution strategies\n- Prevention approaches\n\nBuild error-handling instincts" }, { "title": "Quick Start", "body": "# Analyze sessions (reads agent .jsonl files from ~/.openclaw/agents/)\ncd ~/.openclaw/workspace/skills/openclaw-continuous-learning\nnode scripts/analyze.mjs\n\n# List learned instincts\nnode scripts/analyze.mjs instincts\n\n# Show optimizations\nnode scripts/analyze.mjs list\n\n# Show error patterns\nnode scripts/analyze.mjs errors" }, { "title": "1. Create storage directory", "body": "mkdir -p ~/.openclaw/workspace/memory/learning" }, { "title": "2. Schedule analysis", "body": "Add to cron for periodic analysis:\n\n{\n \"id\": \"continuous-learning\",\n \"schedule\": \"0 22 * * *\"\n}" }, { "title": "3. Integrate with daily tips", "body": "Connect to daily summary for optimization delivery." }, { "title": "File Structure", "body": "~/.openclaw/workspace/\n└── memory/\n └── learning/\n ├── instincts.jsonl # Atomic learnings\n ├── patterns.json # Aggregated patterns\n └── optimizations.json # Suggestions" }, { "title": "Example Output", "body": "🧠 Learning Report\n\nPatterns Detected:\n- prefer-simplicity (0.7) ↑2\n- test-first (0.5) ↑1\n- commit-often (0.3) new\n\nConfidence Changes:\n- minimal-code: 0.5 → 0.7\n\nSuggested:\n1. Prioritize simple solutions\n2. Add pre-commit hooks\n3. Enable stricter typing" }, { "title": "Best Practices", "body": "Start simple - Few patterns, low confidence\nValidate often - Check if patterns still hold\nReview suggestions - Don't auto-apply everything\nTrack confidence - Update based on results\nExport/share - Build library of common patterns" }, { "title": "FAQ", "body": "How is this different from memory?\nMemory stores facts. This learns behavioral patterns and preferences.\n\nHow long to see results?\nDepends on session volume. Typically 1-2 weeks for meaningful patterns.\n\nIs it safe to auto-apply?\nOnly high-confidence (0.7+) patterns. Always review suggestions first." }, { "title": "Related Skills", "body": "skill-engineer - Quality-gated skill development\ncompound-engineering - Session review and learning\nmemory-setup - Memory configuration\nopenclaw-daily-tips - Daily optimization tips\n\nVersion: 1.1.0\nInspired by: Anthropic's continuous learning patterns, Claude Code homunculus" } ], "body": "Continuous Learning for AI Agents\n\nAn instinct-based learning system that helps AI agents improve themselves through observation and pattern detection.\n\nWhat This Skill Does\nAnalyzes session history - Reviews agent interactions and outputs\nDetects patterns - Identifies recurring behaviors, preferences, workflows\nCreates instincts - Atomic learnings with confidence scores\nSuggests optimizations - Based on observed behavior patterns\nEnables self-evolution - Converts insights into improvements\nWhen to Use\n\nUse when:\n\nBuilding self-improving AI agents\nWant agent to learn from interactions\nDiscovering optimization opportunities\nCreating adaptive automation\nTracking behavioral patterns\n\nSkip when:\n\nStatic, unchanging behavior preferred\nNo session history available\nSimple, deterministic workflows only\nArchitecture\n~/.openclaw/agents/ (session .jsonl files)\n │\n ▼\n┌───────────────────────────────────────────┐\n│ analyze.mjs │\n│ • Reads session history │\n│ • Extracts tool calls & errors │\n│ • Detects patterns │\n└───────────────────────────────────────────┘\n │\n ▼\n┌───────────────────────────────────────────┐\n│ memory/learning/ │\n│ • instincts.jsonl (atomic learnings) │\n│ • patterns.json (aggregated) │\n│ • optimizations.json (suggestions) │\n└───────────────────────────────────────────┘\n\nExternal Feedback (Sub-Skill)\n\nThis skill works with agent-self-improvement (ClawHub) for external user feedback capture:\n\nInternal Learning: Session analysis (this skill)\nExternal Learning: User feedback via SKILL:agent-self-improvement\nCombined Usage\n# Nightly: Internal analysis\nSKILL:openclaw-continuous-learning --analyze\n\n# After any output: Capture feedback\nSKILL:agent-self-improvement --job --feedback \"\"\n\n# Daily: Generate combined improvements\nSKILL:agent-self-improvement --improve all\n\nFeedback Flow\nUser Response → agent-self-improvement → Directive Hints\n ↓\nSession Analysis → openclaw-continuous-learning → Internal Patterns\n ↓\nCombined Insights → Agent Optimization\n\n\nBoth skills store learnings in memory/learning/ and can reference each other's data.\n\nConfidence Scoring\nScore\tMeaning\tBehavior\n0.3\tTentative\tSuggested but not enforced\n0.5\tModerate\tApplied when relevant\n0.7\tStrong\tAuto-approved\n0.9\tCore behavior\tAlways apply\n\nConfidence increases when:\n\nPattern observed repeatedly\nUser doesn't correct behavior\nMultiple observations agree\n\nConfidence decreases when:\n\nUser explicitly corrects\nPattern not observed recently\nContradicting evidence appears\nKey Concepts\nInstincts\n\nAn instinct is a small learned behavior:\n\nid: prefer-simplicity\ntrigger: \"when solving problems\"\nconfidence: 0.7\ndomain: problem_solving\n---\n# Prefer Simple Solutions\n\n## Action\nAlways choose the simplest solution that meets requirements.\n\n## Evidence\n- Observed preference for minimal code\n- User corrected over-engineered approaches\n\nPatterns\n\nAggregated observations grouped by category:\n\ncode_style\ntesting\ngit\ndebugging\nworkflow\ncommunication\nOptimizations\n\nActionable improvements derived from patterns.\n\nUse Cases\n1. Agent Self-Improvement\nAgent observes its own sessions:\n- What works consistently?\n- What gets corrected?\n- What patterns emerge?\n\nCreates instincts → Applies high-confidence patterns\n\n2. User Preference Learning\nLearn user preferences from interactions:\n- Coding style preferences\n- Communication preferences\n- Workflow preferences\n\nAdapt behavior accordingly\n\n3. Performance Optimization\nDetect performance patterns:\n- Slow operations\n- Bottlenecks\n- Optimization opportunities\n\nSuggest improvements\n\n4. Error Pattern Detection\nTrack error patterns:\n- Common failures\n- Resolution strategies\n- Prevention approaches\n\nBuild error-handling instincts\n\nQuick Start\n# Analyze sessions (reads agent .jsonl files from ~/.openclaw/agents/)\ncd ~/.openclaw/workspace/skills/openclaw-continuous-learning\nnode scripts/analyze.mjs\n\n# List learned instincts\nnode scripts/analyze.mjs instincts\n\n# Show optimizations\nnode scripts/analyze.mjs list\n\n# Show error patterns\nnode scripts/analyze.mjs errors\n\nSetup\n1. Create storage directory\nmkdir -p ~/.openclaw/workspace/memory/learning\n\n2. Schedule analysis\n\nAdd to cron for periodic analysis:\n\n{\n \"id\": \"continuous-learning\",\n \"schedule\": \"0 22 * * *\"\n}\n\n3. Integrate with daily tips\n\nConnect to daily summary for optimization delivery.\n\nFile Structure\n~/.openclaw/workspace/\n└── memory/\n └── learning/\n ├── instincts.jsonl # Atomic learnings\n ├── patterns.json # Aggregated patterns\n └── optimizations.json # Suggestions\n\nExample Output\n🧠 Learning Report\n\nPatterns Detected:\n- prefer-simplicity (0.7) ↑2\n- test-first (0.5) ↑1\n- commit-often (0.3) new\n\nConfidence Changes:\n- minimal-code: 0.5 → 0.7\n\nSuggested:\n1. Prioritize simple solutions\n2. Add pre-commit hooks\n3. Enable stricter typing\n\nBest Practices\nStart simple - Few patterns, low confidence\nValidate often - Check if patterns still hold\nReview suggestions - Don't auto-apply everything\nTrack confidence - Update based on results\nExport/share - Build library of common patterns\nFAQ\n\nHow is this different from memory? Memory stores facts. This learns behavioral patterns and preferences.\n\nHow long to see results? Depends on session volume. Typically 1-2 weeks for meaningful patterns.\n\nIs it safe to auto-apply? Only high-confidence (0.7+) patterns. Always review suggestions first.\n\nRelated Skills\nskill-engineer - Quality-gated skill development\ncompound-engineering - Session review and learning\nmemory-setup - Memory configuration\nopenclaw-daily-tips - Daily optimization tips\n\nVersion: 1.1.0\nInspired by: Anthropic's continuous learning patterns, Claude Code homunculus" }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/adelpro/openclaw-continuous-learning", "publisherUrl": "https://clawhub.ai/adelpro/openclaw-continuous-learning", "owner": "adelpro", "version": "1.3.0", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/openclaw-continuous-learning", "downloadUrl": "https://openagent3.xyz/downloads/openclaw-continuous-learning", "agentUrl": "https://openagent3.xyz/skills/openclaw-continuous-learning/agent", "manifestUrl": "https://openagent3.xyz/skills/openclaw-continuous-learning/agent.json", "briefUrl": "https://openagent3.xyz/skills/openclaw-continuous-learning/agent.md" } }