{"id":43657529,"url":"https://github.com/lekt9/openclaw-foundry","last_synced_at":"2026-02-05T22:00:54.530Z","repository":{"id":335508885,"uuid":"1145578929","full_name":"lekt9/openclaw-foundry","owner":"lekt9","description":"The forge that forges itself. Self-writing meta-extension for OpenClaw.ai","archived":false,"fork":false,"pushed_at":"2026-01-31T23:16:48.000Z","size":1792,"stargazers_count":71,"open_issues_count":0,"forks_count":8,"subscribers_count":1,"default_branch":"main","last_synced_at":"2026-02-05T07:15:32.499Z","etag":null,"topics":["agents","clawdbot","clawdbot-plugin","moltbot","openclaw","openclaw-skills","self-learning"],"latest_commit_sha":null,"homepage":"https://claw.getfoundry.app","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/lekt9.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-01-30T00:32:48.000Z","updated_at":"2026-02-05T06:57:49.000Z","dependencies_parsed_at":null,"dependency_job_id":"e295b25c-4251-4a87-90e6-7dfcb68d1c33","html_url":"https://github.com/lekt9/openclaw-foundry","commit_stats":null,"previous_names":["lekt9/openclaw-foundry"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/lekt9/openclaw-foundry","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lekt9%2Fopenclaw-foundry","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lekt9%2Fopenclaw-foundry/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lekt9%2Fopenclaw-foundry/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lekt9%2Fopenclaw-foundry/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/lekt9","download_url":"https://codeload.github.com/lekt9/openclaw-foundry/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lekt9%2Fopenclaw-foundry/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29135914,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-05T21:59:57.939Z","status":"ssl_error","status_checked_at":"2026-02-05T21:59:57.628Z","response_time":65,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agents","clawdbot","clawdbot-plugin","moltbot","openclaw","openclaw-skills","self-learning"],"created_at":"2026-02-04T21:09:01.872Z","updated_at":"2026-02-05T22:00:54.524Z","avatar_url":"https://github.com/lekt9.png","language":"TypeScript","funding_links":[],"categories":["Skills vs Plugins: What's the Difference?","Skills \u0026 Plugins","🔧 Skills \u0026 Plugins","Plugins und Integrationen","Community Plugins"],"sub_categories":["Bundled Plugins","Notable Skills \u0026 Plugins","Hacker News","Meta / Self-Improvement"],"readme":"\u003cp align=\"center\"\u003e\n  \u003cimg src=\"assets/logo.png\" alt=\"Foundry\" width=\"120\" /\u003e\n\u003c/p\u003e\n\n# Foundry\n\n**The forge that forges itself.**\n\n[![FDRY](https://img.shields.io/badge/FDRY-Solana-9945FF)](https://dexscreener.com/solana/2jc1lpgy1zjl9uertfdmtnm4kc2ahhydk4tqqqgbjdhh)\n\nFoundry is a self-writing meta-extension for [OpenClaw](https://github.com/lekt9/openclaw) that learns how you work, researches documentation, and writes new capabilities into itself. It observes your workflows, crystallizes patterns into tools, and upgrades itself to match how you operate.\n\n**$FDRY** — [dexscreener](https://dexscreener.com/solana/2jc1lpgy1zjl9uertfdmtnm4kc2ahhydk4tqqqgbjdhh) · Solana\n\n```\n┌─────────────────────────────────────────────────────────────┐\n│                         FOUNDRY                             │\n│                                                             │\n│   Observe ──► Research ──► Learn ──► Write ──► Deploy      │\n│       │          │          │          │          │         │\n│       ▼          ▼          ▼          ▼          ▼         │\n│   workflows   docs.openclaw  patterns  extensions  gateway  │\n│   tool calls  arXiv papers   insights  tools       restart  │\n│   outcomes    GitHub repos   workflows hooks       resume   │\n│                                        skills              │\n└─────────────────────────────────────────────────────────────┘\n```\n\n## OpenClaw vs Foundry\n\n**OpenClaw** (originally Clawdbot) is the platform — an open-source agent runtime with:\n- Gateway, channels, memory, sessions\n- Tool execution and skill loading\n- Model providers and routing\n- The infrastructure everything runs on\n\n**Foundry** is a plugin that runs *on* OpenClaw:\n- Observes how you work → learns your patterns\n- Researches docs → writes new extensions/skills/hooks\n- Has its own learning engine (not part of OpenClaw core)\n- Can modify itself via `foundry_extend_self`\n- Publishes to Foundry Marketplace via x402\n\n```\nOpenClaw (platform)\n├── Gateway\n├── Channels (Discord, Slack, Telegram...)\n├── Skills \u0026 Tools\n└── Plugins\n    └── Foundry (this repo)\n        ├── observes → your workflows\n        ├── researches → docs, papers, repos\n        ├── writes → extensions, skills, hooks\n        ├── learns → from outcomes\n        ├── crystallizes → patterns into tools\n        └── publishes → to marketplace\n```\n\n**Key distinction:** OpenClaw doesn't have built-in self-learning. Foundry adds that capability on top. Foundry is an \"agent that builds agents\" — it uses OpenClaw's infrastructure to create new OpenClaw capabilities, and upgrades itself to match how you work.\n\n## Why Self-Writing Matters\n\nThe key insight isn't \"LLM writes code for you\" — it's \"the system upgrades itself.\"\n\n### Knowledge vs Behavior\n\n| Knowledge (Patterns) | Behavior (Self-Written Code) |\n|---------------------|------------------------------|\n| Stored as text | Baked into the system |\n| LLM must read and apply each time | Runs automatically |\n| Uses tokens every invocation | Zero token cost |\n| Can be forgotten or ignored | Always executes |\n\nA pattern says: *\"When X happens, do Y.\"*\nSelf-written code **does** Y automatically when X happens.\n\n### Workflow Learning\n\nFoundry tracks every workflow you run:\n\n```\nGoal: \"deploy to staging\"\nTools: git → build → test → deploy\nOutcome: success\nDuration: 45s\n```\n\nOver time, patterns emerge. When a pattern hits 5+ uses with 70%+ success rate, Foundry **crystallizes** it into a dedicated tool.\n\nWhat took 8 tool calls now takes 1.\n\n### The Recursive Loop\n\n```\nFoundry observes how you work\n    ↓\nLearns patterns, researches docs\n    ↓\nWrites tool/hook to match your workflow\n    ↓\nThat code becomes part of Foundry\n    ↓\nFoundry is now better at working like you\n    ↓\nBetter Foundry learns more, writes more\n    ↓\nRepeat\n```\n\nThe system that writes the code IS the code being written.\n\n### Why This Compounds\n\n| Traditional Agents | Foundry |\n|-------------------|---------|\n| Same logic every time | Learns your patterns |\n| You adapt to the agent | Agent adapts to you |\n| Each capability is isolated | Each upgrade improves the upgrader |\n| Linear improvement | Compound improvement |\n\n**Example:**\n1. You deploy to staging 5 times using git→build→test→deploy\n2. Foundry recognizes the pattern (87% success rate)\n3. Crystallizes into `deploy_staging` tool\n4. Now \"deploy to staging\" is a single command\n5. You save time → do more deploys → pattern strengthens\n6. Foundry learns variations (deploy to prod, deploy with migrations)\n7. Loop\n\n### The Bet\n\nTraditional software: Human improves software → software does more\n\nFoundry: Software upgrades software → software upgrades faster\n\nThis is **recursive self-improvement** — each capability makes acquiring the next capability easier.\n\n## Features\n\n### Self-Writing Code Generation\n- Writes OpenClaw extensions with tools and hooks\n- Generates API skills following AgentSkills format with YAML frontmatter\n- Generates browser automation skills with CDP integration\n- Generates standalone hooks with HOOK.md + handler.ts pattern\n- Can extend itself with new capabilities\n- Validates code in isolated sandbox before deployment\n\n### Workflow Learning \u0026 Crystallization\n- Tracks goal → tool sequence → outcome for every workflow\n- Extracts keywords from goals for pattern matching\n- Calculates success rates and average durations\n- Crystallizes high-value patterns (5+ uses, 70%+ success) into dedicated tools\n- Suggests relevant patterns when you start similar tasks\n\n### The Overseer\n- Runs autonomously on hourly interval\n- Identifies crystallization candidates\n- Auto-generates tools from high-value patterns\n- Prunes stale patterns (30+ days unused)\n- Tracks tool performance metrics (ADAS-style evolution)\n- Reports actions taken\n\n### Native OpenClaw Integration\n- **AgentSkills Format**: Proper YAML frontmatter with metadata (emoji, requires, events)\n- **Browser Automation**: CDP-based browser tool integration for authenticated workflows\n- **Skill Gating**: Auto-generates requires.config, requires.bins, requires.env for dependencies\n- **Hook System**: Full support for OpenClaw hook events (gateway:startup, command:new, etc.)\n- **ClawdHub Ready**: Skills can be published to the ClawdHub registry\n\n### Proactive Learning\n- Records tool outcomes (success/failure) with context\n- Builds patterns from repeated workflows\n- Shares learnings via the Foundry Marketplace\n- Injects relevant context into agent conversations\n\n### Sandbox Validation\n- Runs generated code in isolated Node process\n- Catches runtime errors before they crash the gateway\n- Static security scanning (blocks shell exec, eval, credential access)\n- Only deploys code that passes all checks\n\n### Restart Resume\n- Saves conversation context before gateway restart\n- Automatically resumes after restart via managed hook\n- No lost work when self-modifying\n\n## Installation\n\n```bash\nopenclaw plugins install @getfoundry/foundry-openclaw\n```\n\nThat's it. This will download, extract, enable, and load Foundry automatically.\n\n---\n\n### Alternative: Manual Config\n\nAdd to `~/.openclaw/openclaw.json`:\n```json\n{\n  \"plugins\": {\n    \"entries\": {\n      \"foundry\": { \"enabled\": true }\n    }\n  }\n}\n```\n\nThen restart:\n```bash\nopenclaw gateway restart\n```\n\n### Option C: GitHub Source\n\nAdd to `~/.openclaw/openclaw.json`:\n\n```json\n{\n  \"plugins\": {\n    \"entries\": {\n      \"foundry\": {\n        \"enabled\": true,\n        \"source\": \"github:lekt9/openclaw-foundry\"\n      }\n    }\n  }\n}\n```\n\n### Option D: Nix (Reproducible)\n\n```bash\nnix run github:lekt9/openclaw-foundry\n```\n\n### Option E: Manual Clone\n\n```bash\ngit clone https://github.com/lekt9/openclaw-foundry ~/.openclaw/extensions/foundry\ncd ~/.openclaw/extensions/foundry \u0026\u0026 npm install\n```\n\nThen restart:\n```bash\nopenclaw gateway restart\n```\n\n### Configuration\n\nFull config options:\n\n```json\n{\n  \"plugins\": {\n    \"entries\": {\n      \"foundry\": {\n        \"enabled\": true,\n        \"source\": \"github:lekt9/openclaw-foundry\",\n        \"config\": {\n          \"autoLearn\": true,\n          \"sources\": {\n            \"docs\": true,\n            \"experience\": true,\n            \"arxiv\": true,\n            \"github\": true\n          },\n          \"marketplace\": {\n            \"autoPublish\": false\n          }\n        }\n      }\n    }\n  }\n}\n```\n\n| Option | Default | Description |\n|--------|---------|-------------|\n| `autoLearn` | `true` | Learn from agent activity automatically |\n| `sources.docs` | `true` | Learn from OpenClaw documentation |\n| `sources.experience` | `true` | Learn from own successes/failures |\n| `marketplace.autoPublish` | `false` | Auto-publish high-value patterns |\n\n## Tools\n\n### Research \u0026 Learning\n\n| Tool | Description |\n|------|-------------|\n| `foundry_research` | Search docs.openclaw.ai for best practices and patterns |\n| `foundry_docs` | Read specific documentation pages (plugin, hooks, tools, etc.) |\n| `foundry_learnings` | View recorded patterns, workflows, insights |\n\n### Code Generation\n\n| Tool | Description |\n|------|-------------|\n| `foundry_implement` | Research + implement a capability end-to-end |\n| `foundry_write_extension` | Write a new OpenClaw extension with tools/hooks |\n| `foundry_write_skill` | Write an API skill package (SKILL.md + api.ts) |\n| `foundry_write_browser_skill` | Write a browser automation skill with CDP integration |\n| `foundry_write_hook` | Write a standalone hook (HOOK.md + handler.ts) |\n| `foundry_add_tool` | Add a tool to an existing extension |\n| `foundry_add_hook` | Add a hook to an existing extension |\n| `foundry_extend_self` | Add capabilities to Foundry itself |\n\n### Management\n\n| Tool | Description |\n|------|-------------|\n| `foundry_list` | List all written extensions and skills |\n| `foundry_restart` | Restart gateway with context preservation |\n| `foundry_publish_ability` | Publish patterns/extensions to Foundry Marketplace |\n| `foundry_marketplace` | Search, browse leaderboard, and install abilities |\n\n## Bundled Skills\n\nFoundry ships with built-in skills that are automatically available:\n\n### `foundry-browser-helper`\nHelper skill for browser automation patterns. Provides guidance on using the OpenClaw `browser` tool effectively.\n\n```\n# Quick reference\nbrowser open https://example.com\nbrowser snapshot           # AI-readable format\nbrowser click ref=btn_submit\nbrowser type ref=input_email \"user@example.com\"\n```\n\n## How It Works\n\n### 1. Observe Phase\n```\nFoundry watches every workflow:\n  - Goal: What the user is trying to do\n  - Tools: Sequence of tool calls\n  - Outcome: Success, failure, or partial\n  - Duration: How long it took\n```\n\n### 2. Research Phase\n```\nUser: \"Add a tool that fetches weather data\"\n\nFoundry:\n  1. Searches docs.openclaw.ai for tool registration patterns\n  2. Finds examples of API-calling tools\n  3. Identifies best practices for error handling\n```\n\n### 3. Learn Phase\n```\nFoundry:\n  1. Records workflow patterns\n  2. Tracks success rates per pattern\n  3. Identifies crystallization candidates\n  4. Builds knowledge base of what works\n```\n\n### 4. Write Phase\n```\nFoundry:\n  1. Generates extension code following patterns\n  2. Includes proper TypeScript types\n  3. Adds error handling and logging\n  4. Validates in isolated sandbox\n```\n\n### 5. Deploy Phase\n```\nFoundry:\n  1. Writes to ~/.openclaw/extensions/\n  2. Creates openclaw.plugin.json\n  3. Triggers gateway restart\n  4. Resumes conversation automatically\n```\n\n## Skill Generation\n\nFoundry generates skills in the AgentSkills format with proper YAML frontmatter:\n\n```yaml\n---\nname: my-api-skill\ndescription: Integrates with My API service\nmetadata: {\"openclaw\":{\"emoji\":\"🔌\",\"requires\":{\"env\":[\"MY_API_KEY\"]}}}\n---\n\n# My API Skill\n\n## Authentication\nThis skill requires the `MY_API_KEY` environment variable.\n\n## Endpoints\n- `GET /users` - List all users\n- `POST /users` - Create a new user\n```\n\n### Browser Skills\n\nBrowser automation skills automatically gate on `browser.enabled`:\n\n```yaml\n---\nname: my-browser-skill\ndescription: Automates login workflow\nmetadata: {\"openclaw\":{\"emoji\":\"🌐\",\"requires\":{\"config\":[\"browser.enabled\"]}}}\n---\n\n# My Browser Skill\n\n## Workflow\n1. Open login page\n2. Fill credentials\n3. Submit form\n4. Verify success\n```\n\n### Standalone Hooks\n\nHooks follow the HOOK.md + handler.ts pattern:\n\n```\nmy-hook/\n├── HOOK.md          # Frontmatter + documentation\n└── handler.ts       # Event handler code\n```\n\n## Sandbox Security\n\nGenerated code is validated before deployment:\n\n### Blocked Patterns (Instant Reject)\n- `child_process` / `exec` / `spawn` — Shell execution\n- `eval()` / `new Function()` — Dynamic code execution\n- `~/.ssh/` / `id_rsa` — SSH key access\n- `~/.aws/` / `aws_secret` — Cloud credentials\n- Exfiltration domains (ngrok, webhook.site, etc.)\n\n### Flagged Patterns (Warning)\n- `process.env` — Environment variable access\n- `fs.readFile` / `fs.writeFile` — Filesystem access\n- Base64 encoding — Potential obfuscation\n\n### Runtime Validation\n```\n1. Write extension to temp directory\n2. Spawn isolated Node process with tsx\n3. Mock OpenClaw API\n4. Try to import and run register()\n5. If fails → reject with error message\n6. If passes → deploy to real extensions directory\n```\n\n## Foundry Marketplace\n\nPublish and download abilities with x402 Solana USDC payments:\n\n```bash\n# Publish a workflow pattern you discovered\nfoundry_publish_ability type=\"pattern\" name=\"Deploy Staging\" patternId=\"wp_123\"\n\n# Search for existing patterns\nfoundry_marketplace action=\"search\" query=\"deploy\" type=\"pattern\"\n\n# See the leaderboard\nfoundry_marketplace action=\"leaderboard\"\n\n# Download and apply\nfoundry_marketplace action=\"install\" id=\"abc123\"\n```\n\n### x402 Protocol\n\nHTTP 402 \"Payment Required\" + Solana USDC:\n\n1. Request a skill download\n2. Server returns 402 with payment requirements\n3. Sign USDC transaction with your wallet\n4. Retry with signed transaction in header\n5. Receive the skill\n\nNo intermediaries. Direct creator payment. Network effects compound.\n\n### Ability Types \u0026 Pricing\n\n| Type | Price | Description |\n|------|-------|-------------|\n| Pattern | FREE | Workflow patterns (crowdsourced) |\n| Technique | $0.02 | Reusable code snippets |\n| Extension | $0.05 | Full OpenClaw plugins |\n| Agent | $0.10 | High-fitness agent designs |\n\n## Configuration\n\n```json\n{\n  \"plugins\": {\n    \"entries\": {\n      \"foundry\": {\n        \"enabled\": true,\n        \"config\": {\n          \"dataDir\": \"~/.openclaw/foundry\",\n          \"openclawPath\": \"/path/to/openclaw\",\n          \"autoLearn\": true,\n          \"sources\": {\n            \"docs\": true,\n            \"experience\": true,\n            \"arxiv\": false,\n            \"github\": false\n          },\n          \"marketplace\": {\n            \"url\": \"https://api.claw.getfoundry.app\",\n            \"autoPublish\": false\n          }\n        }\n      }\n    }\n  }\n}\n```\n\n### Config Options\n\n| Option | Description | Default |\n|--------|-------------|---------|\n| `dataDir` | Directory to store forged artifacts | `~/.openclaw/foundry` |\n| `openclawPath` | Path to OpenClaw installation for local docs | - |\n| `autoLearn` | Automatically learn from agent activity | `true` |\n| `sources.docs` | Learn from OpenClaw documentation | `true` |\n| `sources.experience` | Learn from own successes/failures | `true` |\n| `sources.arxiv` | Learn from arXiv papers | `true` |\n| `sources.github` | Learn from GitHub repos | `true` |\n| `marketplace.url` | Foundry marketplace URL | `https://api.claw.getfoundry.app` |\n| `marketplace.autoPublish` | Auto-publish high-value patterns | `false` |\n\n## Research Foundations\n\nFoundry's self-improvement mechanisms draw from recent advances in autonomous learning agents:\n\n### Self-Improving Code Agents\n\n| Paper | Key Insight | Foundry Application |\n|-------|-------------|---------------------|\n| [Self-Improving Coding Agent](https://arxiv.org/abs/2504.15228) (Robeyns et al., 2025) | Agent systems with coding tools can autonomously edit themselves, achieving 17-53% improvement through \"non-gradient learning via LLM reflection and code updates\" | `foundry_extend_self` — the agent modifies its own codebase |\n| [From Language Models to Practical Self-Improving Computer Agents](https://arxiv.org/abs/2404.11964) (Shinn et al., 2024) | LLM agents can \"systematically generate software to augment themselves\" starting from minimal capabilities | Self-written tools/hooks that expand Foundry's capabilities |\n| [SelfEvolve](https://arxiv.org/abs/2306.02907) (Jiang et al., 2023) | Two-step pipeline: knowledge generation + self-reflection debugging using interpreter feedback | LearningEngine records outcomes → patterns → crystallization |\n\n### Recursive Introspection\n\n| Paper | Key Insight | Foundry Application |\n|-------|-------------|---------------------|\n| [RISE: Recursive Introspection](https://arxiv.org/abs/2407.18219) (Qu et al., 2024) | Iterative fine-tuning teaches models to \"alter responses after unsuccessful attempts\" via multi-turn MDPs | Workflow tracking learns from outcomes, suggests improvements |\n| [HexMachina](https://arxiv.org/abs/2506.04651) (Liu et al., 2025) | \"Artifact-centric continual learning\" — separates discovery from strategy evolution through code refinement | Patterns (knowledge) crystallize into hooks/tools (behavior) |\n\n### Meta-Agent Search\n\n| Paper | Key Insight | Foundry Application |\n|-------|-------------|---------------------|\n| [ADAS: Automated Design of Agentic Systems](https://arxiv.org/abs/2408.08435) (Hu et al., 2024) | Meta-agent iteratively discovers improved agent designs through archive-based evolution | Overseer tracks tool fitness, evolves patterns |\n\n### Core Principle\n\n\u003e \"An agent system, equipped with basic coding tools, can autonomously edit itself, and thereby improve its performance\" — Robeyns et al.\n\nFoundry operationalizes this: the system that writes the code IS the code being written.\n\n## Key Directories\n\n```\n~/.openclaw/foundry/            — Data directory\n  ├── workflows.json            — Recorded workflows\n  ├── workflow-patterns.json    — Crystallization candidates\n  ├── learnings.json            — Patterns, insights, outcomes\n~/.openclaw/extensions/         — Generated extensions go here\n~/.openclaw/skills/             — Generated skills go here\n~/.openclaw/hooks/foundry-resume/ — Restart resume hook\n```\n\n## Development\n\n```bash\n# Type check\nnpx tsc --noEmit\n\n# Test extension locally\nopenclaw gateway restart\ntail -f ~/.openclaw/logs/gateway.log | grep foundry\n```\n\n## License\n\nMIT\n\n---\n\n*Built with OpenClaw. Forged by Foundry.*\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flekt9%2Fopenclaw-foundry","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Flekt9%2Fopenclaw-foundry","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flekt9%2Fopenclaw-foundry/lists"}