{"id":35189821,"url":"https://github.com/eesb99/universal-ideation-v3","last_synced_at":"2026-01-13T21:04:41.817Z","repository":{"id":331045231,"uuid":"1124455216","full_name":"eesb99/universal-ideation-v3","owner":"eesb99","description":"Science-grounded autonomous ideation system with Triple Generator, DARLING learning, 8-dimension scoring, and NovAScore atomic novelty","archived":false,"fork":false,"pushed_at":"2026-01-01T05:50:28.000Z","size":1262,"stargazers_count":3,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-01-02T11:39:13.457Z","etag":null,"topics":["ai-ideation","autonomous-agents","brainstorming","claude","claude-code","creativity","innovation","llm","machine-learning","python"],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/eesb99.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":null,"funding":null,"license":"LICENSE","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":"2025-12-29T03:50:33.000Z","updated_at":"2026-01-01T05:50:31.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/eesb99/universal-ideation-v3","commit_stats":null,"previous_names":["eesb99/universal-ideation-v3"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/eesb99/universal-ideation-v3","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/eesb99%2Funiversal-ideation-v3","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/eesb99%2Funiversal-ideation-v3/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/eesb99%2Funiversal-ideation-v3/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/eesb99%2Funiversal-ideation-v3/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/eesb99","download_url":"https://codeload.github.com/eesb99/universal-ideation-v3/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/eesb99%2Funiversal-ideation-v3/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":28400966,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-01-13T14:36:09.778Z","status":"ssl_error","status_checked_at":"2026-01-13T14:35:19.697Z","response_time":56,"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":["ai-ideation","autonomous-agents","brainstorming","claude","claude-code","creativity","innovation","llm","machine-learning","python"],"created_at":"2025-12-29T05:35:29.255Z","updated_at":"2026-01-13T21:04:41.812Z","avatar_url":"https://github.com/eesb99.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Universal Ideation v3.2\n\nScience-grounded autonomous ideation system with self-improving mechanisms.\n\n## Features\n\n- **Triple Generator** - Explorer, Refiner, Contrarian modes\n- **8-Dimension Scoring** - Novelty, Feasibility, Market, Complexity, Scenario, Contrarian, Surprise, Cross-Domain\n- **DARLING Learning** - Diversity-aware reward calculation\n- **Atomic Novelty (NovAScore)** - 0.94 accuracy claim-level novelty detection\n- **Verification Gates** - Quality checkpoints\n- **Reflection Learning** - Self-improving pattern extraction\n- **Plateau Escape** - Avoid local optima\n- **Web Search Integration** - Real-time market intelligence via Perplexity API\n- **Interview Context Integration** - Enriched ideation using universal-interview skill\n- **Batch Mode** - Generate multiple ideas per API call (5x faster)\n\n## Architecture\n\n![Universal Ideation v3.2 Architecture](docs/architecture.png)\n\n## Installation\n\n### As Claude Code Skill\n\n1. Copy folder to `~/.claude/skills/universal-ideation-v3/`\n2. Install dependencies:\n   ```bash\n   pip install -r requirements.txt\n   ```\n3. Invoke with:\n   ```bash\n   /universal-ideation-v3 \"your domain\"\n   ```\n\n### Optional: Vector Search\n\nFor semantic similarity features:\n```bash\npip install qdrant-client sentence-transformers\ndocker run -p 6333:6333 qdrant/qdrant\n```\n\n## Quick Start\n\n### Via Skill Command\n```bash\n/universal-ideation-v3 \"e-commerce innovation\"\n/universal-ideation-v3 \"sustainable opportunity in Malaysia\"\n/universal-ideation-v3 \"ai start up innovation\"\n/universal-ideation-v3 \"sustainable packaging innovation\"\n```\n\n### Via CLI (Stub Mode)\n```bash\ncd ~/.claude/skills/universal-ideation-v3\npython3 scripts/run_v3.py \"your domain\" --verbose\n```\n\n### Via LLM Runner (Full Mode with Claude API)\n\nRequires Anthropic API key in `~/.env`:\n```bash\nCLAUDE_API_KEY=sk-ant-xxxxx\n```\n\nOptional: Add Perplexity API key for web search:\n```bash\nPERPLEXITY_API_KEY=pplx-xxxxx\n```\n\nRun with full LLM integration + storage:\n```bash\ncd ~/.claude/skills/universal-ideation-v3\npython3 scripts/llm_runner.py \"your domain\" -i 30 -m 30 -v\n```\n\nThis mode:\n- Uses Claude API for idea generation and scoring\n- Stores ideas in SQLite database\n- Stores embeddings in Qdrant for semantic search\n- Exports full v3.2 statistics (DARLING learnings, atomic novelty, etc.)\n\n### With Web Search (Market Intelligence)\n\nEnable real-time market context from Perplexity:\n```bash\npython3 scripts/llm_runner.py \"protein beverages\" -i 10 -m 5 -w -v\n```\n\nThis fetches:\n- **Domain trends** - Current market developments\n- **Market gaps** - Underserved opportunities\n- **Emerging tech** - Relevant technologies\n- **Consumer insights** - Behavior patterns\n\nThe market context is injected into idea generation prompts for more grounded, trend-aware ideas.\n\n### With Interview Context (Deep Domain Understanding)\n\nEnrich ideation with structured interview context from the universal-interview skill:\n\n```bash\n# First, run an interview to build context\ncd ~/.claude/skills/universal-interview\npython3 scripts/interview_runner.py \"sustainable protein beverages\"\n\n# Then use the context in ideation\ncd ~/.claude/skills/universal-ideation-v3\npython3 scripts/llm_runner.py \"protein beverages\" --context-id [ID] -b -n 10 -v\n\n# Or use interactive context selection\npython3 scripts/llm_runner.py \"protein beverages\" -c -b -n 10 -v\n```\n\nThe interview context provides:\n- **Problem Space** - Pain points and user needs\n- **Constraints** - Budget, timeline, regulations\n- **Assumptions** - Hidden beliefs to challenge\n- **Intent** - Strategic goals\n- **Preferences** - What excites vs. bores\n- **Existing Solutions** - Competitors and gaps\n- **Resources** - Assets and capabilities\n\nBoth skills share the database at `~/.claude/data/ideation.db`.\n\n### Batch Mode (5x Faster)\n\nGenerate multiple ideas per API call:\n\n```bash\n# Generate 100 ideas in batch mode\npython3 scripts/llm_runner.py \"your domain\" -b -n 100 -s 10 -v\n\n# With web search\npython3 scripts/llm_runner.py \"your domain\" -b -n 100 -w -v\n\n# With interview context\npython3 scripts/llm_runner.py \"your domain\" -b -n 100 --context-id [ID] -v\n```\n\n| Mode | Ideas/min | 100 ideas |\n|------|-----------|-----------|\n| Standard | ~1 | ~100 min |\n| Batch | ~5 | ~20 min |\n\n## Options\n\n### llm_runner.py (Full Mode)\n\n| Flag | Default | Description |\n|------|---------|-------------|\n| `-i, --iterations` | 15 | Max iterations (standard mode) |\n| `-m, --minutes` | 15 | Max duration (standard mode) |\n| `-t, --threshold` | 60.0 | Acceptance score |\n| `-v, --verbose` | false | Show progress |\n| `-w, --web-search` | false | Enable Perplexity web search |\n| `-b, --batch` | false | Enable batch mode |\n| `-n, --target` | 100 | Target ideas (batch mode) |\n| `-s, --batch-size` | 10 | Ideas per API call (batch mode) |\n| `-c, --context` | false | Interactive interview context selection |\n| `--context-id` | - | Specific interview context ID |\n\n### run_v3.py (v3.3 - LLM Mode)\n\n| Flag | Default | Description |\n|------|---------|-------------|\n| `-i, --iterations` | 30 | Max iterations |\n| `-m, --minutes` | 30 | Max duration |\n| `-t, --threshold` | 65.0 | Acceptance score |\n| `-v, --verbose` | false | Show progress |\n| `--use-llm` | false | Enable Anthropic API for real idea generation |\n| `--llm-model` | claude-sonnet-4-20250514 | Model to use for generation |\n| `--initiative` | - | Load enriched_domain from interview database by ID |\n| `--test` | false | Stub mode for testing |\n\n#### With Interview Integration (v3.3)\n\nLoad context directly from interview database:\n\n```bash\n# Run ideation with interview context\npython3 scripts/run_v3.py --initiative e87f16de-61d0-427c-aa21-2cc7b6d3274d --use-llm -v\n\n# Or with manual domain\npython3 scripts/run_v3.py \"your domain\" --use-llm -v\n```\n\nThe `--initiative` flag queries `~/.claude/data/ideation.db` for the enriched_domain generated by universal-interview.\n\n## Storage\n\n### SQLite Database\n- Location: `~/.claude/data/ideation.db` (shared with universal-interview)\n- Tables: ideas, sessions, learnings, initiatives, interview_responses\n- Persists all accepted ideas with scores\n\n### Qdrant Vector Database (Optional)\n- Location: `localhost:6333`\n- Collection: `universal_ideas` (384-dim embeddings)\n- Enables semantic similarity search\n\nStart Qdrant:\n```bash\ndocker run -d -p 6333:6333 qdrant/qdrant\n```\n\n## Output\n\nResults saved to:\n- `output/ideation_YYYYMMDD_HHMMSS.json` - Full session export\n- `data/ideation.db` - SQLite persistence\n- Qdrant vectors - Semantic embeddings (if enabled)\n\n## Backup\n\nBackup your ideas database (SQLite + Qdrant vectors):\n\n```bash\ncd ~/.claude/skills/universal-ideation-v3\n\n# Full backup (SQLite + Qdrant + JSON)\npython3 scripts/backup.py backup\n\n# Named backup\npython3 scripts/backup.py backup -n \"my_backup\"\n\n# SQLite only (skip Qdrant)\npython3 scripts/backup.py backup --no-qdrant\n\n# View statistics\npython3 scripts/backup.py stats\n\n# List all backups\npython3 scripts/backup.py list\n\n# Export all ideas to JSON\npython3 scripts/backup.py export -o my_ideas.json\n\n# Restore full backup (SQLite + Qdrant)\npython3 scripts/backup.py restore backup_file.db\n\n# Restore SQLite only\npython3 scripts/backup.py restore backup_file.db --no-qdrant\n```\n\n### Backup Files\n\n| File | Contents |\n|------|----------|\n| `*_backup.db` | SQLite database (ideas, sessions, learnings) |\n| `*_qdrant.snapshot` | Qdrant vector embeddings |\n| `*_backup.json` | Full JSON export |\n\nBackups saved to `backups/` folder.\n\n## Testing\n\n```bash\npython -m pytest tests/ -v\n```\n\n## Structure\n\n```\nuniversal-ideation-v3/\n├── SKILL.md              # Skill definition\n├── README.md             # This file\n├── requirements.txt      # Dependencies\n├── scripts/\n│   ├── run_v3.py        # Main orchestrator (stub mode)\n│   ├── llm_runner.py    # LLM-integrated runner (full mode)\n│   ├── backup.py        # Database backup tool\n│   ├── generators/      # Triple Generator\n│   ├── gates/           # Quality gates\n│   ├── evaluators/      # Cognitive diversity\n│   ├── learning/        # DARLING + reflection\n│   ├── escape/          # Plateau escape\n│   ├── novelty/         # Atomic novelty\n│   ├── search/          # Web search (Perplexity)\n│   └── storage/         # Persistence (SQLite + Qdrant)\n├── tests/               # 74 unit tests\n├── data/                # Runtime SQLite\n├── backups/             # Database backups\n└── output/              # Generated results\n```\n\n## License\n\nMIT\n\n## Version\n\nv3.3 (2026-01-01) - LLM integration for run_v3.py, --initiative flag for interview database integration\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Feesb99%2Funiversal-ideation-v3","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Feesb99%2Funiversal-ideation-v3","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Feesb99%2Funiversal-ideation-v3/lists"}