{"id":30397832,"url":"https://github.com/duckboss/client-tools-assistant","last_synced_at":"2025-08-21T14:12:47.791Z","repository":{"id":297744711,"uuid":"997770281","full_name":"DuckBoss/client-tools-assistant","owner":"DuckBoss","description":"A local RAG-based AI assistant to retrieve team workflows and processes information from locally stored documents","archived":false,"fork":false,"pushed_at":"2025-06-07T06:57:47.000Z","size":34,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-06-07T07:26:52.804Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","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/DuckBoss.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}},"created_at":"2025-06-07T06:25:23.000Z","updated_at":"2025-06-07T06:57:51.000Z","dependencies_parsed_at":"2025-06-07T07:36:56.122Z","dependency_job_id":null,"html_url":"https://github.com/DuckBoss/client-tools-assistant","commit_stats":null,"previous_names":["duckboss/client-tools-assistant"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/DuckBoss/client-tools-assistant","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DuckBoss%2Fclient-tools-assistant","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DuckBoss%2Fclient-tools-assistant/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DuckBoss%2Fclient-tools-assistant/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DuckBoss%2Fclient-tools-assistant/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/DuckBoss","download_url":"https://codeload.github.com/DuckBoss/client-tools-assistant/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DuckBoss%2Fclient-tools-assistant/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":271493232,"owners_count":24769117,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-08-21T02:00:08.990Z","response_time":74,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"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":[],"created_at":"2025-08-21T14:12:46.232Z","updated_at":"2025-08-21T14:12:47.776Z","avatar_url":"https://github.com/DuckBoss.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Client Tools Assistant (CTA)\n\nA local RAG-based AI assistant to retrieve team workflows and processes information from locally stored documents.\n\n## ⚠️ Disclaimer\n\n- This **DOES NOT** contain any actual data, it only has the code to use with your own data.\n- This **DOES NOT** provide any trained or finetuned models, it is using out-of-the-box IBM Granite + Nomic Embed\n- This is a generic software, it can be used as a generic rag-based LLM. There is no company data in use.\n- This is a proof-of-concept. None of the code is finalized, and it will probably change a lot.\n- Don't expect any sort of magical working code, I'm not Linus Torvalds.\n\n### 🚀 Primary Technologies Used\n\n- **Ollama**: For the local LLM interface (running locally)\n- **IBM Granite-3.3:2b Model**: For the local LLM model (can be configured)\n- **nomic-embed-text**: For the local embedding model (can be configured)\n- **ChromaDB**: For the local vector database\n- **Docling**: For the local document conversion and formatting\n- **Rich**: For the CLI user experience\n\n### 🔗 Installation and Setup\n\n1. Install required dependencies\n```bash\npip install -r requirements.txt\n```\n\n2. Create a docs folder to store your documents:\n```bash\ncd \u003cmy_repo_clone\u003e\nmkdir ./docs/\n```\n\n3. Run ollama locally:\n```bash\nollama serve\n```\n\n4. Pull required models:\n```bash\nollama pull granite3.3:2b\nollama pull nomic-embed-text:latest\n```\n\n### 🏃 Usage\n\n![demo-image](images/demo.png)\n\n- One-shot prompting: where the prompt is provided through command-line arguments\n\n```bash\n./cta.py How do I format a PR commit message? \n```\n\n- Interactive prompting: where the AI can be continuously prompted\n\n```bash\n./cta.py --interactive\n```\n\n- Prompting with streamed responses: where the AI response is generated in a stream rather than waiting for the entire response to generate\n\n```bash\n./cta.py --stream How do I format a PR commit message?\n```\n\n- Set a custom directory to pull documentation from (uses `./docs/` by default):\n\n```bash\n./cta.py --docs my/path/to/docs/\n```\n\nRefer to the cli help instructions here: `./cta.py --help`\n\n```text\nA local RAG-based AI assistant to retrieve team workflows and processes information.\n\nPositional Arguments:\n  prompt             prompt for the AI assistant\n\nOptions:\n  -h, --help         show this help message and exit\n  -s, --stream       enable streaming generated responses\n  -i, --interactive  enable interactive prompting\n  -d, --docs DOCS    path to local directory of documents to use for rag prompting\n```\n\n### Misc\n\n- You can change the system prompt, llm/embed models, and the default docs directory in the `config.py` file.\n- Supported document types:\n  - PDF\n  - DOCX\n  - XLSX\n  - HTML\n\n## 💡 Next Steps\n\n- **Code Cleanup**: This code needs to be cleaned up... a lot.\n- **Integrations**: A slack bot integration would make this much easier to use.\n- **Prompt Improvements**: This still hallucinates sometimes, improved prompt engineering may help. It's rather simple right now.\n- **Model Improvements**: There's probably a better model I could use that's within data privacy policy.\n- **Automated Document Upload**: A way to automatically hook into new/updated documents instead of manually putting documents into a folder would be helpful.\n- **Stop Generating The Vector Embeddings Every Time**: This can probably speed up a lot by storing the chromadb vector database persistently instead of regenerating it every time.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fduckboss%2Fclient-tools-assistant","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fduckboss%2Fclient-tools-assistant","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fduckboss%2Fclient-tools-assistant/lists"}