{"id":39233357,"url":"https://github.com/mmacy/tech-doc-workflow","last_synced_at":"2026-01-17T23:43:38.460Z","repository":{"id":302564333,"uuid":"1012620378","full_name":"mmacy/tech-doc-workflow","owner":"mmacy","description":"React web app that writes technical docs for your project using the evaluator-optimizer pattern, an LLM-based agent-like looping workflow.","archived":false,"fork":false,"pushed_at":"2025-10-07T18:34:53.000Z","size":1759,"stargazers_count":1,"open_issues_count":3,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-10-12T05:24:31.142Z","etag":null,"topics":["ai","documentation","evaluator-optimizer-pattern","evaluator-optimizer-workflow","llm","nodejs","react","technical-writing","typescript"],"latest_commit_sha":null,"homepage":"","language":"TypeScript","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/mmacy.png","metadata":{"files":{"readme":"README.md","changelog":null,"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}},"created_at":"2025-07-02T15:59:27.000Z","updated_at":"2025-10-07T18:34:57.000Z","dependencies_parsed_at":"2025-07-03T04:31:19.413Z","dependency_job_id":"bcb4190c-346f-4d06-b21a-a0c317d279e4","html_url":"https://github.com/mmacy/tech-doc-workflow","commit_stats":null,"previous_names":["mmacy/tech-doc-workflow"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/mmacy/tech-doc-workflow","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mmacy%2Ftech-doc-workflow","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mmacy%2Ftech-doc-workflow/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mmacy%2Ftech-doc-workflow/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mmacy%2Ftech-doc-workflow/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/mmacy","download_url":"https://codeload.github.com/mmacy/tech-doc-workflow/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/mmacy%2Ftech-doc-workflow/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":28522313,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-01-17T22:11:28.393Z","status":"ssl_error","status_checked_at":"2026-01-17T22:11:27.841Z","response_time":85,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: 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","documentation","evaluator-optimizer-pattern","evaluator-optimizer-workflow","llm","nodejs","react","technical-writing","typescript"],"created_at":"2026-01-17T23:43:38.384Z","updated_at":"2026-01-17T23:43:38.449Z","avatar_url":"https://github.com/mmacy.png","language":"TypeScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Tech Doc Workflow\n\nA React application that uses AI to automate technical documentation authoring through a multi-role [evaluator-optimizer](https://www.anthropic.com/engineering/building-effective-agents#workflow-evaluator-optimizer) workflow.\n\n*Tech Doc Workflow* implements the collaborative writing and review process by using an LLM-based agent-like workflow to replicate the roles typically seen in professional technical writing scenarios:\n\n- **Technical writer** generates and revises content\n- **Technical reviewer** verifies accuracy against source code\n- **Information architect** reviews structure and organization\n- **Technical editor** ensures grammar and style consistency\n\nThe workflow continues iteratively until all reviewers approve the document or maximum review cycles are reached.\n\n![Diagram of the draft revision and feedback flow between the technical writer, technical reviewer, information architect, and technical editor workflow roles](media/workflow-diagram.png)\n\n## Features\n\n- **Multi-role AI workflow**: Technical writer, technical reviewer, information architect, and technical editor agents working collaboratively\n- **Multiple LLM provider support**: OpenAI, Google Gemini, and Azure OpenAI\n- **Document type profiles**: Use the included example templates and guidance for how-to guides and explanation documents or create your own\n- **Iterative review process**: Automatic revision cycles with configurable maximum loops per reviewer\n- **Real-time workflow tracking**: Live status updates and detailed logging of the authoring process\n- **Source code verification**: Technical accuracy validation against optionally provided source code\n- **Downloadable outputs**: Final document in Markdown format and complete review logs\n- **Customizable settings**: Configurable reviewer guidance, style guides, and workflow parameters\n\n## Prerequisites\n\n- Node.js 18.0+\n- npm 8.0+ (bundled with Node.js)\n- Your API key from a supported provider:\n  - [OpenAI API](https://platform.openai.com/api-keys)\n  - [Google Gemini API](https://ai.google.dev/)\n  - [Azure OpenAI API](https://learn.microsoft.com/azure/ai-foundry/)\n  - Anthropic (not yet implemented)\n\n\u003e [!WARNING]\n\u003e This app can cost you real money! You are responsible for any API usage fees incurred by using this application.\n\n## Installation\n\n1. Get the code\n\n    ```shell\n    # Clone the repo and enter its directory\n    git clone https://github.com/mmacy/tech-doc-workflow.git\n    cd tech-doc-workflow\n    ```\n\n2. Install dependencies\n\n    ```shell\n    # Install the project's dependencies\n    npm install\n    ```\n\n3. Run the app\n\n    ```shell\n    # Start the development server\n    npm run dev\n    ```\n\nNavigate to the URL shown in the output, for example `http://127.0.0.1:5173`, and the *Tech Doc Workflow* interface should appear in your browser.\n\n![Web browser showing the main page of the Tech Doc Workflow app during a doc authoring and revision run. The Technical Editor role is in the Waiting state while the Technical Writer integrates the Editor's feedback.](media/workflow-ui-01-main-page.png)\n\n## Usage\n\n1. Run the application and navigate to **Settings** to configure your LLM provider API key\n2. Select a document type profile (**How-to guide** or **Explanation document**) from the dropdown\n3. Paste your draft content (draft text, notes, or existing documentation) into the **Draft content** field\n4. Optionally add authoritative **Supporting content** like source code for content generation and technical accuracy verification\n5. Select **Start authoring workflow** to begin the multi-agent review process\n6. Monitor the real-time progress as the AI roles peform their tasks of writing and reviewing the document\n7. Download the final Markdown document and review logs when the workflow completes\n\n## Settings\n\nExcept for API keys, *Tech Doc Workflow* stores its configuration settings in browser local storage and provides the following customization options. API keys are stored ONLY in memory - you must re-enter your key if you close the browser tab or window.\n\n- **Custom reviewer guidance**: Modify the instructions given to each AI reviewer role\n- **Global style guides**: Set writing and Markdown formatting standards applied across all document types\n- **Document type profiles**: Create custom templates and guidance for specialized document types\n- **Review loop limits**: Adjust the maximum number of revision cycles per reviewer role\n- **LLM provider settings**: Switch between different AI providers\n\nAccess these settings through the **Settings** page using the gear icon in the top-right corner of the main page.\n\n## Security\n\n### API keys\n\nThis application prioritizes the security of your API keys:\n\n- **Memory-only storage**: API keys are stored only in browser memory, never in local storage, session storage, or any persistent storage\n- **No network transmission**: Keys are used only for direct API calls to your chosen provider (OpenAI, Google, or Azure)\n- **Automatic clearing**: Keys are automatically cleared when you close the browser tab or window\n- **No logging**: API keys are never written to logs or exported in any downloadable files\n- **Masked display**: Keys are shown with masking (••••1234) in the user interface\n\n#### Best practices for API key security\n\n- Never share screenshots of the application showing your API keys\n- Use API keys with appropriate usage limits and monitoring enabled at the provider level\n- Regularly rotate your API keys following your organization's security policies\n- Only run this application in secure, trusted environments\n\n### Content\n\n- All document content and workflow logs remain local to your browser\n- No content is transmitted to external servers except for the necessary API calls to your chosen LLM provider\n- Downloaded files contain only the document content and workflow feedback, never API keys\n\n## Troubleshooting\n\n**\"Failed to initialize LLM provider\" error**\n\n- Verify your API key is correctly configured in Settings\n- Ensure your API key has sufficient credits/quota with the provider\n- Check that you've selected the correct provider type for your API key\n\n**Workflow is slow**\n\n- Lower the maximum review passes per reviewer in **Settings** (fewer passes may result in lower quality output)\n- Review the workflow logs to identify recurring feedback patterns and adjust reviewer prompts accordingly\n- Provide more accurate and/or clearer supporting material\n\n**Poor or incomplete final document**\n\n- Ensure your draft content provides sufficient detail for the selected document type\n- Check that supporting content (if provided) is relevant and complete\n- Review custom reviewer guidance settings for overly restrictive criteria\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmmacy%2Ftech-doc-workflow","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fmmacy%2Ftech-doc-workflow","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fmmacy%2Ftech-doc-workflow/lists"}