Add qx-labs/agents-deep-research

agents/qx-labs__agents-deep-research/README.md

@@ -0,0 +1,63 @@
+# Agents Deep Research
+
+**Author:** [QX Labs](https://github.com/qx-labs) · **License:** Apache-2.0
+
+A powerful, autonomous deep-research assistant built on the [OpenAI Agents SDK](https://github.com/openai/openai-agents-python). Point it at any topic and it produces a thorough, well-referenced report — without asking clarifying questions.
+
+## What it does
+
+The agent runs a coordinated multi-agent pipeline that iteratively closes knowledge gaps until the research is complete:
+
+1. **Knowledge Gap Agent** — analyses what is already known and decides what still needs investigation
+2. **Tool Selector Agent** — picks the right tool(s) for each gap (web search, site crawler, or custom tools)
+3. **Tool Agents** (parallel) — execute the selected research actions concurrently
+4. **Observations / Thinking Agent** — reflects on findings and steers the next iteration
+5. **Writer Agent** — synthesises all findings into a coherent, cited report
+
+## Two research modes
+
+| Mode | Best for | Class |
+|---|---|---|
+| **IterativeResearcher** | Shorter reports (≤ ~1 000 words / 5 pages) | `IterativeResearcher` |
+| **DeepResearcher** | Long-form structured reports (20+ pages) | `DeepResearcher` |
+
+The `DeepResearcher` first drafts a full report outline (Planner Agent), then runs parallel `IterativeResearcher` instances per section, and finally proofreads the combined result.
+
+## Example usage
+
+```python
+import asyncio
+from deep_researcher import DeepResearcher
+
+researcher = DeepResearcher(max_iterations=3, max_time_minutes=10)
+report = asyncio.run(researcher.run("Overview of quantum computing"))
+print(report)
+```
+
+Or via CLI:
+
+```bash
+pip install deep-researcher
+deep-researcher --mode deep \
+  --query "Comprehensive overview of quantum computing" \
+  --max-iterations 3 --max-time 10 --verbose
+```
+
+## Compatible models
+
+Works with any OpenAI-API-compatible provider: **OpenAI, Anthropic, Gemini, DeepSeek, Perplexity, Azure OpenAI, OpenRouter, Hugging Face, Ollama, LM Studio**.
+
+## Key features
+
+- 🔄 Fully autonomous iterative research loop
+- 🔀 Parallel section research for long-form reports
+- 🔍 Web search (Serper / OpenAI native) + site crawler built-in
+- 🔧 Extensible with custom tool agents
+- 📊 Optional OpenAI trace monitoring
+- 🤖 Model-agnostic (any OpenAI-spec provider)
+
+## Sample outputs
+
+- [Life and Works of Plato](https://github.com/qx-labs/agents-deep-research/blob/main/examples/sample_output/plato.md) — 7,980 words
+- [Textbook on Quantum Computing](https://github.com/qx-labs/agents-deep-research/blob/main/examples/sample_output/quantum_computing.md) — 5,253 words
+- [Deep-Dive on Tesla](https://github.com/qx-labs/agents-deep-research/blob/main/examples/sample_output/tesla.md) — 4,732 words

agents/qx-labs__agents-deep-research/metadata.json

@@ -0,0 +1,15 @@
+{
+  "name": "agents-deep-research",
+  "author": "qx-labs",
+  "description": "Agentic deep-research assistant using multi-agent iteration: gap analysis, tool selection, web search & crawl, and report synthesis.",
+  "repository": "https://github.com/qx-labs/agents-deep-research",
+  "path": "",
+  "version": "1.0.0",
+  "category": "research",
+  "tags": ["research", "deep-research", "multi-agent", "web-search", "openai-agents-sdk", "report-generation", "iterative"],
+  "license": "Apache-2.0",
+  "model": "gpt-4o",
+  "adapters": ["openai", "system-prompt"],
+  "icon": false,
+  "banner": false
+}