From 8f1451773e4912b942c518d5fb40ec81c4f71d7c Mon Sep 17 00:00:00 2001
From: GAP Promoter <bot@gitagent.sh>
Date: Sat, 30 May 2026 18:56:10 +0000
Subject: [PATCH] Add Cranot/deep-research to the registry

---
 agents/Cranot__deep-research/README.md     | 58 ++++++++++++++++++++++
 agents/Cranot__deep-research/metadata.json | 28 +++++++++++
 2 files changed, 86 insertions(+)
 create mode 100644 agents/Cranot__deep-research/README.md
 create mode 100644 agents/Cranot__deep-research/metadata.json

diff --git a/agents/Cranot__deep-research/README.md b/agents/Cranot__deep-research/README.md
new file mode 100644
index 0000000..4fbeb89
--- /dev/null
+++ b/agents/Cranot__deep-research/README.md
@@ -0,0 +1,58 @@
+# Deep Research
+
+**Fractal exploration of any question through recursive AI agents.**
+
+Deep Research is a multi-agent system that takes any question and transforms it into a comprehensive, synthesized report by recursively decomposing it. An orchestrator agent breaks the question into angles, each angle spawns sub-agents to investigate further, and everything synthesizes back up the tree into one authoritative answer.
+
+## What it does
+
+- **Recursive decomposition** — complex questions are broken into sub-questions automatically, to any depth
+- **Parallel exploration** — all sub-agents run concurrently for fast results
+- **Multi-model ensemble support** — use Claude, Gemini, Kimi, Grok, or Azure OpenAI at different tree levels
+- **Four research strategies**: recursive, Socratic, perspective expansion, and web-grounded
+- **Persistent checkpoints** — resume interrupted research from where it left off
+- **Typed knowledge graphs** — every finding carries epistemic state (Unknown → Explored → Validated → Synthesized)
+
+## Example usage
+
+```bash
+# Basic research (Claude opus orchestrator + haiku researchers)
+deep-research research "What makes startups successful?"
+
+# Socratic mode — improve the question first
+deep-research socratic "How do I become more productive?"
+
+# Perspective expansion
+deep-research perspectives "What makes good leadership?"
+
+# Grounded research with web search
+deep-research research --strategy grounded_research "Current state of AI agents"
+
+# Legacy bash (still works)
+./deep-research.sh -m opus -r haiku -d 2 "Why do smart people make bad decisions?"
+```
+
+## Output
+
+Reports save to `reports/YYYY-MM-DD-question-slug/SYNTHESIS.md` with individual agent outputs in `agents/`.
+
+## Models supported
+
+| Provider | Models |
+|---|---|
+| Claude | opus, sonnet, haiku |
+| Gemini | flash, pro |
+| OpenRouter | grok, grok-3, grok-4 |
+| Kimi | kimi (free, best merger) |
+| Azure OpenAI | gpt5-mini |
+
+## Skills
+
+| Skill | Description |
+|---|---|
+| `recursive_research` | Decompose → spawn parallel researchers → synthesize |
+| `socratic` | Challenge assumptions, surface better questions |
+| `perspective_expander` | Multi-angle analysis with blind spot detection |
+| `grounded_research` | Web-verified, fact-checked answers |
+
+Created by [Dimitris Mitsos](https://github.com/Cranot) & [AgentsKB.com](https://agentskb.com). MIT licensed.
diff --git a/agents/Cranot__deep-research/metadata.json b/agents/Cranot__deep-research/metadata.json
new file mode 100644
index 0000000..5c06b5c
--- /dev/null
+++ b/agents/Cranot__deep-research/metadata.json
@@ -0,0 +1,28 @@
+{
+  "name": "deep-research",
+  "author": "Cranot",
+  "description": "Fractal research agent: decomposes any question into recursive sub-questions, explores all angles in parallel with multiple LLMs, and synthesizes back into a comprehensive report.",
+  "repository": "https://github.com/Cranot/deep-research",
+  "version": "1.0.0",
+  "category": "research",
+  "tags": [
+    "multi-agent",
+    "recursive",
+    "research",
+    "synthesis",
+    "claude",
+    "gemini",
+    "web-search",
+    "knowledge-graph",
+    "parallel-agents",
+    "fractal"
+  ],
+  "license": "MIT",
+  "model": "claude-opus-4-5",
+  "adapters": [
+    "claude-code",
+    "system-prompt"
+  ],
+  "icon": false,
+  "banner": false
+}