@bis-code/deep-think 1.1.1

package/.claude-plugin/marketplace.json

@@ -0,0 +1,18 @@
+{
+  "name": "bis-code-deep-think",
+  "description": "Structured reasoning plugins for Claude Code",
+  "owner": {
+    "name": "Ioan-Sorin Baicoianu",
+    "email": "baicoianuioansorin@gmail.com"
+  },
+  "plugins": [
+    {
+      "name": "deep-think",
+      "description": "Structured reasoning with auto-checkpointing, branching, and reflection",
+      "source": {
+        "source": "github",
+        "repo": "bis-code/mcp-deep-think"
+      }
+    }
+  ]
+}

package/.claude-plugin/plugin.json

@@ -0,0 +1,19 @@
+{
+  "name": "deep-think",
+  "version": "1.1.1",
+  "description": "Structured reasoning with auto-checkpointing, branching, and reflection",
+  "author": {
+    "name": "Ioan-Sorin Baicoianu",
+    "email": "baicoianuioansorin@gmail.com"
+  },
+  "repository": "https://github.com/bis-code/mcp-deep-think",
+  "license": "MIT",
+  "keywords": [
+    "reasoning",
+    "thinking",
+    "structured",
+    "checkpoint",
+    "reflection"
+  ],
+  "mcpServers": "./.mcp.json"
+}

package/.deep-think.json

@@ -0,0 +1,24 @@
+{
+  "thinking": {
+    "defaultStrategy": null,
+    "maxThoughts": 50,
+    "autoCheckpointEvery": 10,
+    "confidenceThreshold": 0.7
+  },
+  "practices": {
+    "rules": [],
+    "antiPatterns": [],
+    "reviewChecklist": []
+  },
+  "strategies": {
+    "custom": []
+  },
+  "reflection": {
+    "circularThreshold": 0.6,
+    "contradictionSensitivity": "medium"
+  },
+  "persistence": {
+    "directory": ".deep-think/sessions",
+    "maxCheckpoints": 10
+  }
+}

package/.mcp.json

@@ -0,0 +1,8 @@
+{
+  "mcpServers": {
+    "deep-think": {
+      "command": "node",
+      "args": ["${CLAUDE_PLUGIN_ROOT}/server/dist/bundle.mjs"]
+    }
+  }
+}

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Ioan-Sorin Baicoianu
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,92 @@
+# Deep Think
+
+A Claude Code plugin for structured reasoning. Think through complex problems with strategies, branching, reflection, and persistent checkpoints.
+
+## Install
+
+```bash
+# Add the marketplace
+/plugin marketplace add bis-code/mcp-deep-think
+
+# Install the plugin
+/plugin install deep-think@bis-code-mcp-deep-think
+```
+
+## How It Works
+
+Deep Think provides 5 MCP tools that work together:
+
+```
+strategize → think → reflect → branch → checkpoint
+     |          |        |         |          |
+  Pick a    Record    Check     Explore    Save/restore
+  reasoning  steps    for       alternate  reasoning
+  framework          issues    paths      state
+```
+
+**Start a session** with `/start` and describe what you're reasoning about. The plugin suggests a strategy and begins the thought chain.
+
+**Auto-checkpointing** saves your reasoning before context compaction and every 10 thoughts (configurable). Resume anytime with `/checkpoints`.
+
+## Commands
+
+| Command | What it does |
+|---------|-------------|
+| `/start` | Launch a reasoning session from a description |
+| `/checkpoints` | View, restore, or delete saved checkpoints |
+| `/constraints` | Manage reasoning rules and anti-patterns |
+| `/practices` | Manage review checklists and custom strategies |
+
+## Skills (auto-triggering)
+
+| Skill | Triggers when... |
+|-------|-----------------|
+| `restore-checkpoint` | Starting a new session and continuing previous reasoning |
+| `manage-constraints` | User says "remember to always check X" or "add a rule about Y" |
+| `manage-practices` | User says "add a review step for..." or "create a strategy for..." |
+
+## Configuration
+
+Create `.deep-think.json` in your project root to customize:
+
+```json
+{
+  "thinking": {
+    "defaultStrategy": "first-principles",
+    "maxThoughts": 50,
+    "autoCheckpointEvery": 10
+  },
+  "practices": {
+    "rules": ["Always consider security implications"],
+    "antiPatterns": ["Premature optimization"],
+    "reviewChecklist": ["Check for edge cases"]
+  },
+  "strategies": {
+    "custom": [{
+      "name": "incident-response",
+      "description": "Structured incident investigation",
+      "steps": ["Identify symptoms", "Gather evidence", "Form hypothesis", "Verify"],
+      "checkpoints": ["Evidence collected?", "Root cause confirmed?"]
+    }]
+  }
+}
+```
+
+Or use commands to manage configuration conversationally:
+- `/constraints` — add rules like "always verify auth boundaries"
+- `/practices` — add review steps or custom strategies
+
+## Built-in Strategies
+
+| Strategy | Use when... |
+|----------|------------|
+| `first-principles` | Conventional wisdom may be misleading |
+| `red-team` | Validating important decisions |
+| `convergent` | Choosing between multiple options |
+| `divergent` | Brainstorming or stuck on a problem |
+| `root-cause` | Debugging or investigating failures |
+| `decision-matrix` | High-stakes decisions with multiple factors |
+
+## License
+
+MIT

package/commands/checkpoints.md

@@ -0,0 +1,13 @@
+---
+name: checkpoints
+description: View, restore, or delete saved deep-think reasoning checkpoints
+---
+
+Show the user their saved reasoning checkpoints.
+
+1. Call the deep-think checkpoint tool with operation "list".
+2. Display each checkpoint with: name, timestamp, project path (or "(unknown)"
+   if not available), thought count, branch count, and active strategy.
+3. Ask the user what they'd like to do — restore one, delete one, or just browse.
+4. For restore: call checkpoint with operation "load" and the chosen name.
+5. For delete: call checkpoint with operation "delete" and the chosen name.

package/commands/constraints.md

@@ -0,0 +1,16 @@
+---
+name: constraints
+description: Manage reasoning rules and anti-patterns in .deep-think.json
+---
+
+Help the user manage their deep-think constraints conversationally.
+Constraints map to two fields in .deep-think.json:
+- practices.rules — things to always check during reasoning
+- practices.antiPatterns — patterns to avoid and flag
+
+1. Read .deep-think.json from the current project directory (create if missing).
+2. Ask the user what they want to do: add, remove, or show constraints.
+3. For add: ask what the rule or anti-pattern is, add it to the appropriate array.
+4. For remove: show current items, let them pick which to remove.
+5. For show: display all rules and anti-patterns.
+6. Write the updated .deep-think.json back to disk.

package/commands/practices.md

@@ -0,0 +1,16 @@
+---
+name: practices
+description: Manage review checklists and custom strategies in .deep-think.json
+---
+
+Help the user manage their deep-think practices conversationally.
+Practices map to two fields in .deep-think.json:
+- practices.reviewChecklist — items to check during reflection
+- strategies.custom — custom reasoning strategies with steps and checkpoints
+
+1. Read .deep-think.json from the current project directory (create if missing).
+2. Ask the user what they want to do: add/remove/show review items or strategies.
+3. For review items: manage the practices.reviewChecklist array.
+4. For strategies: manage strategies.custom array. Each strategy needs:
+   name, description, steps (array), and checkpoints (array).
+5. Write the updated .deep-think.json back to disk.

package/commands/start.md

@@ -0,0 +1,15 @@
+---
+name: start
+description: Launch a deep-think reasoning session from a natural language description
+---
+
+The user wants to start a structured reasoning session. They may have provided
+a topic or question as an argument.
+
+1. Call the deep-think checkpoint tool with operation "list" to check for
+   existing checkpoints. If any are related to the topic, offer to resume.
+2. Based on the topic, suggest an appropriate strategy from the available ones
+   (call strategize with operation "list" to see options, then "set" to activate).
+3. Begin the first thought in the chain using the think tool.
+
+If no topic was provided, ask the user what they want to reason about.

package/hooks/hooks.json

@@ -0,0 +1,11 @@
+{
+  "hooks": {
+    "PreCompact": [{
+      "matcher": "",
+      "hooks": [{
+        "type": "prompt",
+        "prompt": "Before compaction, if you have been using deep-think tools in this session and have recorded thoughts, save the current reasoning state by calling the deep-think checkpoint tool with operation 'save'. Use a descriptive name based on what you were reasoning about."
+      }]
+    }]
+  }
+}

package/package.json

@@ -0,0 +1,34 @@
+{
+  "name": "@bis-code/deep-think",
+  "version": "1.1.1",
+  "description": "Claude Code plugin for structured reasoning with auto-checkpointing, branching, and reflection",
+  "license": "MIT",
+  "author": "Ioan-Sorin Baicoianu <baicoianuioansorin@gmail.com>",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/bis-code/mcp-deep-think.git"
+  },
+  "keywords": [
+    "claude-code",
+    "plugin",
+    "reasoning",
+    "thinking",
+    "mcp",
+    "structured-reasoning",
+    "checkpoint"
+  ],
+  "files": [
+    ".claude-plugin",
+    "server/dist",
+    "server/package.json",
+    "commands",
+    "skills",
+    "hooks",
+    "rules",
+    ".mcp.json",
+    ".deep-think.json"
+  ],
+  "scripts": {
+    "prepublishOnly": "cd server && npm install && npm run build"
+  }
+}

package/rules/deep-think-workflow.md

@@ -0,0 +1,35 @@
+# Deep Think Workflow
+
+## When to Use Deep Think
+
+Use the deep-think reasoning tools for:
+- Complex decisions with multiple viable approaches
+- Problems where the full scope isn't clear initially
+- Multi-module changes that need careful planning
+- Analysis that might need course correction
+- Architecture or design decisions with trade-offs
+
+Skip deep-think for: simple fixes, documentation, dependency updates, single-file changes with obvious intent.
+
+## Recommended Workflow
+
+1. **Strategize** — Pick a reasoning framework that fits the problem (first-principles, red-team, convergent, divergent, root-cause, decision-matrix)
+2. **Think** — Work through the problem step by step, tracking confidence and assumptions
+3. **Reflect** — Check your reasoning for circular logic, contradictions, and gaps
+4. **Branch** — Explore alternatives when you hit a decision point
+5. **Conclude** — Merge branches, document the decision
+
+This workflow is a guide, not a mandate. Adapt it to the problem at hand.
+
+## Checkpointing
+
+- Checkpoints are saved automatically before context compaction
+- The server also auto-saves every N thoughts (default: 10, configurable)
+- Use `/deep-think:checkpoints` to view, restore, or delete saved reasoning state
+- Use `/deep-think:start` to begin a new reasoning session (checks for existing checkpoints first)
+
+## Customization
+
+- `/deep-think:constraints` — Add project-specific rules and anti-patterns
+- `/deep-think:practices` — Add review checklist items and custom strategies
+- Or edit `.deep-think.json` directly in your project root