conclude
Analyze the entire thought graph and produce a comprehensive summary-conclusion with action items and graph health report.
LABSCO SUMMARY
Reasoning you can go back to — the graph is still there after the restart.
What separates this from a linear thinking tool is that the earlier work stays addressable: branch a thesis and its antithesis in parallel, challenge the strongest node with the devil's advocate pass, prune what went nowhere, then conclude over what survived. Node aliases and the suggested-next-call on every response keep the chaining cheap, which matters when one analysis runs to dozens of calls.
What it is
A reasoning server that keeps thoughts in a directed acyclic graph with branching, merging, and cross-edges. 17 tools work over that graph, under 10 reasoning strategies including an auto mode that selects one from the content and the current state of the graph.
What you get
- think — add a thought under a chosen strategy: sequential, dialectic, parallel, analogical, abductive, first_principles, counterfactual, systems_thinking, mcts, or auto
- evaluate and metacog — confidence scoring with support and contradiction analysis, plus detection of stuck states, stagnation, and declining confidence
- graph and prune — trace paths through the reasoning, and cut dead ends, redundancy, and unproductive branches
- simulate_devils_advocate, cross_disciplinary_synthesis, temporal_projection, ethical_framework_evaluation, emotional_intelligence_analysis, social_impact_analysis, and explain_decision
- conclude — a graph summary carrying the primary finding, action items, and graph health
- Node aliases last, best, and root, accepted anywhere a node ID is
- Every response comes back with a status, a confidence, and a nextSuggested call, so chaining does not need guesswork
- Sessions auto-save under ~/.deep-thinker/sessions/ and resume by name after a restart
Requirements
No account and no key. Run it over stdio with npx deep-thinker, or npm install -g deep-thinker and launch the deep-thinker command. Sessions are written under your home directory, so the graph outlives the process.
Setup effort
One command — npx deep-thinker
USE CASES
01
Deciding between two architectures without anchoring on the first one
Put both sides in as a dialectic, have the strongest one attacked, then summarise what actually survived.
02
Working out why production started throwing errors
Add an observation and its candidate explanations abductively, score them against the evidence, and close on the best one.
03
Noticing you are stuck before losing another hour
Ask for the stuck-state read, cut the branches that stopped producing, and restart from a different strategy.
04
Picking up yesterday's analysis
Resume the saved session by name, trace the path that got you there, and read the current conclusion.
05
Explaining a decision to someone who was not in the room
Render the reasoning graph and get the written account of why the surviving branch won.
17 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
cross_disciplinary_synthesis
Combine insights from multiple domains to generate novel perspectives. Creates analogical mappings between domains.
emotional_intelligence_analysis
Analyze emotional tone, stakeholder emotions, and social dynamics of thoughts. Provides empathy and persuasion insights.
ethical_framework_evaluation
Evaluate a thought or decision through multiple ethical frameworks (deontological, consequentialist, virtue ethics, rights-based).
evaluate
Evaluate the thinking process: score confidence, generate critiques, and assess overall graph health. Provides detailed analysis of weak spots and strong reasoning paths.
explain_decision
Generate human-understandable explanation of a decision path. Shows which factors contributed most and why.
graph
Query and visualize the thought graph. View the DAG structure, find paths, inspect branches, and get statistics.
help
Discover deep-thinker tools and learn usage workflows. Shows tools grouped by category with quick-start examples.
metacog
Metacognitive operations: view the current thinking state, get strategy suggestions, switch strategies, and receive guidance on improving reasoning. The system automatically detects stuck states and recommends actions.
optimize_prompt
Node Zero (PromptOptimizer): Transform vague/raw prompts into optimized Super Prompts with routing recommendations. Entry point for the reasoning DAG.
prune
Prune and optimize the thought graph. Remove dead ends, consolidate redundant branches, and optimize reasoning paths. Helps maintain graph efficiency during deep reasoning.
reset
Reset the thought graph and metacognitive state. Start a fresh reasoning session, save current session, or resume a saved session.
simulate_devils_advocate
Generate counterarguments and opposing viewpoints for a given thought. Automatically creates antithesis nodes.
social_impact_analysis
Analyze social impact, stakeholder emotions, group cohesion, and persuasion effectiveness of a thought or decision.
temporal_projection
Project thoughts into future or past scenarios. Analyze how conclusions change over time.
think
Add a thought to the cognitive graph using the current strategy. Supports sequential, dialectic, parallel, analogical, and abductive reasoning strategies. Each thought becomes a node in a DAG with confidence scoring, edges, and metadata.
visualize_thought_graph
Generate visual representation of the thought graph as SVG or ASCII. Supports highlighting paths, branches, and confidence levels.
BEFORE YOU INSTALL
What to know first
Tools you get
17 tools when we connected.
Requires
npx on PATH
Where it runs
The documented client runs the server on your machine.
CONTEXT COST
Moderate Standing cost
Rated by the size of the 17 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
TERMINAL
npx deep-thinkerMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"deep-thinker": {
"command": "npx",
"args": [
"-y",
"deep-thinker"
]
}
}
}
