"A task involving critical or creative thinking" matching MCP servers:

A validation layer for AI coding assistants that enforces explicit LLM evaluations on plans, code diffs, and tests to ensure safer and higher-quality code.

Codebase Q&A agent with git-history awareness. Explore a repo with list_files, read_file, and search, plus git log, blame, and diff and a repo map to answer "why did this change?" questions. Runs locally as an MCP server for Claude Code and other MCP clients.

A lean, local knowledge graph that joins a repo's code to its aSPARK delivery artifacts, enabling agents to trace user stories to code and assess impact of changes, served over MCP.

Enables executing arbitrary Joern/CPGQL queries against a loaded Code Property Graph, loading source code in multiple languages (PHP, Java, JavaScript, Python, C/C++) to build new CPGs, and inspecting loaded projects and the active workspace. It communicates with a running Joern server over its synchronous HTTP query endpoint, so no state polling is required.

Zero-dependency MCP server inspector for protocol conformance, tool annotations, and capability checks.

An AI debugging agent MCP server that enables autonomous plan-act-observe debugging workflows, allowing repository exploration, code inspection, human-approved edits, and test execution through structured MCP tools.

A unified developer toolkit for AI-assisted workflows. Task timing, doc drift detection, env validation, secret scanning, port conflict resolution, AI context generation, and license auditing — one MCP server, one install.

Provides coding agents with file inventory, ranked search, location-aware reading, code packing, and note writing over the workspace folder through MCP, without involving a model.

Integrates Google Gemini 3.1 with Claude Code to provide web search, large codebase analysis, multimodal image understanding, and creative brainstorming, leveraging Gemini's 1M token context and Google Search grounding.

locidx

Analyzes codebases from local directories, GitHub, and Azure DevOps, providing intelligent context to AI coding assistants through repository structure, critical files, and semantic maps.

Marrow is an MCP server that solves agent amnesia. It gives coding agents persistent task tracking, versioned docs, and semantic search over your actual source code (via a tree-sitter powered background indexer) — so a new session, or a different model entirely, can pick up exactly where the last one left off. Built for multi-agent, multi-session software delivery.

A Model Context Protocol (MCP) server that helps AI coding assistants identify critical design issues in code, rather than just focusing on cosmetic problems when asked to improve code.

Deterministic repository context packing for AI coding agents: selects, compresses, and budgets only the files a task needs. Measured 83% fewer input tokens at the same task coverage, fully local, no LLM in the loop.

MCP server for DeepSeek AI models (Chat + Reasoner). Supports multi-turn sessions, model fallback with circuit breaker, function calling, thinking mode, JSON output, multimodal input, and cost tracking.

AI-powered MCP server that analyzes project structures to create optimized task architectures for mise, enabling intelligent task management, dependency mapping, and automated task suggestions.

Provides a memory server for coding agents over MCP, capturing and retrieving the meaning and rationale behind code decisions down to the function level. It enables agents to search product memory, get task context, and propose new memories to avoid re-deriving or re-breaking past decisions.

Enables coding agents to retrieve only the JavaScript/TypeScript code needed for a given task within a token budget, using TypeScript compiler analysis and a decision model to judge relevance.

Enables AI coding assistants to run pre-publish security and code-quality scans across 17 categories, generate a 0–100 security score and Markdown report, quickly check critical vulnerabilities, and receive fix/refactor suggestions by finding ID.

A Model Context Protocol server that enhances AI-generated code quality through comprehensive analysis across 10 critical dimensions, helping identify issues before they become problems.

Enables intelligent code analysis and search across repositories using the CodeRank algorithm (inspired by PageRank) to identify critical modules, trace dependencies, find code hotspots, and perform context-aware keyword searches with importance-ranked results.

Enables extraction of mathematical content from TeX papers and conversion to Lean code through a structured intermediate representation. Supports project scaffolding, entity management, and task tracking for mathematical formalization workflows.

Enables full-stack compiler development across the hw-native-sys workspace by combining repository operations (git health, code search, task running) with a knowledge layer for architecture docs, task routing, and abstraction/pass pipeline information.

An AI-powered Model Context Protocol server for Claude Code that provides code intelligence tools including codebase analysis, task management, component generation, and deployment configuration.

An MCP server that enables persistent memory, structured thinking sessions, and project-based knowledge management for Claude. It includes specialized coding tools for package discovery and reinvention prevention by validating code against existing libraries and APIs.

Enables MCP-compatible coding agents to request a small, ranked, explainable context package from a repository for a given task, returning the same JSON output as the CLI.

Scans generated JavaScript/TypeScript code to detect AI-hallucinated or suspicious npm package imports before installation, providing a verdict of CLEAN, REVIEW, or BLOCK.

Enables AI editors such as Cursor, Claude Code, Windsurf and VS Code to scan a project for common AI-introduced security mistakes — leaked keys, open Supabase/Firebase data, missing auth, unverified Stripe webhooks, SQL injection and more — and return findings with exact fix instructions. It also provides a guided prompt that walks an agent through fixing critical issues one by one and verifying each fix by re-scanning, without ever editing files or rotating keys itself.

Enables planning safe file or export renames from a caller-supplied module graph, returning direct and indirect referencers, a mechanical edit order, and risk flags without reading, renaming, or executing anything.

Task-adaptive repository context for coding agents, enabling budgeted evidence retrieval without API keys or network calls.