Agent tools

A repo-resident team memory system shared through Git, helping engineers and their AI agents store, retrieve, and hand over architecture decisions and project context locally.

OpenCode plugin giving coding agents persistent memory backed by a local vector database, enabling cross-session context retention and semantic retrieval for ongoing development tasks.

Builds an executable semantic context layer for data and analytics agents, letting assistants query company metrics accurately through MCP-based retrieval with shared business definitions.

Minimal persistent-memory toolkit for AI agents using a compact 426-token prompt and plug-and-play script for recall, logging, and token-level merge retrieval.

Rust-based computer-use layer that gives AI agents reliable desktop control. It reads application UI structure through OS accessibility trees rather than pixels, enabling stable references and safe, retryable actions.

Extends the Pi coding agent with web search and content extraction, aggregating multiple search and scraping services for zero-configuration lookup, page parsing, and video understanding.

MCP bridge connecting AI assistants such as Cursor, Copilot, and Claude to Cheat Engine for natural-language reverse engineering, pointer scanning, and memory analysis.

MCP-based repository toolkit for AI agents providing context-efficient file reading, writing, code retrieval, and background task management to reduce token usage.

Open-source LLM knowledge-compilation engine converting raw multi-format documents into structured, local-first wikis with citations and cross-indexes as a transparent, improvable alternative to traditional RAG.

Multi-agent knowledge engine providing grounded Q&A over codebases with file paths and line numbers. Works with Claude Code, Cursor, Codex CLI, and other AI IDEs.

Personal context store helps AI coding assistant users capture, search, and reuse project knowledge across repositories and agents, offering desktop GUI, MCP service, skills, and CLI integration.

A Zotero plugin exposing literature libraries to AI assistants over MCP, supporting paper search, metadata management, full-text analysis, and intelligent question answering.

An agent-native reverse-engineering lab combining a large knowledge base with MCP tools and automation for CTF challenges, APK and PE analysis, debugging, and firmware inspection.

Enterprise multimodal parsing and ontology-driven knowledge engine converting raw content into reusable vector, graph, retrieval, and business knowledge services for organizations.

A secure coding runtime service exposing code execution to any AI agent via MCP, letting MCP clients delegate file operations, shell commands, and programming tasks under controlled permissions.

Executable memory tool giving coding agents instant recall over local machine history, searching past agent sessions with source attribution like git blame for AI work.

A Rust-based self-hostable web scraping, crawling, and search API with a Firecrawl-compatible interface and MCP server support, providing structured web data to AI agents from a lightweight single-binary deployment.

An MCP-based scholarly retrieval tool for AI agents that resolves DOIs and retrieves academic PDFs, handling both open-access sources and authenticated paid-access content.

A tool that maintains a self-updating repository wiki, capturing project conversations, decisions, and context locally so coding agents can retrieve them.

A Model Context Protocol server that exposes Google Ads account data to AI assistants, enabling agents to query campaigns and account information programmatically on behalf of advertisers and marketing developers.

Enables login-free, key-free fetching of X/Twitter tweets, replies, timelines, and articles through multiple backends, with archiving and routing designed for programmatic use by AI agents.

Collection of MCP tools and skills for engineering-simulation agents, supporting model invocation, solver execution, monitoring, and lightweight result viewing workflows.

Local-first, Rust-based MCP server that gives AI agents detailed PDF understanding through structured text extraction, tables, OCR, and visual evidence. Returns page-level citations to support verifiable answers for document-heavy workflows.

Local, zero-dependency memory tool for coding agents, providing cross-session historical search, automatic recall, trust scopes, curated notes, and impact reports.

A metadata-driven CLI built for AI agent bots to call backend services via exec, handling messaging, docs, drive and fleet tasks with structured JSON envelopes.

Provides a Java-based cognitive memory engine for AI agents, building memory graphs and insight trees to support multi-channel context recall for long-running proactive assistants.

An AI toolkit for building reliable browser automations, giving coding agents live browser inspection and network forensics to support robust web integrations.

Toolkit transforming unstructured documents into validated, rich, queryable knowledge graphs to support high-accuracy Q&A in finance, chemistry, and legal workflows.

A free, local personal memory engine that records conversation details verbatim and supplies relevant memory slices each turn to Cursor, Claude Code, and other agents.

Parses repositories into persistent system maps for code intelligence, letting humans and AI agents use semantic queries to understand structure instead of manual searching and guesswork.

Local MCP memory service for coding agents that records issues, attempts, fixes, and decisions, warning developers before retrying failed approaches to save tokens and preserve privacy.

Context engineering infrastructure that provides unified MCP tools, secure third-party integrations, and one-click connectivity. Enables developers to give AI agents reliable, efficient access to real-world services and data sources.

Gives AI agents a controllable real browser via an MCP-native control plane with reusable login sessions, human approval, auditing, and local execution.

An open-source ontology-based semantic context layer that helps AI agents make more accurate, consistent, and explainable decisions by supplying retrieved and validated knowledge-graph context over MCP.

Provides a free web-based sub-agent workflow that uses a Chrome bridge to retrieve and summarize web content across multiple model providers.

An agent-native control plane for ComfyUI exposing MCP tools and skills that let developers author workflows, edit live graphs in natural language, and manage models, custom nodes, and media generation.

Exposes KiCAD 10 functionality as MCP tools through a single Rust binary, letting LLM assistants perform schematic editing, layout, routing, design review, and manufacturing tasks.

Open-source memory system for AI users that turns an Obsidian vault into persistent working memory using plain Markdown templates, avoiding vector databases for transparent long-term context management.

Local PDF and document knowledge base using libSQL with vector search for semantic retrieval, also usable as a memory-retrieval service for AI agents.

A CLI that runs multiple coding agents on the same task in isolated Git worktree sandboxes, orchestrates peer review among them, then replays candidates and merges the one passing tests.

An on-device MCP server exposing 57 tools so AI models can remotely see and operate your Android phone.

Unified data plugin for AI agent builders that adds web search, social media retrieval, and crypto on-chain data access through a single integration without extra configuration.

Stealth Chromium engine with fingerprint patches and anti-detection handling, giving scraper and browser-agent developers a drop-in execution base for passing verification checks reliably.

MCP server enabling AI agents to control Linux desktops via accessibility trees, screenshots, clicks, scrolling, and semantic actions using AT-SPI, GNOME Shell, and Wayland portals.

Builds a self-maintaining, Obsidian-compatible knowledge base that turns raw sources into an interlinked wiki using Native Open Knowledge Format v0.2, queryable by Pi and other agents via MCP.

Chinese-first enterprise RAG knowledge base covering controllable parsing, governance, chunking, hybrid retrieval, reranking, citations, GraphRAG, evaluation and Dify gateway integration.

Records the user's screen entirely locally and compiles activity into structured frames served over MCP, giving AI agents searchable episodic memory without cloud processing or LLM intermediation.

Acts as an independent verification layer for code changes and agent behavior, surfacing important issues through terminal use, CI integration, and skill-triggered reviews.