Exposes 16 local stdio MCP tools that give Codex-style clients project-isolated persistent memory, local ONNX-based semantic search, and durable task tracking that survives across sessions and concurrent clients, without executing agents, hosting a model, or contacting model providers.

Enables MCP hosts to spawn, supervise, and collect results from parallel panels of pinned-model OpenCode v2 agents for delegated tasks, reviews, and cross-model comparisons. Supports per-member roles, models, permissions, follow-up prompts, cancellation, and persisted result retrieval.

Enables agents to reach pump.fun's full API surface — official developer routes, bonding-curve reads, Solana RPC, and unsigned trade building — by writing JavaScript against one merged OpenAPI document, using just two tools (search and execute). Model-authored code runs in a network-less sandbox, so multi-step questions resolve in a single round trip while credentials stay on the host Worker.

Enables coding agents to navigate and verify a contract-backed semantic graph of TypeScript backends, inspect change impact, run checks, explain failures, and draft human-reviewable contracts through bounded MCP tools and matching JSON CLI output.

Connects MCP clients to a public CVE activity feed and private product watches, letting agents search advisories, save watched products, check and acknowledge matches, and track new or updated vulnerability records.

Enables people and AI agents to create, preview, publish, withdraw, and discover offers and requests across goods, services, trips, and delivery, with semantic matching and secure shared rooms for negotiation.

It enables AI agents to access live web data through page fetching, search, prebuilt scrapers, datasets, and remote browsers, using free direct requests and local Chrome rendering with Bright Data only as a fallback for bot walls. It also adds budget controls, caching, pagination, cancellation, retries, SSRF protection, and both stdio and HTTP transports.

Mounts a headless, low-memory browser for agents with a text-first tool interface, enabling navigation, semantic tree/markdown extraction, structured-data scraping, screenshots, form interaction, and session capture. Reads and interactions are returned as semantic trees, markdown, or CSS-selector JSON rather than pixels, which is cheaper for a model to reason over.

Enables AI agents to run web searches and fetch page content by driving a real Chromium browser over CDP, so every request carries an existing profile's cookies, login state, and fingerprint. It also lets agents click elements, fill and submit forms, read the current tab, and check browser/CDP status, returning accurate results from logged-in pages without being treated as an anonymous crawler.

Enables agents to crawl, search, map, and extract structured evidence from websites, SPAs, PDFs, Office files, YouTube videos, GitHub repositories, ArXiv papers, shopping and travel listings, and CVE advisories via local or cloud-backed tools.

Enables agents to search local Chinese text memories and fetch individual records through stdio tools, using bigram FTS5 recall with optional vector rank fusion.

Enables agents to search the web with ranked content-bearing results, extract clean text from pages, crawl sites under configured limits, and map site structure via the Tavily API.

Enables agents to run neural web searches, gather token-budgeted code context and documentation excerpts, and crawl individual pages through the Exa API.

Enables agents and humans to fetch YouTube transcripts, search within them, inspect video metadata, and monitor channel uploads through bounded, resumable MCP tools.

Enables agents to query the Brave Search API across web, local, news, image, and video indexes through five MCP tools, returning consistent result details.

Enables AI agents to read and write a local-first Markdown/Obsidian notebook via a single capture call that routes text into the right notebook, board, or daily journal entry and returns related notes, alongside search, note and card management, and numeric metric series. Access is governed by per-agent tokens with read/write/admin scopes, optional folder restrictions, and an audit log of every action.

Enables AI coding agents to read GitHub repositories, files, commits, issues, pull requests, releases, Actions, and search results through 18 read-only MCP tools. It reuses existing credentials, routes around unreliable networks, and caches responses on disk to stay within rate limits.

Give your AI agents persistent, collective memory — with deduplicating absorb, supersession lineage, semantic search, and a graph UI. Speaks MCP.

Multi-language repository map for agents and humans: MCP tools, dependency graphs, symbol indexes, impact review, and bounded source evidence.

A structural code search engine for Al agents.

Automated harness evolution for AI agents. A Claude Code plugin that iteratively optimizes system prompts, routing, retrieval, and orchestration code using full-trace counterfactual diagnosis. Based on Meta-Harness (Lee et al., 2026).

A self-hosted MCP memory server that gives every MCP-compatible agent a shared, persistent memory layer, so preferences, project conventions, and past decisions carry across sessions and across different agents and vendors. It combines hybrid keyword-plus-vector retrieval with automatic context injection, zero-config sync, and selective injection of memories by agent and priority.

Enables AI coding agents to retrieve compact structural codebase context through repository architecture summaries, AST symbol extraction, cyclomatic complexity scoring, and TODO/FIXME hotspot search.

Enables AI agents and LLMs to persist facts, preferences, and observations in a local SQLite store and to build and traverse subject-predicate-object knowledge graphs across sessions. It provides fuzzy keyword search and graph queries so assistants retain context and recall prior project history without repeating prompts.