MCP Finder

Issues, PRs, repos and code search from your agent — the default first MCP for any dev workflow.

Sandboxed read/write access to local directories. The reference server for file operations.

Drive a real browser: navigate, click, screenshot, scrape. The workhorse for web automation and E2E checks.

Query your database with schema-aware SQL from the agent. Read-only mode available.

Fetch URLs and convert pages to markdown for the model — the minimal web-reading server.

Pull queries, impressions and indexing status from Google Search Console into your agent — real SEO data you own.

Create and update posts programmatically — pairs with keyword datasets for content pipelines.

Knowledge-graph memory that persists across sessions — give your agent long-term recall.

Inspect a local SQLite database from an agent while keeping the data close to the project.

Run managed browser sessions for research, testing, and workflows that need a real web runtime.

Read error issues and performance context so an agent can start from production evidence.

Turn product conversations into issues, cycles, and status updates without leaving the agent workflow.

Inspect products, customers, and payment context for billing-aware product and support work.

Search shared documents and turn an internal knowledge base into grounded context for an agent.