steadyfetch
Reliable web fetching for AI agents — MCP server with retry, circuit breaker, caching, and anti-bot bypass
Documentation
SteadyFetch
Reliable web fetching for AI agents. Stop losing hours to Cloudflare blocks, timeouts, and flaky scrapes.
SteadyFetch is an MCP server that gives your AI agents reliable web access with automatic retry, circuit breaker protection, caching, and anti-bot bypass — out of the box.
The Problem
Every AI agent that touches the web hits the same wall:
- Cloudflare blocks your requests
- Sites return CAPTCHAs
- Pages timeout or load partially
- Rate limits kill your batch jobs
- You waste hours debugging flaky scrapes
The Solution
One MCP tool call. SteadyFetch handles the rest.
Agent calls fetch_url("https://example.com")
→ Checks cache (instant if hit)
→ Checks circuit breaker (fail-fast if domain is down)
→ Stealth browser fetch with anti-bot bypass
→ On failure: retry with exponential backoff
→ Fallback: plain HTTP fetch
→ Cache the result
→ Return clean markdown + raw HTMLTools
| Tool | Description |
|---|---|
| `fetch_url` | Full reliability fetch — returns markdown + HTML |
| `fetch_markdown` | Returns only clean markdown, optimized for LLMs |
| `check_domain` | Circuit breaker status for a domain |
| `cache_stats` | Cache utilization metrics |
| `clear_cache` | Flush the cache for fresh data |
Quick Start
Install from PyPI
pip install steadyfetch
steadyfetchThen connect from Claude Desktop:
{
"mcpServers": {
"steadyfetch": {
"command": "steadyfetch"
}
}
}Self-host with Docker
docker build -t steadyfetch .
docker run -p 8200:8200 steadyfetchConfiguration
All settings via environment variables:
| Variable | Default | Description |
|---|---|---|
| `PORT` | 8200 | Server port |
| `STEADYFETCH_MAX_RETRIES` | 3 | Retry attempts per URL |
| `STEADYFETCH_CIRCUIT_THRESHOLD` | 5 | Failures before circuit opens |
| `STEADYFETCH_CIRCUIT_COOLDOWN` | 120 | Seconds before retrying a broken domain |
| `STEADYFETCH_CACHE_TTL` | 3600 | Cache lifetime in seconds |
| `STEADYFETCH_TIMEOUT` | 30000 | Page load timeout in ms |
How It Works
Retry with backoff — Exponential backoff + jitter prevents retry storms. 3 browser attempts before falling back to HTTP.
Circuit breaker — Per-domain failure tracking. After 5 consecutive failures, the domain is circuit-broken for 2 minutes. Prevents wasting time on sites that are blocking you.
Caching — Disk-backed cache with configurable TTL. Repeat fetches are instant. 500MB default limit.
Anti-bot bypass — Stealth browser with magic mode, navigator patching, and human-like behavior simulation via Crawl4AI.
Graceful degradation — If the browser can't get through, falls back to plain HTTP. If HTTP fails, returns a clear error with domain health status. Never hangs, never silently fails.
Free vs Pro
| Tool | Free | Pro ($19/mo) |
|---|---|---|
| `fetch_url` | Yes (no JS render, no cache) | Yes (full: JS render + cache + anti-bot) |
| `check_domain` | Yes | Yes |
| `fetch_markdown` | - | Yes |
| `cache_stats` | - | Yes |
| `clear_cache` | - | Yes |
Free tier gives you basic HTTP fetching and domain health checks. Pro unlocks JS rendering, anti-bot bypass, caching, and clean markdown output.
**Upgrade to Pro on MCPize** — $19/mo or $190/yr.
License
MIT
Frequently asked questions
What is steadyfetch?
steadyfetch is Reliable web fetching for AI agents — MCP server with retry, circuit breaker, caching, and anti-bot bypass
How do I install steadyfetch?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is steadyfetch open source?
Yes — it is hosted on GitHub at https://github.com/carsonlabs/steadyfetch.
Related MCP tools
Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory across sessions with a self-hosted knowledge graph engine.
Automate browser based workflows with AI
Hindsight: Agent Memory That Learns
A privacy-first app that strips AI watermarks from content you own.
Agent framework and applications built upon Qwen>=3.0, featuring Function Calling, MCP, Code Interpreter, RAG, Chrome extension, etc.
The power of Claude Code / GeminiCLI / CodexCLI + [Gemini / OpenAI / OpenRouter / Azure / Grok / Ollama / Custom Model / All Of The Above] working as one.
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP