Propagation breaker for agent fleets
Egress screening and outbreak telemetry against self-replicating prompt injections
- Pricing
- Paid
- Platforms
- API
- Runs
- Remotely
- Version
- 1.0.12
- Listing updated
- 2 days ago
About Propagation breaker for agent fleets
Propagation breaker for agent fleets is a paid MCP server for security. It runs as a hosted remote server, so there is nothing to install locally.
Where most defences score one input at a time, this product treats agent worms as a population problem. A payload that slips past each filter can still spread through a fleet if each infected agent passes it to more than one other, so the server measures spread as well as content.
The egress screen compares outbound messages with inbound context using blended shingle and token similarity, with no model call on the hot path. High-similarity instruction forwards are quarantined, and a borderline band is held closed, never silently passed. Fingerprint telemetry alerts when one payload appears in three or more distinct agents and reports an R estimate. Shared-store writes are data-only unless elevated by an HMAC token bound to the key and agent.
The vendor reports lab results (26 of 26 tests, 903 microseconds per message) and says there is no confirmed in-the-wild outbreak. Only LangGraph is wired in; other adapters are thin wrappers over the hosted API.
Install
Run in a terminal
claude mcp add --transport http propagation-breaker-agent-worms https://textforagents.com/products/propagation-breaker-agent-worms/mcpAdd to Settings → Connectors → Add custom connector
https://textforagents.com/products/propagation-breaker-agent-worms/mcpAdd to ~/.cursor/mcp.json
{
"mcpServers": {
"propagation-breaker-agent-worms": {
"url": "https://textforagents.com/products/propagation-breaker-agent-worms/mcp"
}
}
}Add to .vscode/mcp.json
{
"servers": {
"propagation-breaker-agent-worms": {
"type": "http",
"url": "https://textforagents.com/products/propagation-breaker-agent-worms/mcp"
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"propagation-breaker-agent-worms": {
"serverUrl": "https://textforagents.com/products/propagation-breaker-agent-worms/mcp"
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"propagation-breaker-agent-worms": {
"type": "http",
"url": "https://textforagents.com/products/propagation-breaker-agent-worms/mcp"
}
}
}Tools this server provides (6)
Taken from the project documentation; the running server may differ.
- egress_check
- Score outbound agent output and quarantine instruction-like forwards.
- fingerprint_query
- Ask whether a payload fingerprint has been seen propagating.
- store_write
- Write to the shared store as data only.
- store_read
- Read from the shared store with read classes.
- telemetry_report
- Report fingerprint telemetry and outbreak alerts.
- controls_verify
- Verify the breaker's controls are intact.
What it does
- Egress quarantine
- Holds or blocks outbound messages that copy instructions from inbound context.
- Outbreak telemetry
- Alerts when a fingerprint reaches three or more agents, with an R estimate.
- Data-only shared store
- Writes cannot become instructions without a bound cryptographic token.
- Fingerprint query
- Agents can ask whether a payload has been seen propagating.
- Control verification
- A sealed registry detects tampering with the breaker's own controls.
- LangGraph adapter
- Screens messages inside StateGraph transitions.
Good for
- Screen messages between agents for worm-like payloads
- Watch a fleet for propagation of one payload
- Prevent shared memory writes from becoming instructions
- Check a forwarded instruction before acting on it
Pricing
Paid, from $19 a month
Fleet $19/mo, Fleet Pro $49/mo, fingerprint query $0.001; no perpetual free tier
Prices change. Check textforagents.com before you buy.
Propagation breaker for agent fleets FAQ
What does it cost?
Fleet is $19 a month for up to 50 agents, Fleet Pro is $49 for up to 500, and a fingerprint query is $0.001 via x402.
Is there a free tier?
No perpetual one; the first fleet outbreak scan is free, once per operator.
Does it use an LLM to judge messages?
No. The egress screen makes no inference call on the hot path.
What are its limits?
Detection is proven for the lab shapes only, and an agent that never consults the service is outside the screen.
Is Propagation breaker for agent fleets free?
No. Propagation breaker for agent fleets is a paid product. Paid plans start at $19 a month.
How do I add Propagation breaker for agent fleets to Claude Code?
Run this command in your terminal: claude mcp add --transport http propagation-breaker-agent-worms https://textforagents.com/products/propagation-breaker-agent-worms/mcp
What are the best alternatives to Propagation breaker for agent fleets?
The closest alternatives to Propagation breaker for agent fleets are REA, Agent Roast, Midplane, Leash and Tinplate Edge AI Safety, Security & EU Compliance.
Alternatives to Propagation breaker for agent fleets

REA
MCP serverCLI and MCP server that lets coding agents reverse engineer apps and binaries
Runs locally- Open source
- 464K GitHub stars
- Developer tools
Agent Roast
MCP serverOffline review that flags agent changes which weaken tests, assertions or CI
Runs locally- Open source
- 4 Hacker News points
- Developer tools

Midplane
MCP serverPostgres gateway for AI agents with table grants, column masking and approval for writes
Runs locally- Open source
- 5 GitHub stars
- Databases

Leash
MCP serverReview what your AI touched, approve risky actions and undo supported changes
- 1 tool
- Remote
- Free
- 1 GitHub star
- Security
Zero-log MCP server with 49 tools for AI guardrails, secret scanning and EU compliance
- 49 tools
- Remote
- Freemium
- Security

SecureCloud
MCP serverSentinelOne and IT security catalog, prices, quotes and checkout links for Israel
- 7 tools
- Remote
- Free
- Security