The agent budget gate standard.

Pre-flight hard-stop before token-1. The model that loops cannot bypass it because the gate lives outside its context.

# one line, zero refactor
$ pip install burnstop

import burnstop
burnstop.install(budget_envelope_usd=5.00)

Why this exists.

Four public incidents, one missing primitive. The same shape every time.

Jul 2025
Replit AI
prod DB wiped
DROP DATABASE during code freeze. 1,206 records gone.
Nov 2025
LangChain A2A
$47,000
11-day clarification loop. No per-conversation budget.
2026
LangGraph
$4,200
Weekend autonomous refactor. No alerting until Monday.
Apr 2026
Cursor + Claude
PocketOS DB
9-second DB + backup wipe. 30h outage for rental clients.

Ecosystem.

Use the core directly or wire it into your framework with a 1-line install.

burnstop

Core: pre-flight gate + decomposition for Anthropic SDK.

pip install burnstop

github →

burnstop-langchain

LangChain CallbackHandler. Drop-in for CallbackManager.

pip install burnstop-langchain

github →

burnstop-langgraph

LangGraph node wrapper + shared middleware.

pip install burnstop-langgraph

github →

burnstop-crewai

Per-agent or per-crew shared budget envelope.

pip install burnstop-crewai

github →

burnstop-autogen

ConversableAgent + GroupChat shared receipt.

pip install burnstop-autogen

github →

agentic-receipts

DID-signed audit chain. SOC2 / ISO 42001 / EU AI Act.

pip install agentic-receipts

github →

How it compares.

Honest matrix. No marketing tricks.

ToolFiresLayerBest at
burnstopBEFORE token-1pre-callHard-stop $-envelope outside model context
Helicone / LangfuseAFTER calltelemetryPer-call observability
Lakera / Protect AIDURING (prompt)content guardPrompt-injection / PII
BraintrustAFTER, offlineevalCompare model variants on golden datasets
LangChain max_iterationsDURINGloop counterStop runaway turn count

Full comparison in COMPARISON.md.