paytriage
Enables AI agents to analyze payment gateway error messages and receive structured triage results, including likely causes, recommended checks, and retry safety guidance.
README
paytriage
Reads a payment gateway error and tells you what it probably means, what to check next, and whether retrying is safe.
Point it at a log line, an error string, or a whole log file. It answers in plain language instead of a code you have to go look up.
No dependencies. Python 3.8 and up. Works as a CLI, as a library, and as an MCP server an agent can call.
Why
Support and implementation people spend a lot of time turning response_code="51"
into "the customer's card did not have the money, this is not your bug." That
translation lives in people's heads and in bookmarked PDFs. This puts it in a file
you can read, grep, and correct.
The timeout case is the one that matters most. A timeout is not a decline. The transaction may already be authorized on the other side, so a blind retry is how you double-charge someone. paytriage says that out loud every time.
Install
git clone https://github.com/Flyingmiata-droid/paytriage.git
cd paytriage
pip install -e .
Or skip installing and run it in place with python -m paytriage.cli.
Use it from the command line
$ paytriage 'AUTH DECLINED response_code="51" insufficient funds'
[low] Insufficient funds (decline)
cause: The issuer declined the authorization because the account did not have
enough available balance or credit.
check: Nothing to fix on the integration side. Ask the cardholder to use
another card or retry later.
matched response code: 51
safe to retry: no
A whole file, rolled up:
$ paytriage -f examples/gateway.log --summary
lines: 11 matched: 9 unmatched: 2
by severity:
low: 2
medium: 3
high: 2
critical: 2
by signature:
decline.insufficient_funds: 1
transport.tls_handshake: 1
format.signature_mismatch: 1
...
Add --json to any call to get machine-readable output.
Use it as a library
from paytriage import triage_line
result = triage_line("SSLError: handshake failure tlsv1 alert protocol version")
result.severity # "critical"
result.retryable # True
result.findings[0].label # "TLS handshake failure"
result.findings[0].fix # what to check
Use it as an MCP server
The server speaks JSON-RPC over stdio and exposes one tool,
triage_payment_error, so an agent can hand it an error and get structured output
back.
python -m paytriage.mcp_server
Claude Desktop or Claude Code config:
{
"mcpServers": {
"paytriage": {
"command": "python",
"args": ["-m", "paytriage.mcp_server"]
}
}
}
What it covers
Twenty four signatures across six families:
- Declines: insufficient funds, do not honor, expired card, pick up card, invalid number
- Verification: AVS mismatch, CVV mismatch
- Transport: TLS handshake, certificate expired or untrusted, timeout, connection refused or reset, DNS
- API: 401, 403, 429, 5xx
- Message format: XML parse, JSON parse, signature or HMAC mismatch, missing required field
- Business rules: duplicate transaction, unsupported currency, bad amount, invalid token
Matching is by response code where one is present, and by pattern otherwise. A line can match more than one signature, and results come back worst first.
What it is not
- Not fraud scoring, and not a risk engine.
- Not connected to any gateway. It reads text you already have, and it makes no network calls.
- Not exhaustive. It is a first-pass triage helper. The issuer's own documentation is still the authority on any specific code.
- Not a decision maker. Nothing here should auto-retry a payment on its own.
Adding a signature
Everything lives in one table in paytriage/signatures.py. A signature is a
pattern or a set of codes, plus the cause, the fix, and a severity. Add an entry,
add a test, open a pull request. Corrections to the existing wording are just as
welcome as new entries.
Tests
python -m unittest discover -s tests -t .
23 tests, no network, no fixtures to download.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.