Skip to main content

Automatic GitHub issue creation + PR fix on runtime errors

Project description

Splat

PyPI version Python versions License

Automatic GitHub issue creation on application crashes with optional Claude Code auto-fix.

Features

  • Zero-config crash reporting - Add two lines, get GitHub issues on errors
  • True deduplication - Checks GitHub before creating, works across restarts/instances
  • Rich context - Rolling log buffer, framework-specific request context, custom metadata
  • Auto-fix ready - One command installs a GitHub Action that uses Claude Code to fix reported bugs

Installation

pip install py-splat

Quick Start

Basic Usage

from splat import Splat

splat = Splat(repo="owner/repo", token="ghp_...")

try:
    do_something()
except Exception as e:
    await splat.report(e)  # async - use in async context

With Context

try:
    process_order(user)
except Exception as e:
    await splat.report(
        e,
        context={
            "user_id": user.id,
            "endpoint": "/checkout",
        },
    )

Flask Integration

from flask import Flask
from splat.middleware.flask import SplatFlask

app = Flask(__name__)
splat = SplatFlask()
splat.init_app(app, repo="owner/repo", token="ghp_...")

FastAPI Integration

from fastapi import FastAPI
from splat.middleware.fastapi import SplatMiddleware

app = FastAPI()
app.add_middleware(SplatMiddleware, repo="owner/repo", token="ghp_...")

Configuration

Environment Variables

Variable Description Default
SPLAT_GITHUB_TOKEN GitHub API token (required)
SPLAT_GITHUB_REPO Repository (owner/repo) (required)
SPLAT_ENABLED Enable/disable reporting true
SPLAT_LOG_BUFFER_SIZE Log buffer capacity 200
SPLAT_LABELS Issue labels (comma-separated) bug,splat
SPLAT_VERCEL_SECRET Vercel webhook verification secret (optional)
SPLAT_VERCEL_WEBHOOK_PATH Vercel webhook endpoint path /splat/logs
SPLAT_VERCEL_LOG_TTL Seconds to keep Vercel logs in memory 60

pyproject.toml

[tool.splat]
repo = "owner/repo"
labels = ["bug", "splat", "auto-fix"]
log_buffer_size = 500

Precedence

Configuration is loaded with precedence: Programmatic > pyproject.toml > Environment Variables > Defaults

CLI

# Interactive setup wizard
splat init

# Install auto-fix GitHub Action
splat install-autofix

Auto-Fix with Claude Code

Splat can automatically create PRs to fix reported bugs using Claude Code:

  1. Install the auto-fix workflow:

    splat install-autofix
    
  2. Add ANTHROPIC_API_KEY to your repository secrets

  3. Add auto-fix to your labels:

    [tool.splat]
    labels = ["bug", "splat", "auto-fix"]
    

When an error is reported, Claude Code will analyze the issue, write a failing test, implement a fix, and open a PR.

Vercel Log Integration

Splat can capture Vercel runtime logs via Log Drain webhooks, providing precise logs for each failed request.

Setup

  1. Configure your app:

    Flask (webhook auto-registered):

    from flask import Flask
    from splat.middleware.flask import SplatFlask
    
    app = Flask(__name__)
    splat = SplatFlask()
    splat.init_app(app, repo="owner/repo", token="ghp_...")
    # Automatically registers /splat/logs webhook
    

    FastAPI (manual middleware wrapping to share Splat instance):

    from fastapi import FastAPI
    from splat.middleware.fastapi import SplatMiddleware
    from splat.webhooks.fastapi import create_webhook_router
    
    app = FastAPI()
    # Add webhook router first
    middleware = SplatMiddleware(app, repo="owner/repo", token="ghp_...")
    app.include_router(create_webhook_router(middleware.splat))
    # Use middleware.app or wrap app manually if needed
    
  2. Add Log Drain in Vercel:

    • Go to your project's Settings → Log Drains
    • Create a new Log Drain:
      • Delivery format: JSON
      • Endpoint: https://your-app.vercel.app/splat/logs
      • Sources: Lambda, Edge (runtime logs)
  3. (Recommended) Add verification secret:

    # Generate a secret
    export SPLAT_VERCEL_SECRET=$(openssl rand -hex 32)
    # Add to your environment variables and Vercel Log Drain config
    

How It Works

  1. Vercel pushes runtime logs to your /splat/logs endpoint
  2. Splat stores logs keyed by request ID (from x-vercel-id header)
  3. When an error occurs, middleware captures the request ID
  4. splat.report() fetches logs for that specific request
  5. GitHub issue includes the exact logs for the failed request

Logs automatically expire after 60 seconds (configurable via SPLAT_VERCEL_LOG_TTL).

Issue Format

Created issues include:

  • Error type, message, file, line, and function
  • Full traceback
  • Optional context (custom metadata you provide)
  • Recent logs (last 200 entries by default)
  • Unique signature for deduplication

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

py_splat-0.1.5.tar.gz (99.6 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

py_splat-0.1.5-py3-none-any.whl (36.4 kB view details)

Uploaded Python 3

File details

Details for the file py_splat-0.1.5.tar.gz.

File metadata

  • Download URL: py_splat-0.1.5.tar.gz
  • Upload date:
  • Size: 99.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for py_splat-0.1.5.tar.gz
Algorithm Hash digest
SHA256 d358570deeeb3066ea388dbd8fde605e513c9ed93a8eb5c9d76ca7e58b2dadc5
MD5 99e6d5e1bae81ba9dddfe9b5a5ba54c3
BLAKE2b-256 6477b0aa36773ff4163a5dbef51244a121d57602e4251bc36ed98664181dbe13

See more details on using hashes here.

Provenance

The following attestation bundles were made for py_splat-0.1.5.tar.gz:

Publisher: publish.yml on andreaslordos/splat

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file py_splat-0.1.5-py3-none-any.whl.

File metadata

  • Download URL: py_splat-0.1.5-py3-none-any.whl
  • Upload date:
  • Size: 36.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for py_splat-0.1.5-py3-none-any.whl
Algorithm Hash digest
SHA256 8545b83abb95bfa28d2adec4cd0cb61778feed6afb5dc49d1bb79609f90450b3
MD5 7b51497e12e6b03c7b03d126d07bbcbd
BLAKE2b-256 964bdcb78387c5fc666ca5ccfe9c18b547219f11ec473c00585d2babb3d7d0a6

See more details on using hashes here.

Provenance

The following attestation bundles were made for py_splat-0.1.5-py3-none-any.whl:

Publisher: publish.yml on andreaslordos/splat

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page