Reusable automation patterns with low-code contracts, deterministic fixtures, and tested Python implementations.
Automation Kit helps turn common business workflows — CRM imports, webhook routing, email triage, lead enrichment, calendar booking, Slack alerts, and creative-asset preparation — into repeatable automation patterns that can be inspected, tested, and run locally before connecting real services.
Each pattern includes:
- a low-code-style
workflow.jsoncontract - synthetic fixtures and expected output
- a deterministic Python implementation
- regression tests
- notes on when to keep the workflow in n8n/Make/Zapier versus when to move logic into code
The repo is safe to run from a clean checkout: all shipped examples use synthetic data, mock clients, and disabled-by-default live service boundaries.
Use Automation Kit when you need to:
- turn a messy workflow idea into a repeatable automation pattern
- compare low-code and Python implementations before choosing a delivery path
- test CRM, webhook, email, calendar, Slack, enrichment, or creative workflows without live credentials
- package reusable automation logic for companion case-study repositories
- expose fixture-safe automation capabilities through CLI, API, Docker, or MCP
Automation Kit is not a hosted SaaS app and does not connect to live third-party accounts by default.
| # | Pattern | Workflow problem | Output |
|---|---|---|---|
| 1 | csv-to-crm |
Import, normalize, dedupe, and upsert lead rows | Mock CRM upsert summary |
| 2 | email-parser |
Classify inbound emails into support, sales, billing, spam, or review queues | Routed classification result |
| 3 | lead-enrichment |
Enrich B2B prospect data with firmographic records | Firmographic records and manual-research flags |
| 4 | calendar-booking |
Booking requests and availability checks | Event decision and confirmation message |
| 5 | webhook-router |
Third-party event fan-out and dead-letter routing | Typed handler result or dead-letter queue item |
| 6 | slack-alerts |
Ops alerts and team notifications | Severity-routed mock Slack messages |
| 7 | product-creative-pack |
Ecommerce creative asset preparation | Prompt pack, ComfyUI manifest, mock assets, review packet |
Every pattern lives under patterns/<name>/ with:
workflow.json— declarative workflow structure.fixtures/— synthetic inputs plusexpected_output.json.python/main.py— runnable Python equivalent.python/test_*.py— pattern-specific regression tests.README.md— implementation notes and automation fit.
See docs/pattern-index.md for inputs, outputs, fit, and live-service boundaries for every pattern.
Requires Python 3.11+.
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
auto-kit list-patterns
auto-kit validate-all
auto-kit run patterns/csv-to-crmExpected validation result:
7 pattern(s): 7 passed, 0 failed
Source-checkout module invocation also works without installing the console script:
PYTHONPATH=src python -m auto_kit.cli validate-allauto-kit list-patterns
auto-kit validate patterns/webhook-router
auto-kit run patterns/webhook-routerPYTHONPATH=src uvicorn auto_kit.api:app --host 127.0.0.1 --port 8000
curl -fsS http://127.0.0.1:8000/health
curl -fsS http://127.0.0.1:8000/patternsOpenAPI docs are available at http://127.0.0.1:8000/docs.
auto-kit mcp-validate
auto-kit mcp-servedocker build -t automation-kit .
docker run --rm automation-kit validate-allEvery pattern follows the same contract:
patterns/<name>/
workflow.json low-code-style workflow contract
README.md workflow fit, boundaries, and implementation notes
fixtures/
input.* synthetic input data
expected_output.json expected deterministic result
python/
main.py runnable Python equivalent
test_<name>.py pattern-specific regression tests
The repository validates this structure through auto-kit validate-all and tests/test_all_patterns.py.
Automation Kit is designed for the decision point before implementation:
- If the workflow is simple, visual, and owned by non-developers, keep it in n8n, Make, or Zapier.
- If it needs validation, deduplication, replay safety, CI, or auditability, move the risky logic into tested Python.
- If it needs both, keep the workflow contract visible and test the risky logic locally.
Automation Kit keeps both views visible: workflow.json explains the low-code flow, while python/main.py proves the same logic can run deterministically under tests.
Automation Kit is safe to run from a clean checkout:
- all shipped data is synthetic
- mock clients are used by default
- live services require explicit environment opt-in
.env.examplecontains empty credential placeholders- API and MCP responses report
fixture_safe=trueandlive_services_used=false - ComfyUI/network calls are disabled unless
AUTO_KIT_USE_LIVE_SERVICES=trueand a local endpoint are explicitly supplied
The product-creative-pack pattern models an ecommerce creative workflow: product brief, prompt pack, ComfyUI job manifest, deterministic mock assets, and a human review packet. It does not generate images by default.
Live ComfyUI calls are isolated behind auto_kit.comfyui_client.ComfyUIClient, which raises before any network operation unless AUTO_KIT_USE_LIVE_SERVICES=true and COMFYUI_BASE_URL are supplied in the local runtime environment.
Automation Kit is the reusable pattern layer. Companion case-study repositories apply the same pattern discipline to one concrete workflow problem at a time.
Public companion case studies include:
api-webhook-bridge— safe API/webhook builds with validation, mapping, idempotency, audit, and dead-letter behavior.automation-debugger— broken automation diagnosis, normalized event inspection, safe replay/refusal, and fix reports.
See docs/case-studies/api-webhook-bridge.md and docs/case-studies/automation-debugger.md for the Automation Kit relationship.
docs/pattern-index.md— inputs, outputs, automation value, and fit for every pattern.docs/architecture.md— package boundaries, runtime flow, and live-service isolation.docs/api.md— local FastAPI/OpenAPI surface for fixture-safe pattern runs.docs/mcp.md— fixture-safe MCP surface and capability registry.docs/deployment.md— local Docker/API runtime, healthcheck, and cloud-free boundary.docs/public-readiness.md— public-surface audit criteria and release hygiene checks.docs/proof-spoke-architecture.md— companion case-study architecture and readiness contract.EVIDENCE.md— latest verification snapshot.
These screenshots show the current CLI validation, pattern outputs, architecture, quality gates, and companion case-study relationship.
python -m pytest -q
python -m ruff check .
python -m mypy src
PYTHONPATH=src python -m auto_kit.cli validate-all
PYTHONPATH=src python -m auto_kit.cli mcp-validate
docker build -t automation-kit:local .
docker run --rm automation-kit:local validate-allSee .env.example:
| Variable | Default | Description |
|---|---|---|
AUTO_KIT_USE_LIVE_SERVICES |
false |
Explicit opt-in gate for live API clients. |
AUTO_KIT_MOCK_SEED |
42 |
Seed for deterministic mock data. |
COMFYUI_BASE_URL |
empty | Optional local or cloud ComfyUI base URL. |
COMFY_CLOUD_API_KEY |
empty | Optional Comfy Cloud key supplied only outside git. |
src/auto_kit/
api.py
capability_registry.py
cli.py
comfyui_client.py
fixtures.py
mcp_server.py
mock_clients.py
models.py
pattern_runner.py
registry/
workflow_schema.py
patterns/
<pattern-name>/
workflow.json
README.md
fixtures/
python/
docs/
architecture.md
api.md
case-studies/
deployment.md
mcp.md
pattern-index.md
proof-spoke-architecture.md
screenshots/
examples/
api-requests/
api-responses/
mcp/
registry/
capabilities.yaml
sectors.yaml
MIT License. See LICENSE.
Part of Stefan's automation tools catalog.