Skip to content

execution-layer.md

Latest commit

 

History

History
148 lines (103 loc) · 2.91 KB

execution-layer.md

File metadata and controls

148 lines (103 loc) · 2.91 KB
Raw

Execution Layer

TrendPilot AI is no longer only a documentation starter kit. It includes a small local execution layer that can:

  • validate a source-log CSV
  • calculate product opportunity scores
  • generate a Markdown daily trend report
  • run unit tests against scoring and report logic
  • provide optional integration examples for Firecrawl and Google Sheets

The execution layer is intentionally lightweight. It does not scrape private data, send messages, auto-publish reports, or promise sales or profit.

Install

python -m pip install -e .

Validate CSV records

python -m trendpilot validate --input examples/sample-source-log.csv

Generate a Markdown report

python -m trendpilot report \
  --input examples/sample-source-log.csv \
  --output examples/generated-daily-report.md \
  --limit 5

Score a product idea

python -m trendpilot score \
  --score-id TP-DEMO-001 \
  --product-idea "Portable desk organizer" \
  --category "Home office" \
  --target-region "United States" \
  --target-audience "Remote workers and students" \
  --demand 4 \
  --social 3 \
  --price 3 \
  --supplier 4 \
  --differentiation 3 \
  --competition 4 \
  --shipping 2 \
  --compliance 1 \
  --risk-level Low \
  --confidence Medium

Run tests

python -m unittest discover -s tests

or:

make test

Optional n8n local runtime

A basic docker-compose.yml is included for users who want to experiment with n8n locally.

cp .env.example .env
# edit .env first
mkdir -p data outputs
docker compose up -d

The included .pseudo.json workflow files are planning references, not import-ready n8n exports.

Optional integration examples

See:

These examples require your own credentials. Never commit real API keys or service account files.

Developer Workflow

Install the package locally:

make install

Install development tools for linting and schema validation:

make dev

Run the full local check:

make ci

The score shortcut now reads from a JSON input file instead of using a very long command:

make score

Input file:

examples/sample-product-score-input.json

Output file:

examples/generated-product-score.json

CI Coverage

The GitHub Actions workflow in .github/workflows/ci.yml runs:

  • editable package install
  • Ruff lint check
  • Python unit tests
  • JSON Schema sample validation
  • CLI smoke tests for validate, report, and score

Legacy Install Compatibility

The project uses pyproject.toml as the primary package configuration and includes a minimal setup.py only to keep older setuptools/pip workflows from failing.