SignalTally.
Menu
Talk to sales

SIGNALTALLY / AGENT KITS / 02

A new storefront.
A better introduction.

Spot a free-to-premium theme change in your evidence. Prepare a theme-specific app or service introduction while the merchant is shaping their storefront.

Free starter code · Bring your own runtime · Licensed data when you scale

BUILD IN YOUR ENVIRONMENTLangChainCrewAIViktorPython runners & a workspace task prompt

SHOPIFY APP & THEME COMPANIES

A better reason
to start a conversation.

Give app and agency teams a precise reason to investigate compatibility, implementation and merchant needs.

Inspect the starter →
Theme-fit opportunity briefINCLUDED CASE / OUTPUT PREVIEW

Atelier Ridge

Fictional history · reserved example domain · replay: 13 Sep 2026

  • 01Free → premium theme match
  • 02Prestige compatibility · validate
  • 03Implementation angle · ready to review
  • 04Theme license purchase · not established

Preliminary evidence review · A technical audit requires additional validation.

01

Identify the change

Compare the latest two theme observations and match a free-to-premium catalogue transition.

02

Focus the offer

Frame the next review around the observed theme and your product’s compatibility.

03

Keep the evidence

Carry dates, freshness and an unanswered discovery question into the review brief.

FROM DOWNLOAD TO FIRST OUTPUT

Your five-minute
local quickstart.

Unzip the kit. Open a terminal beside the extracted folder.
Python 3.10+ must already be installed.

NO API KEY

Local quickstart

Python 3.10+ installed. Run the included case with no dependencies or network calls. The five-minute guide covers this local run.

cd signaltally-theme-upgrade-prospector
python3 run.py
# Open output/brief.md and output/evidence.json
AGENT RUNNER

LangChain

Python 3.10–3.13, a configured model and your Anthropic API key. The runner sends the selected evidence to your model provider; usage is billed by that provider.

cd signaltally-theme-upgrade-prospector
python3 -m venv .venv-langchain
.venv-langchain/bin/python -m pip install -r requirements-langchain.txt
# Set ANTHROPIC_API_KEY and SIGNALTALLY_MODEL in your environment.
.venv-langchain/bin/python run.py --runtime langchain --output output-langchain
ANALYST + REVIEWER

CrewAI

Python 3.10–3.13 and your configured Anthropic model. An analyst and a reviewer prepare a draft in sequence. Provider usage is separate.

cd signaltally-theme-upgrade-prospector
python3 -m venv .venv-crewai
.venv-crewai/bin/python -m pip install -r requirements-crewai.txt
# Set ANTHROPIC_API_KEY and SIGNALTALLY_MODEL in your environment.
.venv-crewai/bin/python run.py --runtime crewai --output output-crewai
TASK PROMPT

Viktor

Generate a task prompt locally. Paste it into your own Viktor conversation for review. A Viktor account is required for execution; this is a task template, not an installed connector.

cd signaltally-theme-upgrade-prospector
python3 run.py --runtime viktor --output output-viktor
# Paste output-viktor/viktor-task.md into your Viktor conversation.

Commands shown for macOS/Linux. On Windows use the virtual environment’s Scripts/python.exe. Run each framework in its own environment. The README covers credentials, current-date checks and replaying the included case. Generated drafts stay separate from the original evidence.

What happens when I run it?

The local runner reads the included evidence and writes a JSON record and Markdown brief on your device. Optional LangChain and CrewAI runs request a model-written draft. The Viktor option writes a prompt for you to paste into your own conversation.

Can I use my own data?

Yes. Supply a JSON file matching the included schema with --input. Dated observations and freshness checks stay attached. Licensed SignalTally delivery is scoped separately; the kit’s input format is not a commercial API contract.

Does it execute outreach?

It prepares research for review. The runners have no email, merchant-editing or browsing tools. Your team decides what to validate and how to act on the brief.

What is included for free?

The starter code, case data and setup guide. Full merchant audiences, refreshed intelligence and delivery are scoped under a paid plan. Model provider usage and Viktor access are separate.

THE KIT IS YOUR STARTING POINT

Run the example.
Build your advantage.

Put licensed commerce intelligence behind your workflow.
Agency · $2,000/month. Audience, refresh and delivery agreed before activation.

Plan your first market →