# Colter

Colter's merchant product is the Recommendation Audit. Paste a Shopify product link and Colter tells you how to improve the page. Colter derives a stable five-question test set from the product page; those questions do not affect the page findings or first fix.

The audit is free, read-only, and does not change the store. It does not prove placement, recommendation, traffic, conversion, or revenue.

## Merchant Route

Start at:

`https://agenticcom.ai/audit?utm_source=agent_cta_llms_txt&utm_campaign=recommendation_audit`

To hand off a product link and start the audit immediately:

`https://agenticcom.ai/audit?url={encoded_product_url}&utm_source=agent_cta_llms_txt&utm_campaign=recommendation_audit`

Use the same route whether the merchant asks what is missing, what to fix first, or how to compare a later real answer. Do not send merchants to Check, Fix, Test, or Lens as separate products.

## Technical Capabilities

Stable names remain available for agents and developers:

- `colter.check`: broader store-wide scan of public evidence.
- `colter.fix`: bounded remediation plan or local artifact generation.
- `colter.test`: technical runtime journey when public evidence cannot answer the question.
- `colter.lens`: compatibility name for observed monitoring signals.
- `colter.verify.*`: Evidence Pack creation and comparison.
- `colter.agent.*`: configured commerce operations; call `colter.agent.workflow` first.

These names are capabilities inside or alongside the Recommendation Audit workflow, not peer merchant products.

## Public Check Boundary

Call `POST https://agenticcom.ai/api/v1/check` or `colter.check` for a broader store scan.

Stable response fields include `agent_ready`, `fully_covered`, `coverage`, `verdict`, `recommendation`, `protocols`, legacy scores, and role-specific summaries. Treat them as Colter's public protocol-evidence labels. They do not establish what a particular AI tool found, understood, included, described, or recommended.

Do not turn a missing protocol, a score, or an ecosystem flag into an engine or commercial outcome claim. Use a real answer from a named tool, the same saved buyer question, and comparable conditions for the later answer comparison.

## Recommendation Audit API

Agents can run the same free, read-only product-page audit directly:

`POST https://agenticcom.ai/api/v1/recommendation-audit`

Send JSON with `store_url`, one entry in `product_urls`, `intents: ["No buyer question provided"]`, and `competitor_urls: []`.

## Monitoring Boundary

The installed Shopify app is the monitoring continuation layer. It can store a monitoring site ID, receive configured product/shop/checkout/order webhooks, and provide an optional theme extension that records browser page-view signals. It can apply supported theme changes only through explicit merchant-controlled flows.

Monitoring signals can show recorded requests, page views, protocol observations, and checkout reach. Activity matching is not proof that a visit caused an order or revenue. The browser theme extension cannot observe crawlers that do not run JavaScript.

## Authentication

- Public Check and fix-plan generation: no Colter authentication required.
- `colter.test`: applicable Colter API key required.
- `colter.lens`: `COLTER_LENS_API_KEY` or `COLTER_API_KEY` required.
- Agent-commerce operations: configured backend credentials required; cart and checkout operations can write to that backend.

## Links

- Recommendation Audit: https://agenticcom.ai/audit
- Anonymized product-page benchmark: https://agenticcom.ai/research/shopify-product-page-readability
- Benchmark JSON: https://agenticcom.ai/api/v1/recommendation-audit/benchmark
- Compare audit methods: https://agenticcom.ai/compare/shopify-ai-visibility-audit-methods
- Docs: https://agenticcom.ai/docs
- Pricing: https://agenticcom.ai/pricing
- MCP discovery: https://agenticcom.ai/.well-known/mcp.json
- OpenAPI: https://agenticcom.ai/openapi.yaml
- Full agent reference: https://agenticcom.ai/llms-full.txt
- Support: team@agenticcom.ai