Metadata-Version: 2.4
Name: eclipse-agentready
Version: 0.1.0
Summary: AgentReady (T6 CLI edition) — audit a public storefront for AI shopping-agent readiness from CI: UCP profile, live MCP endpoint, llms.txt, robots AI policy, Product structured data, sitemap, platform, HTTPS/canonical.
Author: Cyan Eclipse Technologies LLC
License: MIT
Project-URL: Homepage, https://cetech.ai
Project-URL: Repository, https://github.com/CYAN-ECLIPSE-TECHNOLOGIES-LLC/agent-ready-mcp
Project-URL: Vendor, https://cetech.ai
Keywords: agent-readiness,storefront,e-commerce,ci,audit,mcp
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Internet :: WWW/HTTP
Classifier: Topic :: Software Development :: Quality Assurance
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

# eclipse-agentready

**What this is:** `eclipse-agentready` audits **storefronts** — public
e-commerce sites — for AI shopping-agent readiness: can an agent
(ChatGPT, Claude, Perplexity, Gemini…) discover the store, read its
catalog in machine-readable form, and get correct price/stock data?

**What this is not:** it does **not** audit repositories, and it does
**not** grade MCP servers. (The unrelated PyPI project named
`agentready` scores *code repositories* for AI-assisted development —
a different tool for a different job. This one is the CLI edition of
Cyan Eclipse Technologies' **AgentReady** storefront auditor, also
available as an MCP server.)

## Use — a CI step for merchants and storefront DevOps

```console
$ pip install eclipse-agentready
$ eclipse-agentready audit https://shop.example --fail-under 60
$ eclipse-agentready audit https://shop.example --format json --out report.json
$ eclipse-agentready list-checks
```

Exit codes: `0` = audit completed and (if `--fail-under N` given) the
score is at least N · `1` = score below N · `2` = the audit itself
could not run.

## The 8 checks (weights sum to 100)

| Check | Weight |
|---|---|
| Live MCP endpoint answers JSON-RPC | 20 |
| Product structured data (JSON-LD with price + availability) | 20 |
| UCP profile published (/.well-known/ucp) | 15 |
| HTTPS + canonical sanity | 12 |
| robots.txt allows AI crawlers | 10 |
| Sitemap present with product URLs | 10 |
| llms.txt present | 8 |
| Platform identified | 5 |

Scoring: pass = full weight, warn = half, fail = 0, n/a = excluded.
Grade: A ≥ 90 · B ≥ 75 · C ≥ 60 · D ≥ 40 · F < 40. Every check reports
observed evidence only, plus a concrete fix hint when it does not fully
pass. The CLI and the MCP server run the *identical* check catalog and
scoring path — same evidence in, same grade out.

## Honest scope

- **Public-surface audit only.** All checks read public endpoints with
  no credentials, a declared user agent, 10s timeouts, and redirect
  following.
- **No authenticated checkout execution** — this tool never logs in,
  never buys, and cannot verify a full agent purchase. A store can
  score well here and still fail at the checkout step (and vice versa).
- **No private data** is collected or stored; audit output is the
  caller's. Evidence is a point-in-time observation; storefronts change.
- A grade is not an endorsement and not a Bazaar listing claim.

The merchant's path to agent-commerce readiness starts with a grade
you can watch: run it in CI on every deploy. A hosted monitoring tier
(scheduled re-audits, score-change alerts) is on the product roadmap
and is **not yet launched**.

Powered by Cyan Eclipse Technologies LLC (cetech.ai) — a Veteran-Owned
Small Business. MIT licensed. Not affiliated with Shopify, WooCommerce,
Automattic, or any audited store; platform names describe detection
results only.
