# Goose ARWP — Get Found.

Canonical HTML: https://dkharlanau.github.io/agent-ready-web-profile/product/

**Goose ARWP** is the public product. **Agent-Ready Web Profile (ARWP)** is the technical foundation.

Goose helps a site become easier for Search and AI systems to discover, understand, verify, cite and route to useful actions. It does this as an evidence-backed operating loop rather than a one-time readiness score.

> **Tell Goose the goal. Inspect. Decide. Change. Prove. Watch.**

```text
WHAT SHOULD THIS SITE ACHIEVE?
            ↓
         INSPECT
            ↓
          DECIDE
        top 1–3 moves
            ↓
          CHANGE
       safely / reviewably
            ↓
           PROVE
    implementation + deployment
            ↓
          MEASURE
 provider-native Search / AI evidence
            ↓
       WATCH / REVISE
```

## What Goose answers

For a site owner, the useful questions are simple:

1. Why am I not being found for the things that matter?
2. What should I do first?
3. Which change can Goose prepare safely?
4. Is that exact change really live?
5. What Search/AI evidence changed afterwards?
6. Should I keep, revise, continue measuring, revert or retire the intervention?

Advanced technical modules remain available, but the first-run journey should not require choosing one.

## What makes Goose different

Goose keeps the evidence chain connected:

```text
OWNER GOAL / SITE BOUNDARY
          ↓
PUBLIC + REPOSITORY EVIDENCE
          ↓
APPLICABILITY
          ↓
RECOMMENDATION
          ↓
EXACT SOURCE / REPOSITORY SURFACE
          ↓
CHANGE
          ↓
VERIFICATION + DEPLOYMENT PARITY
          ↓
PROVIDER-NATIVE OUTCOME EVIDENCE
          ↓
REVIEW DECISION
```

The chain works backwards too. When upstream guidance changes or a recommendation becomes stale, Goose can identify which sites and prior changes need re-review.

## No Goose score

Search and AI systems expose different evidence. Goose does not flatten it into one 0–100 score.

A useful evidence funnel is:

```text
ACCESS
  ↓
INDEX / ELIGIBILITY
  ↓
EXPOSURE
  ↓
CITATION / REFERENCE
  ↓
BRAND MENTION
  ↓
VISIT
  ↓
USEFUL ACTION
```

Google generative-AI impressions are not Bing citations. Bing citations are not visits. OAI-SearchBot access is not a ChatGPT citation. Missing evidence stays unknown rather than becoming zero.

## Product priorities now

### P0 — Proof before more product

Goose currently prioritizes real site evidence loops over new architecture. The target is several owned sites with explicit experiment state, verified deployment parity, provider-native evidence and reviewed keep/revise/stop decisions. Neutral and negative outcomes remain visible.

### P0 — Portfolio Proof Board

Show what is actually planned, on hold, observing, reviewed or stopped across the portfolio from committed evidence rather than hand-written success claims.

### P0 — Provider-native Search & AI measurement

Keep Google, Bing, OpenAI/referral and other owner evidence semantically correct instead of manufacturing a universal AI-visibility metric.

### P1 — One first-run journey

A future first-run flow should accept a URL plus only the owner context that materially changes recommendations, then return a concise Get Found Brief with the top one to three next moves.

### P1 — Surface retirement

New pages/modules need a real user job, consumer, verification path and retirement/consolidation condition. Project maturity is not page count.

## Technical foundation

Behind the simple public journey, ARWP provides the deeper engines:

- Site Focus and Intent Ownership;
- Search/AI Growth and recommendation registries;
- Resolver and agent-web interoperability;
- Repository Mapper and stack-aware transformations;
- BraidGraph evidence-to-change lineage;
- Evidence and Change Receipts;
- Controlled Cohorts and Winner Observatory;
- Recommendation Review and Trend Intelligence;
- Project Maturity Surfaces and portfolio review tooling.

BraidGraph is an advanced internal primitive, not a user-facing requirement.

## Search / GEO boundary

Goose does not claim that special AI markup, `llms.txt`, schema volume or policy-page count creates ranking or citation gains. Google currently states that foundational SEO remains relevant to AI Overviews and AI Mode and that no special AI optimization is required for eligibility.

Goose therefore prioritizes:

- useful, unique, non-commodity content;
- intent ownership before URL expansion;
- genuine original data/tools/examples where they add standalone value;
- crawl/index eligibility;
- clear visible identity and fact parity;
- evidence/provenance and meaningful freshness;
- real provider-native measurement after deployment.

## Availability

- GitHub: https://github.com/dkharlanau/agent-ready-web-profile
- npm: https://www.npmjs.com/package/agent-ready-web-profile
- License: PolyForm Strict 1.0.0
- Source availability: public; commercial use, redistribution and derivative works require separate permission

## Key product docs

- `ROADMAP.md` — proof-first milestones and execution order.
- `docs/PRODUCT-LINE.md` — public journey, internal engines, packaging and stop rules.
- `research/product/2026-09-09-goose-self-interview.md` — multi-role product stress test behind the current priorities.
- `docs/BRAIDGRAPH.md` — advanced evidence-to-change graph design.
- `docs/MATURITY-PROFILE.md` — project identity/governance/evidence maturity pattern.

## Structured-data boundary

The public software entity should use truthful SoftwareApplication/WebApplication semantics. Do not fabricate ratings, reviews, awards, customer counts or other eligibility fields merely to make the project look mature.