docs: stop saying aux-audit does not exist

onboarding/appendix-a-vocabulary.md

@@ -14,7 +14,7 @@ Two tables. One for GitHub words (useful to R1 and R4 if they venture there). On
 | **PR / pull request** | A proposed change to a repo, visible and reviewable before it lands. |
 | **Fork** | Your own copy of a repo. What you do when you want to extend or dissent. |
 | **Star** | A bookmark. Used as a weak popularity signal. |
-| **CLI** | A tool you run from a terminal. `npx aux-audit run ./spec.yaml` is a CLI invocation. |
+| **CLI** | A tool you run from a terminal. `node packages/aux-audit/dist/cli.js run ./spec.yaml` is a CLI invocation; `npx aux-audit` is the same after npm publish. |
 | **CI / CI pipeline** | Automation that runs on every code change. Where `aux-audit` becomes merge-blocking. |
 | **SARIF** | A standard file format for code-scan findings. What CI tools read to surface violations. |
 | **YAML** | A human-readable config format. Every AUX schema is YAML. |

onboarding/for-ctos.md

@@ -4,7 +4,7 @@
 
 > **AUX** is an open standard for agentic UX published at **github.com/auxfirst** — schemas and heuristics are **CC BY 4.0**, tooling is **MIT**. Fork it, cite it, extend it.
 
-This is the 5-minute version. The long version is **[ARCHITECTURE.md](../ARCHITECTURE.md)**. The executable version is **[aux-audit](../repos/aux-audit/README.md)**.
+This is the 5-minute version. The long version is **[ARCHITECTURE.md](../ARCHITECTURE.md)**. The executable version is **[aux-audit](../packages/aux-audit/README.md)** (GitHub Action: `auxfirst/trustkit@v0.2`).
 
 ---
 
@@ -42,11 +42,20 @@ A Trust Contract without guarantees-backed-by-tests is a marketing page. Don't s
 
 `aux-audit` runs the 10 heuristics + 4 trust stages against an agent spec and emits a score, a grade, and named violations. It emits SARIF, which means it can block merges in CI the same way your static-analyzer does.
 
-> *v0.1 — CLI contract published, binary in active development. The command below describes the intended invocation; follow [aux-audit](../repos/aux-audit/README.md) for the ship date.*
+> *v0.1 ships as a GitHub Action. It grades the spec, not the running product.*
+
+```yaml
+- uses: auxfirst/trustkit@v0.2
+  with:
+    spec: ./spec.yaml
+    fail-on: high
+```
+
+Copy-paste workflow: **[forwardables/A3-aux-audit.yml](forwardables/A3-aux-audit.yml)**. From a clone of this repo, without npm:
 
 ```bash
-# v0.1 contract
-npx aux-audit run ./spec.yaml --format sarif --fail-on high
+npm --prefix packages/aux-audit ci && npm --prefix packages/aux-audit run build
+node packages/aux-audit/dist/cli.js run ./spec.yaml --format sarif --fail-on high
 ```
 
 What this gives you:

onboarding/for-engineering.md

@@ -36,16 +36,17 @@ Full schema: **[schemas/agent-spec.schema.yaml](../schemas/agent-spec.schema.yam
 
 ## 2 · Running aux-audit
 
-> **Status — v0.1 contract · CLI in development.** The commands below describe the v0.1 invocation. The binary isn't on npm yet; follow [aux-audit](../repos/aux-audit/README.md) for the ship date. You can still adopt the spec format today and wire the workflow file in advance.
+> **Status — v0.1 implemented.** In CI use the Action. Locally, clone this repo and build — `npx aux-audit` works after the package is on npm.
 
-Locally (v0.1 contract):
+Locally (from a clone of [auxfirst/trustkit](https://github.com/auxfirst/trustkit)):
 
 ```bash
-npx aux-audit run ./spec.yaml
+npm --prefix packages/aux-audit ci && npm --prefix packages/aux-audit run build
+node packages/aux-audit/dist/cli.js run ./spec.yaml
 # → score, grade, named gaps, recommendations
 ```
 
-In CI (v0.1 contract):
+In CI:
 
 ```yaml
 # .github/workflows/aux-audit.yml
@@ -58,9 +59,10 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
-      - uses: actions/setup-node@v4
-        with: { node-version: 20 }
-      - run: npx aux-audit run ./spec.yaml --format sarif --out audit.sarif --fail-on high
+      - uses: auxfirst/trustkit@v0.2
+        with:
+          spec: ./spec.yaml
+          fail-on: high
       - uses: github/codeql-action/upload-sarif@v3
         if: always()
         with: { sarif_file: audit.sarif }

packages/aux-audit/README.md

@@ -3,6 +3,11 @@
 > Score an agent product against the [10 AUX Heuristics](../../schemas/aux-heuristics.yaml) and the [Trust Architecture](../../schemas/trust-architecture.yaml). Score, grade, violations — reproducible.
 
 ```bash
+# today — from a clone of auxfirst/trustkit
+npm --prefix packages/aux-audit ci && npm --prefix packages/aux-audit run build
+node packages/aux-audit/dist/cli.js run ./agent-spec.yaml
+
+# after npm publish
 npx aux-audit run ./agent-spec.yaml
 ```
 
@@ -34,6 +39,16 @@ Two consequences, both deliberate:
 
 ## Install
 
+**Works today** (from a clone of this repository):
+
+```bash
+npm --prefix packages/aux-audit ci
+npm --prefix packages/aux-audit run build
+node packages/aux-audit/dist/cli.js run ./agent-spec.yaml
+```
+
+**After `npm publish`** (optional — the GitHub Action does not need this):
+
 ```bash
 npx aux-audit run ./agent-spec.yaml     # no install
 npm install --save-dev aux-audit        # or pin it

profile/README.md

@@ -24,7 +24,7 @@ If you ship AI products, you are no longer designing screens. You are designing
 | Entry point | What it is | Use when |
 |---|---|---|
 | **[aux-frameworks](../repos/aux-frameworks)** | Methodology as code. Audits, taxonomies, scorecards as YAML. | You want the standard. |
-| **[aux-audit](../repos/aux-audit)** | `npx aux-audit` — a CLI (v0.1 contract · binary in development) that scores an agent product against the 10 heuristics. | You want a grade, not an opinion. |
+| **[aux-audit](../packages/aux-audit)** | GitHub Action `auxfirst/trustkit@v0.2` — scores an agent spec against the 10 heuristics. | You want a grade, not an opinion. |
 | **[trust-gap-classifier](../repos/trust-gap-classifier)** | A taxonomy + detector for how agent trust breaks. | You want shared language for failures. |
 | **[agent-memory-policy](../repos/agent-memory-policy)** | Generate memory governance policy + system-prompt snippets from a config. | Product, compliance, and devs need to agree. |
 | **[agent-ux-teardowns](../repos/agent-ux-teardowns)** | Structured teardowns of shipped AI products. | You want receipts, not vibes. |
@@ -53,13 +53,21 @@ This is the **proof layer** for the category — every claim ships with a schema
 
 Every tool ships as a CLI. Every framework ships as YAML/JSON. Every claim ships with a reproducible scenario.
 
+```yaml
+- uses: auxfirst/trustkit@v0.2
+  with:
+    spec: ./agent-spec.yaml
+    fail-on: high
+```
+
+From a clone, without npm:
+
 ```bash
-# aux-audit v0.1 contract — CLI in active development
-npx aux-audit run ./agent-spec.json
-# → score: 72  grade: B  issues: [trust_gap, memory_amnesia]
+npm --prefix packages/aux-audit ci && npm --prefix packages/aux-audit run build
+node packages/aux-audit/dist/cli.js run ./agent-spec.yaml
 ```
 
-The spec format ([`agent-spec.schema.yaml`](../schemas/agent-spec.schema.yaml)) and rule set ([`aux-heuristics.yaml`](../schemas/aux-heuristics.yaml)) are published today. The binary follows — see [`aux-audit`](../repos/aux-audit).
+The spec format ([`agent-spec.schema.yaml`](../schemas/agent-spec.schema.yaml)) and rule set ([`aux-heuristics.yaml`](../schemas/aux-heuristics.yaml)) drive the CLI. See [`packages/aux-audit`](../packages/aux-audit).
 
 ## For product & design
 

repos/aux-audit/README.md

@@ -5,6 +5,11 @@
 > **Status — v0.1 implemented.** The CLI lives at **[`packages/aux-audit`](../../packages/aux-audit/)** in this repository: validator, rule set, scorer, and Markdown/JSON/SARIF output, with a GitHub Action wrapper at [`action.yml`](../../action.yml). From a clone: `npm --prefix packages/aux-audit ci && npm --prefix packages/aux-audit run build`. `npx aux-audit` works after the package is published to npm.
 
 ```bash
+# today — from a clone of this repo
+npm --prefix packages/aux-audit ci && npm --prefix packages/aux-audit run build
+node packages/aux-audit/dist/cli.js run ./agent-spec.yaml
+
+# after npm publish
 npx aux-audit run ./agent-spec.yaml
 ```
 
@@ -45,10 +50,27 @@ Because AUX claims need to be **inspectable and reproducible.** A score that can
 
 ## Install
 
+**Works today**
+
+```bash
+# CI
+- uses: auxfirst/trustkit@v0.2
+  with:
+    spec: ./agent-spec.yaml
+    fail-on: high
+
+# local, from a clone of auxfirst/trustkit
+npm --prefix packages/aux-audit ci
+npm --prefix packages/aux-audit run build
+node packages/aux-audit/dist/cli.js run ./spec.yaml
+```
+
+**After `npm publish`** (optional — not required for the Action)
+
 ```bash
 npm i -g aux-audit
 # or one-shot:
-npx aux-audit run ./spec.json
+npx aux-audit run ./spec.yaml
 ```
 
 ## Agent spec format
@@ -67,8 +89,11 @@ flows: ["./flows/handoff.md", "./flows/refund.md"]
 ## Use in CI
 
 ```yaml
-# .github/workflows/aux-audit.yml
-- run: npx aux-audit run ./spec.yaml --format sarif --fail-on high
+# .github/workflows/aux-audit.yml — copy-paste: onboarding/forwardables/A3-aux-audit.yml
+- uses: auxfirst/trustkit@v0.2
+  with:
+    spec: ./spec.yaml
+    fail-on: high
 ```
 
 Failing the build on `severity: high` makes AUX a merge-blocking check, not a slide.