You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Checks that fail cause the test to fail. Checks that warn still pass (they're informational). Checks skipped due to unmet dependencies or config filtering show as skipped.
168
+
169
+
### Running a subset of checks
170
+
171
+
If your platform doesn't support certain checks (for example, you can't serve markdown), you can limit which checks run via the config:
132
172
133
173
```yaml
134
174
url: https://docs.example.com
135
175
checks:
136
176
- llms-txt-exists
137
177
- llms-txt-valid
138
178
- llms-txt-size
179
+
- http-status-codes
180
+
- auth-gate-detection
139
181
```
140
182
141
-
Then in your test file:
183
+
Only the listed checks will run. The rest show as skipped in the test output.
184
+
185
+
### Config resolution
186
+
187
+
The helpers look for `agent-docs.config.yml` (or `.yaml`) starting from `process.cwd()` and walking up the directory tree, so the config works whether your test file is at the project root or in a subdirectory. You can also pass an explicit directory:
188
+
189
+
```ts
190
+
describeAgentDocsPerCheck(__dirname);
191
+
```
192
+
193
+
### Timeouts
194
+
195
+
The helpers set a 120-second timeout on the check run automatically. No vitest timeout configuration is needed.
196
+
197
+
### Summary helper
198
+
199
+
If you don't need per-check granularity, `describeAgentDocs` provides a simpler two-test suite (one to run checks, one to assert no failures):
142
200
143
201
```ts
144
202
import { describeAgentDocs } from 'afdocs/helpers';
145
203
146
204
describeAgentDocs();
147
205
```
148
206
149
-
This reads the config and generates one test assertion covering all specified checks.
150
-
151
207
### Direct imports
152
208
209
+
For full control, use the programmatic API directly:
0 commit comments