Published 2026-09-22 · Tested 2026-09-22
Sentry
CSentry received 8 PASS votes and passed 4 of five agent surface checks. The clearest finding came from the find the exact limits task.
Three AI models, GPT 5.6 Sol, Claude Opus 5, and DeepSeek v4 Flash, each read Sentry’s public documentation independently and attempted five first-hour developer jobs: capture the first error, find the exact limits, recover from a 429, verify a webhook, use the JavaScript SDK.
No accounts, API calls, or code execution were used. Every verdict came from public pages and every published quotation passed a live verification check.
Freshness
How rechecks work- Category
- Observability & analytics
- Tested
- Quotes verified
- Surface rechecked
- Not yet rechecked
76.2% · 61/80 · AI Agent Readiness Score · reading 30 pts · surface 50 pts
Agent surface checks · 40/50
The Reading Test
| Task | GPT 5.6 Sol | Opus 5 | DeepSeek v4F | Consensus |
|---|---|---|---|---|
| Capture the first error | PASS | PASS | PASS | PASS |
| Find the exact limits | FAIL | FAIL | PARTIAL | FAIL |
| Recover from a 429 | PARTIAL | PASS | PARTIAL | PARTIAL |
| Verify a webhook | PASS | PASS | PASS | PASS |
| Use the JavaScript SDK | PARTIAL | PARTIAL | PASS | PARTIAL |
docs platform: Custom Next.js (unscored) · verified 2026-09-22
What to fix first
These 4 fixes could add up to 19 points to the AI Agent Readiness Score. The list ranks each fix by the points it would add. How the ranking works
- 1 +10 points llms-full.txt check failed
Found: HTTP 404 returning the documentation HTML shell.
Fix: Publish llms-full.txt at the docs root with the full text of every docs page in one plain-text file.
- 2
Found: The API rate limits page names no request number or window length for any endpoint, and no docs page states the Developer plan quota.
Fix: Publish the request maximum and window size for each endpoint on the API rate limits page, and list the Developer plan quota beside the paid plans.
Evidence: docs.sentry.io/
api/ ratelimits (opens in a new tab) - 3
Found: The quota guide names both 429 causes and the Retry-After header, but splits the recovery steps across pages and never gives one ordered correction.
Fix: Add one ordered recovery procedure to the manage event stream guide that runs from the 429 response through Retry-After to increasing the quota or filtering events.
Evidence: docs.sentry.io/
pricing/ quotas/ manage-event-stream-guide (opens in a new tab) - 4
Found: The JavaScript usage page ships the import line with an unsubstituted placeholder instead of the package name, so the name comes from another page.
Fix: Replace the placeholder in the import line on the JavaScript usage page with @sentry/browser so the captureException example imports a real package name.
Evidence: docs.sentry.io/
platforms/ javascript/ usage (opens in a new tab)
What the docs get right
- Capture the first error: 3 PASS votes. Four legacy Browser guide URLs failed, but the current JavaScript page documents installation, initialization, testing, and issue viewing.
- Verify a webhook: 3 PASS votes. The integration guide covers creation, endpoint configuration, event payloads, delivery headers, response timing, and HMAC-SHA256 verification code.
- 4 of 5 agent surface checks. Present: llms.txt, markdown mirrors, an MCP server, docs AI. Missing: llms-full.txt.
Capture the first error
PASSPASS consensus from 3 PASS.
Four legacy Browser guide URLs failed, but the current JavaScript page documents installation, initialization, testing, and issue viewing. One page carries prerequisites, install, Sentry.init with a DSN, and a throwing-button verification step, so a zero-to-first-error path is complete without leaving it. The JavaScript SDK page documents a complete zero-to-first-error path: install via npm, initialize with a DSN, then verify with a button that throws an error and check the Issues page.
Find the exact limits
FAILFAIL consensus from 1 PARTIAL, 2 FAIL.
The API page gives headers but no numeric endpoint limits; pricing gives paid baselines but omits the Developer plan quota. The API rate limit page names no requests-per-second or concurrency number for any endpoint, and the quota pages state no per-plan event allowance, deferring instead to the off-docs sentry.io/pricing/ page. Plan quotas are exact (50k errors, 5GB logs, 5M spans per paid plan) and consistent across the pricing and quota pages, but the API rate limits page gives no exact numeric limits for any endpoint, just says each has its own window and maximum.
Recover from a 429
PARTIALPARTIAL consensus from 1 PASS, 2 PARTIAL.
The obsolete miscellaneous page failed; current Usage and quota pages identify causes and require cross-referencing for cause-specific recovery. The same page names both causes of a 429 (exceeding the org quota and exceeding a per-key project rate limit), the Retry-After header, the instruction not to retry, and the increase-quota and filtering routes to a fix. The docs name two causes (quota exceeded, rate limit hit) and give a partial recovery path (increase quota, drop events until Retry-After expires), but the recovery actions are split across pages and a single complete step-by-step correction from recognizing a 429 to resolving it is not presented in one place.
Verify a webhook
PASSPASS consensus from 3 PASS.
The integration guide covers creation, endpoint configuration, event payloads, delivery headers, response timing, and HMAC-SHA256 verification code. Creation under Settings > Developer Settings with a webhook URL, the full header set, the resource and action payload shape, and a runnable HMAC-SHA256 verification snippet are all documented across two linked pages. The integration platform docs tell the developer to create an integration at Settings > Developer Settings, specify the webhook URL, then the webhooks page documents the Sentry-Hook-Signature header and provides a complete code snippet to verify it with HMAC-SHA256.
Use the JavaScript SDK
PARTIALPARTIAL consensus from 1 PASS, 2 PARTIAL.
The official package and API agree, but the capture example uses <sdk-package-name> and depends on setup from another page. The captureException example is correct and matches the Sentry.captureException form used elsewhere, but its import line ships an unsubstituted build token instead of a package name, and that page never states the package, so the name has to be cross-referenced from the getting-started page. The official @sentry/browser package is identified on the JavaScript platform page, and the usage page provides a complete minimal working example (import, try-catch, Sentry.captureException) that matches the Sentry.init API form shown on the setup page.
The receipt
Each endpoint has its own maximum number of requests and window size.
The API page gives headers but no numeric endpoint limits; pricing gives paid baselines but omits the Developer plan quota.
Agent surface notes
Initialize returned HTTP 401 with a Bearer OAuth challenge naming resource metadata at mcp.sentry.dev/.well-known/oauth-protected-resource/mcp.
Clicking the docs header Ask AI button opened the Kapa widget shadow root with the input placeholder Ask me a question about Sentry.
Show the score
Paste this into a readme:
[](https://docsforagents.com/reports/sentry-docs-ai-agent-readiness/) Method note
This is a reading test of public documentation, not an execution test. No accounts were created and no API calls were run. The AI Agent Readiness Score counts fifteen reading votes at PASS 2, PARTIAL 1, and FAIL 0, for 30 possible points. Five agent surface checks add 10 points each. The total is 80. Consensus chips show each row majority and do not affect scoring. The panel split on 3 of five tasks. Quotes shown here were re-fetched and confirmed verbatim on 2026-09-22.
Put another docs site through the battery.
Nominate a docs site