# Recall.ai docs: AI Agent Readiness Score 80.0% (B-)

**80.0% · 64/80 · AI Agent Readiness Score · 24/30 reading points · 40/50 agent surface points**

Recall.ai received 9 PASS votes and passed 4 of five agent surface checks. The clearest finding came from the find the exact limits task.

- Tested: 2026-08-19
- Published: 2026-09-21
- Battery: v1
- Scoring: reading 30 pts · surface 50 pts
- Docs: https://docs.recall.ai/

Three AI models, GPT 5.6 Sol, Claude Opus 5, and DeepSeek v4 Flash, each read Recall.ai’s public documentation independently and attempted five first-hour developer jobs: record the first meeting, find the exact limits, recover from 429 and 507, verify a bot webhook, list bots with pagination.

No accounts, API calls, or code execution were used. Every verdict came from public pages and every published quotation passed a live verification check. Preflight substitutions: T5: Pagination and filtering on List Bots replaced SDK coverage because no primary Meeting Bots SDK is documented.

## Freshness

[How rechecks work](https://docsforagents.com/methodology/#freshness)

- Category: [Voice & speech](https://docsforagents.com/grades/?category=voice)
- Tested: 2026-08-19
- Quotes verified: 2026-08-19
- Surface rechecked: 2026-09-21

No change since the test.

5 of 5 quoted passages still appear on the live pages.

## Agent surface checks · 40/50

| Check | Verdict | Points |
| --- | --- | --- |
| llms.txt | PASS | 10 |
| llms-full.txt | FAIL | 0 |
| Markdown mirror | PASS | 10 |
| MCP server | PASS | 10 |
| Docs AI | PASS | 10 |

## The Reading Test · 24/30

| Task | GPT 5.6 Sol | Opus 5 | DeepSeek v4F | Consensus |
| --- | --- | --- | --- | --- |
| Record the first meeting | PASS | PASS | PASS | PASS |
| Find the exact limits | PARTIAL | PARTIAL | PASS | PARTIAL |
| Recover from 429 and 507 | PASS | PASS | PARTIAL | PASS |
| Verify a bot webhook | PASS | PASS | PASS | PASS |
| List bots with pagination | PARTIAL | PARTIAL | PARTIAL | PARTIAL |

Docs platform: ReadMe (unscored) · verified 2026-08-19

## What to fix first

These 3 fixes could add up to 15 points to the AI Agent Readiness Score. The list ranks each fix by the points it would add. [How the ranking works](https://docsforagents.com/methodology/#what-to-fix-first)

1. **+10 points · llms-full.txt · check failed**

   **Found:** HTTP 404 with the ReadMe not-found page.

   **Fix:** Publish llms-full.txt at the docs root with the full text of every docs page in one plain-text file.

   **Evidence:** [docs.recall.ai/llms-full.txt](https://docs.recall.ai/llms-full.txt)

2. **+3 points · List bots with pagination · PARTIAL**

   **Found:** List Bots gives use_cursor no accepted value, and its response schema shows only offset pagination. In the panel's reading, its date filter format also conflicts with the scheduling page.

   **Fix:** On List Bots, document the use_cursor value, the cursor-mode response, a traversal example, and one join_at_after format.

   **Evidence:** [docs.recall.ai/reference/bot_list](https://docs.recall.ai/reference/bot_list)

3. **+2 points · Find the exact limits · PARTIAL**

   **Found:** The errors page sends readers to the dashboard for ad-hoc concurrency, while the scheduling page cites a 30-bot limit.

   **Fix:** On the scheduling page, state whether 30 is the default ad-hoc limit, and point to Developers > Rate Limits.

   **Evidence:** [docs.recall.ai/docs/creating-and-scheduling-bots](https://docs.recall.ai/docs/creating-and-scheduling-bots)

## What the docs get right

- **Record the first meeting: 3 PASS votes.** The quickstart covers authentication, bot creation, done-state detection, bot retrieval, and the exact media shortcut download path.
- **Verify a bot webhook: 3 PASS votes.** The dashboard setup, event subscription, POST receipt, workspace secret, signed headers, raw-body HMAC, and constant-time comparison are documented.
- **4 of 5 agent surface checks.** Present: llms.txt, markdown mirrors, an MCP server, docs AI. Missing: llms-full.txt.

## Record the first meeting

**PASS**

PASS consensus from 3 PASS.

The quickstart covers authentication, bot creation, done-state detection, bot retrieval, and the exact media shortcut download path. The quickstart runs API key to create-bot curl to done status to recordings[0].media_shortcuts.video_mixed.data.download_url in eight numbered steps with a runnable request and a response sample at each stage. The quickstart maps every step from API key creation through bot creation, waiting for the done status, and retrieving the download_url from media_shortcuts in a single linear path without gaps.

## Find the exact limits

**PARTIAL**

PARTIAL consensus from 1 PASS, 2 PARTIAL.

Create is 120/min/workspace, list is 60/min/workspace, and scheduled concurrency is unlimited, but ad-hoc concurrency is workspace-specific versus 30. Create Bot (120 requests per min per workspace), List Bots (60 requests per min per workspace), and the scheduled-bot concurrency answer (no limit above 10 minutes ahead) are all stated exactly, but the ad-hoc concurrency limit is sent to the dashboard on the errors page while the scheduling page asserts a specific value of 30. Each endpoint documents its own rate limit with scope (per workspace, per min). Scheduled bots have no concurrent limit; ad-hoc bots have a default of 30 concurrent visible in the dashboard. Values and scopes are internally consistent.

## Recover from 429 and 507

**PASS**

PASS consensus from 2 PASS, 1 PARTIAL.

The error guide uses Retry-After for both codes, adds 30-second retries for 507, and recommends scheduling over 10 minutes ahead. The errors page gives Retry-After handling for both codes, states that rate limits are enforced with a sliding window, scopes 507 to ad-hoc bots only, prescribes a 30-second retry loop, and the scheduling page adds the delete-and-recreate correction for a join_at change inside the 10-minute window. 429 handling is fully documented with Retry-After. For 507, the errors page says to use Retry-After but also says 'retry in 30 seconds', and the canonical retry code hardcodes 30s without reading the header, a contradiction that leaves the agent unsure which value to trust.

## Verify a bot webhook

**PASS**

PASS consensus from 3 PASS.

The dashboard setup, event subscription, POST receipt, workspace secret, signed headers, raw-body HMAC, and constant-time comparison are documented. Dashboard endpoint setup, the full bot status event list and payload schema, and a complete TypeScript verifier covering the whsec_ base64 key decode, the id.timestamp.payload signing string, HMAC-SHA256, and a timing-safe compare are all documented, including the separate Svix secret rule for workspaces created before 12/15/2025. The complete flow is documented: create workspace secret in dashboard, subscribe to bot.* events via Svix webhook in dashboard, receive webhook with HMAC headers, and verify using the provided TypeScript function with timing-safe comparison.

## List bots with pagination

**PARTIAL**

PARTIAL consensus from 3 PARTIAL.

The endpoint exposes time filters and cursor mode, but conflicts on date format and gives no cursor traversal example. Filtering is fully specified (join_at_after, join_at_before, meeting_url, platform, status, metadata__<key>), but cursor traversal is not: use_cursor is typed only as a string with no accepted value, and the documented 200 response schema PaginatedBotGetList returns count, next, and previous with offset examples such as ?page=4 and carries no cursor field. Time-window filtering (join_at_after, join_at_before) and offset pagination (page parameter with next URL) are documented. Cursor-based pagination is mentioned via a use_cursor parameter but with no explanation of what value it accepts, how the response differs, or how to traverse pages, the schema only shows offset-style next/previous fields.

## The receipt

> There are no limits on concurrent bots live in meetings scheduled more than 10 minutes in advance.

Create is 120/min/workspace, list is 60/min/workspace, and scheduled concurrency is unlimited, but ad-hoc concurrency is workspace-specific versus 30.

- [docs.recall.ai/docs/creating-and-scheduling-bots](https://docs.recall.ai/docs/creating-and-scheduling-bots)

## Agent surface notes

Initialize returned a valid OAuth-protected MCP challenge with documented scopes.

The live docs expose a public docs-only Ask AI control.

## 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 2 of five tasks. Quotes shown here were re-fetched and confirmed verbatim on 2026-08-19.

Methodology: https://docsforagents.com/methodology/

Canonical URL: https://docsforagents.com/reports/recall-ai-docs-ai-agent-readiness/