# Grab — API Agent Score

> Score: 82/100 (Grade: A) | Domain: grab.com | Rubric: 1.0.0 | Checked: November 9, 2026

Grab scored 82/100 (A), classified "ai-ready".

[View full report](https://www.postman.com/ai/ai-ready-apis/company/grab)
[All organizations](https://www.postman.com/ai/ai-ready-apis/llms.txt)

## Summary

- Overall: 82/100 (A)
- Classification: ai-ready

## Category Scores

- API Design: 97/100 (A+)
- Developer Experience: 65/100 (C)
- Agent Discovery: 80/100 (A)
- Agent Understanding: 92/100 (A+)
- Agent Usability: 76/100 (B)

## Check Results

### API Design

- [pass] Machine-readable, versioned contract — The API declares version "1.1.3" and uses url-based versioning, so an agent can pin to a stable contract. Investigated: spec 100%, docs 100%.
- [pass] Auth declared & discoverable — The MCP server at https://grabgpu.com/mcp accepts unauthenticated requests, so an agent knows it needs no credentials. Investigated: mcp 100%, spec 0%, docs 0%.
- [warn] Schema coverage & depth — 0/1 (0%) spec-enum params are realized as constrained types (not bare strings). Investigated: sdk 86%, spec 50%.
- [pass] Security & governance hygiene — Code is formatter-clean with zero linter errors and zero style warnings. Investigated: sdk 100%, wellknown 50%, spec 0%.
- [pass] Example coverage — 100% of parameters and responses (119 of 119) include example values, grounding an agent in real payload shapes. Investigated: spec 100%, sdk 100%, docs 0%.

### Developer Experience

- [pass] Self-service developer portal — Signup page found at https://www.grab.com/sg/start/, but no free tier or sandbox is documented in docs/pricing, so an agent may hit a paywall. Investigated: docs 100%.
- [pass] Description completeness — 1036/1036 operations carry a description. Investigated: sdk 100%, spec 0%.
- [warn] Quickstart present — Quickstart page is reachable at https://developers.grab.com/docs/quickstart/ (22 variants scanned) but has no runnable code sample in its HTML or .md variant, so there's nothing to copy and run. Investigated: docs 50%.
- [fail] Code samples in docs — No code samples detected across 1 sampled docs pages, so a coding agent gets no ready-to-use examples. Investigated: docs 0%.
- [fail] Changelog published — No changelog link or mention appears in the spec, so an agent can't tell what changed before something breaks. Investigated: spec 0%, docs 0%.

### Agent Discovery

- [pass] Docs reachable, not hard auth-gated — All 1 pages are publicly accessible. Investigated: docs 100%.
- [pass] Registry & SDK presence — Indexed on Context7 (cavaliergopher/grab, 290 snippets). Investigated: docs 100%, sdk 100%, cli 0%, mcp 0%, wellknown 0%.
- [warn] llms.txt present, valid & comprehensive — No llms.txt found at any candidate location (https://developers.grab.com/docs/llms.txt, https://developers.grab.com/llms.txt). Investigated: docs 50%, sdk 0%.
- [warn] Crawlable / AEO — robots.txt blocks every monitored AI crawler from the docs paths, so agents can't cite or read your documentation. Investigated: wellknown 25%.

### Agent Understanding

- [pass] Machine-readable errors (RFC 9457) — Documents 7 distinct 4xx/5xx error response codes, so an agent can branch on specific failures instead of guessing. Investigated: docs 100%, sdk 100%, spec 50%.
- [pass] Operation purpose clarity — 100% of operations (30 of 30) have both a clear summary and a descriptive name, so an agent can pick the right endpoint. Investigated: spec 100%, sdk 100%.
- [pass] Agent-navigable, token-efficient docs — All 1 pages contain server-rendered content. Investigated: docs 100%.
- [warn] Agent instructions file (AGENTS.md) — No agents.md or skill.md context file found (nice-to-have for agent operation). Investigated: sdk 50%, wellknown 0%.
- [warn] Docs structured data — Only OpenGraph/meta tags across 1 assessed pages — no JSON-LD for AI answer engines to cite. Investigated: docs 50%.
- [skip] Description consistency across surfaces — Only 1 surface description(s) with ≥6 tokens available; need at least 2 to compare.

### Agent Usability

- [pass] Idempotency documented — Built-in retry machinery is present (grep-derived). Investigated: sdk 100%, spec 0%, docs 0%.
- [pass] Sandbox separation — Spec declares a sandbox server but no distinguishable test credentials, so agents can't tell test calls from live ones. Investigated: spec 100%, docs 100%, sdk 100%.
- [pass] Runnable collection with test scripts — README is present and includes a runnable quickstart. Investigated: sdk 100%, platform 0%.
- [fail] Rate-limit signaling — A 429 response is documented, but no rate-limit response headers accompany it, so an agent can only react to a throttle instead of anticipating one. Investigated: spec 0%, docs 0%.
- [skip] Pagination documented & consistent — No surface produced evidence for this capability in this run.

## Executive Summary

Grab's API program shows strong fundamentals in design, agent understanding, and discoverability — partners and agents can find and parse the API surface reliably. The two areas that need immediate attention are developer experience and operational reliability: there is no published changelog, so partners and their agents cannot anticipate breaking changes, and there are no rate-limit signaling headers, meaning agents can only react to throttling after it has already disrupted a workflow. Prioritizing a changelog and proactive rate-limit headers, alongside adding ready-to-run code samples, will materially shorten the time partners spend integrating and reduce the support burden on your team.