# AgentScore — catalogue complet

Rubrique v0.2.0 (draft). 18 outil(s).

## Criteria

| Criterion | Weight | Method | Question |
| --- | --- | --- | --- |
| Offload Value | 15 | judged | Does the service let the agent hand off work that would be expensive to do itself through inference, or is it just CRUD in disguise? |
| Interaction Cost | 15 | measured | How many tokens does it cost me just to TALK to this API to complete its category's canonical task? |
| Error Recoverability | 15 | probed | When I hallucinate a parameter, does the error let me correct myself on the next turn, or do I give up? |
| Doc Legibility | 12 | measured | Can I read this documentation without running JavaScript, and without spending 80% of my tokens on markup? |
| Auth Friction | 10 | measured | How many steps between "I have a key" and "my first call succeeds"? |
| Safety & Reversibility | 10 | probed | If I retry after a timeout, do I charge the customer twice? |
| Payload Efficiency | 8 | probed | Can I ask for only the fields I need, or do I have to swallow 200 attributes to extract one? |
| Task Atomicity | 6 | measured | How many calls for a single business intent? |
| Prior Knowledge Coverage | 5 | measured | Do I already know this API by heart, or will I confidently hallucinate an outdated version? |
| Async Compatibility | 2 | probed | If the operation is asynchronous, can I get the result without hosting a web server? |
| Tool-Native (MCP) | 2 | binary | Does the vendor publish an official, maintained MCP server? |

A criterion missing from an entry is excluded from the computation and the weights are renormalised. It is never counted as a zero.

---
# Appwrite

> Open-source backend for web and mobile apps.

- **Category**: backend-database
- **Éditeur** : Appwrite
- **Site** : https://appwrite.io
- **Pricing model**: freemium
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Appwrite API `appwrite-rest-v1`

- **Type** : rest
- **Version** : v1
- **Documentation** : https://appwrite.io/docs/references/cloud/server-rest/databases

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 35.5 / 100
**Coverage**: 86 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 78 |
| Offload Value | 15 | 35 |
| Doc Legibility | 12 | 17 |
| Task Atomicity | 6 | 27 |
| Interaction Cost | 15 | 17 |
| Payload Efficiency | 8 | 5 |
| Error Recoverability | 15 | 39 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/appwrite

---

# Braintree

> PayPal's payment gateway, with a GraphQL API.

- **Category**: payments
- **Éditeur** : Braintree (PayPal)
- **Site** : https://www.braintreepayments.com
- **Pricing model**: per_transaction
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Braintree GraphQL API `braintree-graphql`

- **Type** : graphql
- **Base URL** : https://payments.braintree-api.com/graphql
- **Documentation** : https://developer.paypal.com/braintree/graphql/
- **Authentification** : basic

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 58.7 / 100
**Coverage**: 88 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 85 |
| Offload Value | 15 | 94 |
| Doc Legibility | 12 | 17 |
| Task Atomicity | 6 | 50 |
| Interaction Cost | 15 | 38 |
| Error Recoverability | 15 | 61 |
| Safety & Reversibility | 10 | 42 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/braintree

---

# Cloudflare D1

> Serverless SQLite databases on Cloudflare's network.

- **Category**: backend-database
- **Éditeur** : Cloudflare
- **Site** : https://www.cloudflare.com
- **Pricing model**: freemium
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Cloudflare API — D1 `cloudflare-d1-rest-v4`

- **Type** : rest
- **Version** : v4
- **Base URL** : https://api.cloudflare.com/client/v4
- **Documentation** : https://developers.cloudflare.com/api/resources/d1/subresources/database/methods/query/
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 46.2 / 100
**Coverage**: 78 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 30 |
| Doc Legibility | 12 | 41 |
| Task Atomicity | 6 | 57 |
| Interaction Cost | 15 | 18 |
| Error Recoverability | 15 | 40 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/cloudflare-d1

---

# Exa

> Web search and content retrieval built for AI.

- **Category**: web-search
- **Éditeur** : Exa
- **Site** : https://exa.ai
- **Pricing model**: usage_based
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Exa API `exa-rest`

- **Type** : rest
- **Base URL** : https://api.exa.ai
- **Documentation** : https://exa.ai/docs/reference/search
- **Authentification** : api_key_header

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 80.8 / 100
**Coverage**: 78 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 90 |
| Offload Value | 15 | 92 |
| Doc Legibility | 12 | 57 |
| Task Atomicity | 6 | 100 |
| Interaction Cost | 15 | 100 |
| Error Recoverability | 15 | 49 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/exa

---

# Firebase Firestore

> Real-time document database.

- **Category**: backend-database
- **Éditeur** : Google
- **Site** : https://firebase.google.com
- **Pricing model**: freemium
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Firestore REST API `firebase-firestore-rest-v1`

- **Type** : rest
- **Version** : v1
- **Documentation** : https://firebase.google.com/docs/firestore/use-rest-api

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 61.8 / 100
**Coverage**: 86 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 80 |
| Offload Value | 15 | 35 |
| Doc Legibility | 12 | 31 |
| Task Atomicity | 6 | 100 |
| Interaction Cost | 15 | 100 |
| Payload Efficiency | 8 | 33 |
| Error Recoverability | 15 | 50 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/firebase

---

# Firecrawl

> Turn websites into LLM-ready data, with search built in.

- **Category**: web-search
- **Éditeur** : Firecrawl
- **Site** : https://www.firecrawl.dev
- **Pricing model**: usage_based
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Firecrawl API `firecrawl-rest-v1`

- **Type** : rest
- **Version** : v1
- **Base URL** : https://api.firecrawl.dev/v1
- **Documentation** : https://docs.firecrawl.dev/api-reference/endpoint/search
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 74.1 / 100
**Coverage**: 78 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 90 |
| Doc Legibility | 12 | 48 |
| Task Atomicity | 6 | 75 |
| Interaction Cost | 15 | 75 |
| Error Recoverability | 15 | 55 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/firecrawl

---

# Paddle

> Merchant of record and billing for software companies.

- **Category**: payments
- **Éditeur** : Paddle
- **Site** : https://www.paddle.com
- **Pricing model**: per_transaction
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Paddle API `paddle-rest-v1`

- **Type** : rest
- **Version** : v1
- **Base URL** : https://api.paddle.com
- **Documentation** : https://developer.paddle.com/api-reference/overview

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 58.7 / 100
**Coverage**: 77 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 92 |
| Doc Legibility | 12 | 17 |
| Payload Efficiency | 8 | 53 |
| Async Compatibility | 2 | 100 |
| Error Recoverability | 15 | 63 |
| Safety & Reversibility | 10 | 42 |
| Prior Knowledge Coverage | 5 | 0 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/paddle

---

# Pinecone

> Managed, serverless vector database.

- **Category**: vector-database
- **Éditeur** : Pinecone
- **Site** : https://www.pinecone.io
- **Pricing model**: usage_based
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Pinecone API `pinecone-rest`

- **Type** : rest
- **Base URL** : https://api.pinecone.io
- **Documentation** : https://docs.pinecone.io/reference/api/introduction
- **Authentification** : api_key_header

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 62.5 / 100
**Coverage**: 71 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 90 |
| Doc Legibility | 12 | 17 |
| Task Atomicity | 6 | 75 |
| Interaction Cost | 15 | 80 |
| Payload Efficiency | 8 | 52 |
| Error Recoverability | 15 | 51 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/pinecone

---

# Postmark

> Transactional email with high deliverability.

- **Category**: transactional-email
- **Éditeur** : Postmark
- **Site** : https://postmarkapp.com
- **Pricing model**: per_email
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Postmark API `postmark-rest-v1`

- **Type** : rest
- **Base URL** : https://api.postmarkapp.com
- **Documentation** : https://postmarkapp.com/developer/api/overview

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 59.0 / 100
**Coverage**: 98 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 90 |
| Offload Value | 15 | 91 |
| Doc Legibility | 12 | 50 |
| Task Atomicity | 6 | 33 |
| Interaction Cost | 15 | 65 |
| Payload Efficiency | 8 | 5 |
| Async Compatibility | 2 | 100 |
| Error Recoverability | 15 | 39 |
| Safety & Reversibility | 10 | 42 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/postmark

---

# Qdrant

> Open-source vector search engine, managed in Qdrant Cloud.

- **Category**: vector-database
- **Éditeur** : Qdrant
- **Site** : https://qdrant.tech
- **Pricing model**: freemium
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Qdrant REST API `qdrant-cloud-rest`

- **Type** : rest
- **Documentation** : https://api.qdrant.tech/api-reference
- **Authentification** : api_key_header

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 69.5 / 100
**Coverage**: 88 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Tool-Native (MCP) | 2 | 100 |
| Auth Friction | 10 | 90 |
| Offload Value | 15 | 84 |
| Doc Legibility | 12 | 17 |
| Task Atomicity | 6 | 100 |
| Interaction Cost | 15 | 100 |
| Payload Efficiency | 8 | 52 |
| Error Recoverability | 15 | 36 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/qdrant

---

# Resend

> Transactional email API for developers.

- **Category**: transactional-email
- **Éditeur** : Resend
- **Site** : https://resend.com
- **Pricing model**: freemium_per_email
- **Rubrique** : v0.2.0

Modern email API, popular with the new generation of developer tools. Its API surface is deliberately small.

## Surfaces evaluated

### Resend API (REST) `resend-rest-v1`

- **Type** : rest
- **Version** : v1
- **Base URL** : https://api.resend.com
- **Documentation** : https://resend.com/docs/api-reference/introduction
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 76.4 / 100
**Coverage**: 98 %
**Measured**: 2026-09-21 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 92 |
| Doc Legibility | 12 | 54 |
| Task Atomicity | 6 | 100 |
| Interaction Cost | 15 | 100 |
| Payload Efficiency | 8 | 63 |
| Async Compatibility | 2 | 100 |
| Error Recoverability | 15 | 42 |
| Safety & Reversibility | 10 | 58 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/resend

---

# Twilio SendGrid

> Email delivery platform at scale.

- **Category**: transactional-email
- **Éditeur** : Twilio
- **Site** : https://www.twilio.com
- **Pricing model**: freemium_per_email
- **Rubrique** : v0.2.0

## Surfaces evaluated

### SendGrid API `sendgrid-rest-v3`

- **Type** : rest
- **Version** : v3
- **Base URL** : https://api.sendgrid.com/v3
- **Documentation** : https://www.twilio.com/docs/sendgrid/api-reference

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 52.6 / 100
**Coverage**: 98 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 93 |
| Doc Legibility | 12 | 17 |
| Task Atomicity | 6 | 50 |
| Interaction Cost | 15 | 51 |
| Payload Efficiency | 8 | 5 |
| Async Compatibility | 2 | 0 |
| Error Recoverability | 15 | 39 |
| Safety & Reversibility | 10 | 42 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/sendgrid

---

# Square

> Payments, point of sale and commerce APIs for sellers.

- **Category**: payments
- **Éditeur** : Square
- **Site** : https://squareup.com
- **Pricing model**: per_transaction
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Square API `square-rest-v2`

- **Type** : rest
- **Version** : v2
- **Base URL** : https://connect.squareup.com/v2
- **Documentation** : https://developer.squareup.com/reference/square
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 75.0 / 100
**Coverage**: 98 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 96 |
| Doc Legibility | 12 | 17 |
| Task Atomicity | 6 | 100 |
| Interaction Cost | 15 | 100 |
| Payload Efficiency | 8 | 5 |
| Async Compatibility | 2 | 100 |
| Error Recoverability | 15 | 61 |
| Safety & Reversibility | 10 | 100 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/square

---

# Stripe

> Payment infrastructure for the internet.

- **Category**: payments
- **Éditeur** : Stripe
- **Site** : https://stripe.com
- **Pricing model**: per_transaction
- **Rubrique** : v0.2.0

Reference payments API, often cited as the gold standard of API documentation.

## Surfaces evaluated

### Stripe API (REST) `stripe-rest-v1`

- **Type** : rest
- **Version** : v1
- **Base URL** : https://api.stripe.com/v1
- **Documentation** : https://docs.stripe.com/api
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 77.2 / 100
**Coverage**: 98 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 97 |
| Doc Legibility | 12 | 24 |
| Task Atomicity | 6 | 100 |
| Interaction Cost | 15 | 67 |
| Payload Efficiency | 8 | 46 |
| Async Compatibility | 2 | 100 |
| Error Recoverability | 15 | 80 |
| Safety & Reversibility | 10 | 100 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/stripe

---

# Supabase

> Postgres backend with an auto-generated REST API.

- **Category**: backend-database
- **Éditeur** : Supabase
- **Site** : https://supabase.com
- **Pricing model**: freemium
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Supabase REST (PostgREST) `supabase-rest-v1`

- **Type** : rest
- **Version** : v1
- **Documentation** : https://supabase.com/docs/guides/api

**Statut de notation** : `partial`

Partial benchmark: coverage is below the publication threshold, so no overall score is computed. The criteria present in `criteria` are valid and usable individually.

**Overall score**: not published (insufficient coverage) / 100
**Coverage**: 65 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 88 |
| Offload Value | 15 | 48 |
| Doc Legibility | 12 | 51 |
| Payload Efficiency | 8 | 5 |
| Error Recoverability | 15 | 53 |
| Prior Knowledge Coverage | 5 | 0 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/supabase

---

# Tavily

> Search, extract and crawl API for AI agents.

- **Category**: web-search
- **Éditeur** : Tavily
- **Site** : https://tavily.com
- **Pricing model**: usage_based
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Tavily API `tavily-rest`

- **Type** : rest
- **Base URL** : https://api.tavily.com
- **Documentation** : https://docs.tavily.com/documentation/api-reference/endpoint/search
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 55.7 / 100
**Coverage**: 78 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 88 |
| Doc Legibility | 12 | 48 |
| Task Atomicity | 6 | 16 |
| Interaction Cost | 15 | 6 |
| Error Recoverability | 15 | 54 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/tavily

---

# Turso

> SQLite databases at the edge, queried over HTTP.

- **Category**: backend-database
- **Éditeur** : Turso
- **Site** : https://turso.tech
- **Pricing model**: freemium
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Turso HTTP API (Hrana over HTTP) `turso-http-v2`

- **Type** : rest
- **Version** : v2
- **Documentation** : https://docs.turso.tech/sdk/http/reference
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 65.1 / 100
**Coverage**: 78 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 62 |
| Doc Legibility | 12 | 31 |
| Task Atomicity | 6 | 67 |
| Interaction Cost | 15 | 77 |
| Error Recoverability | 15 | 51 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/turso

---

# Weaviate

> Open-source vector database, managed in Weaviate Cloud.

- **Category**: vector-database
- **Éditeur** : Weaviate
- **Site** : https://weaviate.io
- **Pricing model**: freemium
- **Rubrique** : v0.2.0

## Surfaces evaluated

### Weaviate REST API `weaviate-cloud-rest-v1`

- **Type** : rest
- **Version** : v1
- **Documentation** : https://docs.weaviate.io/weaviate/api/rest
- **Authentification** : bearer_static

**Statut de notation** : `scored`

Complete benchmark. `overall` is a weighted average of the applicable criteria, comparable to other surfaces in the same category and the same rubric version.

**Overall score**: 55.0 / 100
**Coverage**: 86 %
**Measured**: 2026-10-05 (oldest measurement still counted)

| Criterion | Weight | Score |
| --- | --- | --- |
| Auth Friction | 10 | 95 |
| Offload Value | 15 | 78 |
| Doc Legibility | 12 | 51 |
| Task Atomicity | 6 | 38 |
| Interaction Cost | 15 | 17 |
| Payload Efficiency | 8 | 52 |
| Error Recoverability | 15 | 40 |
| Prior Knowledge Coverage | 5 | 100 |

## Methodology

Scores are derived from real runs and probes, never from reviews. A missing criterion is excluded from the computation and the weights are renormalised — it is never counted as a zero.

- Full rubric: https://agentsco.re/api/v1/rubric
- Methodology: https://agentsco.re/methodology
- Cette fiche en JSON : https://agentsco.re/api/v1/tools/weaviate