AXRank MCP — Tool reference

Complete parameter reference for every tool the AXRank MCP exposes. The server is stateless, requires no auth for public tools, and speaks Streamable HTTP at https://axrank.ai/mcp.

All tools are declared with full inputSchema descriptions so agents can self-document from the server without this page.

find_services_for

Public

Given a capability or use case, return the highest-ranking services from AXRank along with each service's summary findings. Use this when picking a provider for a new integration. Results are filtered for relevance and then ordered by AXRank score (autonomous-mode aggregate, 0–100).

Parameter	Type	Required	Description
capability	string	Yes	Free-form capability or use case.e.g. "send email", "track exceptions", "issue tracking", "payments", "auth"
limit	number (integer, 1–20)	No	Maximum number of services to return. Default 5.

Example call

find_services_for({
  capability: "send transactional email",
  limit: 3
})

Example response

Found 3 services for "send transactional email":

• Postmark — AXRank 78/100
  Categories: Email (match)
  About: Transactional email delivery with high deliverability.
  ...

• Sendgrid — AXRank 65/100
  ...

get_service

Public

Look up a single service by name or slug and return its full agent-facing record: AXRank score, categories, public summary, and per-dimension narrative. Use this after find_services_for to drill into a specific candidate.

Parameter	Type	Required	Description
name	string	Yes	Service name or slug. Case-insensitive.e.g. "linear", "Sentry", "stripe", "workos"

Example call

get_service({ name: "linear" })

Example response

• Linear — AXRank 72/100
  Categories: Project management (match)
  About: Issue tracking, planning, and agent workflows for product development teams.
  Summary: Linear exposes a mature GraphQL API and an official MCP...
  Findings: Discoverable 80 · Comprehensible 75 · Reliable 68 ...
  Root URL: https://linear.app
  Leaderboard: https://axrank.ai/s/linear
  Evals run: 3; last updated 2026-05-23

compare_services

Public

Compare two or more services for a given capability. Returns each service's AXRank score, categories, and summary findings, ordered by score. Use this when choosing between specific candidates the user has already narrowed down.

Parameter	Type	Required	Description
services	string[] (2–8 items)	Yes	Two or more service names or slugs to compare.e.g. ["linear", "jira"], ["stripe", "paddle", "lemon-squeezy"]
capability	string	No	Optional capability or use case. Used to phrase the comparison output.

Example call

compare_services({
  services: ["linear", "jira"],
  capability: "issue tracking"
})

Example response

Comparing for "issue tracking", ordered by AXRank score:

• Linear — AXRank 72/100
  ...

• Jira — AXRank 54/100
  ...

Winner: Linear (72). Rank is the autonomous-mode aggregate on the most recent eval.

report_outcome

Public

Closed-loop telemetry for AXRank. Report whether the agent succeeded at a task with a service. Per AXRank's data policy this collects only {service, task_category, model, succeeded, failure_mode} — no prompts, no user IDs, no proprietary content. Default-on for individual developers, default-off for enterprise harnesses.

Parameter	Type	Required	Description
service	string	Yes	Service name or slug the agent was working with.
task_category	string	Yes	Short label for what the agent was trying to do.e.g. "create-issue", "send-email", "list-projects"
succeeded	boolean	Yes	Whether the agent completed the task end-to-end.
model	string	No	Model identifier the agent ran on.e.g. "claude-opus-4-7", "gpt-5.5"
failure_mode	string	No	Short failure-mode label if the task failed. Either a known AXRank failure ID (e.g. C2, V1) or a free-form description.

Example call

report_outcome({
  service: "linear",
  task_category: "create-issue",
  succeeded: false,
  model: "claude-opus-4-7",
  failure_mode: "V1"
})

Example response

Recorded outcome for linear: failed.

get_eval_report

Owner / Admin only

Requires a valid api_key. Obtain by claiming your service at axrank.ai/claim.

Returns a structured markdown eval report for a service: score, per-dimension breakdown, top failure modes, and tasks attempted. The report is formatted as markdown so it can be pasted directly into a CLAUDE.md, system prompt, or agent context file to give the agent real findings to act on.

Parameter	Type	Required	Description
service	string	Yes	Service name or slug.e.g. "linear", "Stripe", "workos"
api_key	string	Yes	Owner API key (issued when you claim your service) or the AXRank admin key. The call is rejected if the key does not match.

Example call

get_eval_report({
  service: "linear",
  api_key: "axr_owner_..."
})

Example response

# Linear — AXRank Eval Report (Authorized)

**Score:** 72/100 (autonomous mode)
**Evals run:** 3
**Last evaluated:** 2026-05-23
**Leaderboard:** https://axrank.ai/s/linear
**Access:** owner/admin verified ✓

## By dimension
| Dimension     | Score   | Passed | Probed |
|---------------|---------|--------|--------|
| Discoverable  | 80/100  | 4      | 5      |
| Comprehensible| 75/100  | 3      | 4      |
| Reliable      | 68/100  | 3      | 4      |
| Composable    | 70/100  | 2      | 3      |
| Recoverable   | 60/100  | 3      | 5      |
| Efficient     | 65/100  | 3      | 4      |
| **Aggregate** | **72/100** | 18  | 25     |

## Task results

### ✓ Create an issue via the API
**Capability:** create-issue
**Dimensions:** comprehensible, reliable
**Rubric:** Agent calls POST /issues and receives a 201 with an issue id.

### ✗ Search issues by label
**Capability:** search-issues
**Dimensions:** comprehensible, composable
**Rubric:** Agent returns issues filtered to a specific label.
**Failure mode:** C2 (secondary: D3)
**Classification reasoning:** The filter parameter name is not documented...
**Grader reasoning:** Agent used 'tag' instead of 'labelIds'; API returned 400.

## Failure mode summary
| ID | Count |
|----|-------|
| C2 | 3     |
| D3 | 2     |
| R1 | 1     |

## Narrative
Linear's surface is highly discoverable but agents burn context on...

---
*Authorized findings report for Linear. AXRank — Agent Experience, graded.*

Server metadata

Endpoint

https://axrank.ai/mcp

Transport

Streamable HTTP

Auth

None (public tools) / api_key (get_eval_report)

Session

Stateless