Eval-backed tool discovery for AI agents, on the auxiliar.ai web-access gateway — one API key for 24 search, scraping, browser-automation and voice APIs, upstream keys injected server-side, usage at each provider's real metered price.
Ask it "what's the best provider for this job?" and it answers from measured public benchmarks — every provider runs the identical task corpus per verb; scorecards carry their run dates; weak scores are published, not hidden.
claude mcp add auxiliar -- npx auxiliar-mcpor in any MCP client config:
{ "mcpServers": { "auxiliar": { "command": "npx", "args": ["auxiliar-mcp"] } } }Related MCP server: cross-validated-search
Tool | What it does |
| Best provider(s) for a job ( |
| The full leaderboard for one verb — every scored provider, raw metrics, run dates. |
| One provider in full: route, pricing, choose/avoid guidance, all its dated scorecards. |
| What the gateway is, how to get a key, how to call it. |
Every response carries the run date behind each number (measured_on, latest_run), the ranking context (rank #n of m), honest caveats, providers excluded_by_constraints (never silently dropped), and gated_not_scored entries for providers that couldn't be scored on the shared corpora.
Recommendations return an executable call pattern:
https://api.auxiliar.ai/{provider}/{provider-native-path}
Authorization: Bearer <your auxiliar API key>Same paths, parameters and responses as each provider's own docs — the gateway injects the upstream key server-side. Get a key (with $5 free credit, no card) at auxiliar.ai.
Benchmark data loads at runtime from auxiliar.ai/evals.json (1h in-memory cache) and falls back to a bundled snapshot offline — responses declare which via data_source. The same data renders the human-readable scorecards at auxiliar.ai/tools. Rankings carry no house incentive: the gateway's fee is flat at credit top-up, so nothing is earned by steering you toward pricier providers.
npm install
npm run build # tsc → dist/ (+ bundled data snapshot)
npm test # unit tests + end-to-end stdio round-trip
npm run update-fallback # refresh src/data/evals-fallback.json from productionReleasing: bump the version in package.json and server.json (two spots) — npm run check-versions (run automatically at prepublish) enforces sync — then npm publish and mcp-publisher publish.
v0.23 (this release) — eval-backed discovery.
v1.0 — in-loop execution: call the providers through the gateway from this MCP (web_search, scrape, extract, crawl, …), routed by the same measured rankings.
MIT
What auxiliar.ai is and how an agent uses it: a web-access API gateway — one API key for 24 search, scraping, browser-automation and voice providers, upstream keys injected server-side, usage at provider-metered prices. Call this to learn how to get a key and call the gateway.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden. It accurately describes the tool as an informational resource with no side effects or behavioral traits beyond providing explanations. There is no contradiction.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise, using two sentences to convey the essence of the tool. It front-loads the purpose and ends with a direct instruction, wasting no words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's low complexity (no parameters, no output schema), the description provides sufficient context about what the tool offers. It could perhaps mention the output format, but is otherwise complete for an informational tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool has zero parameters, so parameter semantics are automatically satisfied. The description adds context about the gateway's capabilities, but this is not parameter-specific. Baseline score of 4 applies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states that this tool provides information about what auxiliar.ai is and how to use it as an API gateway. It distinguishes itself from sibling tools (get_provider, get_scorecard, recommend_tools) by being the introductory tool.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly says 'Call this to learn how to get a key and call the gateway,' which provides clear context for when to use it. It does not mention alternatives or exclusions but implies this is for initial orientation.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Everything about one provider on the auxiliar.ai gateway: route and call pattern, entry pricing, when to choose or avoid it, and all its dated benchmark scorecards across verbs.
| Name | Required | Description | Default |
|---|---|---|---|
| slug | Yes | Provider slug on the gateway (its route is /<slug>). |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description must disclose behavioral traits. It does not mention whether the tool is read-only, requires authentication, or has side effects. The description focuses on the information returned but omits operational behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single sentence that conveys multiple aspects efficiently. It is front-loaded with 'Everything about one provider,' but the density may slightly reduce readability. Still, it is concise and structured.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the lack of output schema, the description lists the key information returned (route, pricing, scorecards), which provides adequate context. However, it does not specify the exact format or structure of the response. For a simple tool with one parameter, this is mostly complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
With 100% schema coverage, the schema already describes the single parameter (slug) fully. The description adds context about the route pattern but does not significantly enhance understanding beyond the schema. Baseline score of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states that the tool provides comprehensive information about a single provider, including route, pricing, and scorecards. It distinguishes itself from sibling tools like 'get_scorecard' and 'recommend_tools' by focusing on a specific provider's full details.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description mentions 'when to choose or avoid it,' implying usage context, but does not explicitly state when to use this tool versus alternatives like 'get_scorecard' or 'about_auxiliar.' No explicit when-not or sibling differentiation.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Full benchmark leaderboard for one job/verb: every scored provider on the auxiliar.ai gateway, best first, with raw measured metrics and run dates. Use to compare all options rather than be handed a pick.
| Name | Required | Description | Default |
|---|---|---|---|
| job | Yes | The verb to rank. search = Web search: query → ranked results | scrape = One URL → clean page content/markdown, anti-bot handled | crawl = Site → enumerate and fetch many pages | extract_ai = Page → structured fields from an AI/natural-language schema | extract_rules = Page → structured fields via CSS/XPath rules | answer = Question → synthesized answer with cited sources | screenshot = URL → rendered page image | scrape_domain = Domain-specific scrapers for hard targets (e-commerce, social) | act = Declarative page interactions (click, fill, scroll) in one call | act_agent = Autonomous natural-language browser agent for multi-step goals | serp = Google SERP verticals (web, news, images, places, scholar) as structured JSON | parse = PDF/document → text (OCR where supported) | watch = URL → change detection / monitoring |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description must disclose behavioral traits. It mentions returning a leaderboard with metrics and dates, but does not disclose whether the tool is read-only, has rate limits, requires authentication, or any other side effects. The description lacks transparency about the tool's behavior beyond its output.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences: the first defines the output, the second gives usage guidance. Every word adds value with no repetition or fluff. It is front-loaded with the key information.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool has one parameter with 100% schema coverage, no output schema, and no annotations, the description adequately explains the return data (leaderboard, metrics, dates). It could mention if there are any limits (e.g., number of providers) but is otherwise complete for a simple lookup tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema coverage is 100% with detailed enum descriptions for the 'job' parameter. The description adds context about the output (leaderboard, metrics, dates) but does not add meaning to the parameter itself beyond the schema. Baseline 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description specifies exactly what the tool does: provides a full benchmark leaderboard for one job/verb, listing providers with metrics and dates. It clearly distinguishes from sibling tools like get_provider and recommend_tools by focusing on comparison rather than selection.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly states the use case: 'Use to compare all options rather than be handed a pick.' This tells the agent when to invoke this tool (when comparing providers) and implies not to use it when a single recommendation is needed.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recommend the best web-access provider(s) on the auxiliar.ai gateway for a job, ranked by measured benchmarks (quality, latency p50, cost per success, error rate — dated runs on a shared public corpus). Returns executable routes, the numbers behind each pick, and honest caveats. Use when deciding which search/scraping/crawling/extraction/browser API to use.
| Name | Required | Description | Default |
|---|---|---|---|
| job | Yes | The job to be done. search = Web search: query → ranked results | scrape = One URL → clean page content/markdown, anti-bot handled | crawl = Site → enumerate and fetch many pages | extract_ai = Page → structured fields from an AI/natural-language schema | extract_rules = Page → structured fields via CSS/XPath rules | answer = Question → synthesized answer with cited sources | screenshot = URL → rendered page image | scrape_domain = Domain-specific scrapers for hard targets (e-commerce, social) | act = Declarative page interactions (click, fill, scroll) in one call | act_agent = Autonomous natural-language browser agent for multi-step goals | serp = Google SERP verticals (web, news, images, places, scholar) as structured JSON | parse = PDF/document → text (OCR where supported) | watch = URL → change detection / monitoring | |
| limit | No | Max candidates to return. Default 3. | |
| max_cost_usd | No | Exclude providers whose measured cost per call/success exceeds this (USD). | |
| optimize_for | No | Ranking priority. balanced/quality = published composite; cost/latency = that measured metric. Default balanced. | |
| max_latency_ms | No | Exclude providers whose measured latency p50 exceeds this. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description discloses the tool's behavior well: it ranks providers based on measured benchmarks, returns executable routes, numbers, and caveats. There is no mention of side effects, but as a recommendation tool, destructive actions are unlikely. The description is transparent about inputs and outputs.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences, each serving a distinct purpose: the first states what the tool does and how it works, the second states what it returns and when to use it. No superfluous information, front-loaded with the core purpose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with 5 parameters and no output schema, the description covers the essential purpose, usage, and return types. It could be more specific about the output structure, but 'executable routes, numbers, and honest caveats' gives adequate guidance. The absence of output schema does not leave critical gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100% and each parameter has a clear description in the schema. The tool description adds little beyond the schema for parameter meanings, but it provides overall context. Baseline 3 is appropriate as the description does not significantly augment parameter understanding.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb ('Recommend') and resource ('web-access providers'), explicitly states the ranking criteria (measured benchmarks), and distinguishes from siblings by focusing on provider comparison for a given job.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description includes 'Use when deciding which search/scraping/crawling/extraction/browser API to use', providing clear context. However, it does not mention when not to use it or explicitly contrast with sibling tools, though the sibling names imply different purposes.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.23.1about_auxiliarget_providerget_scorecardrecommend_toolsScored across 4 tools
Each tool has a clearly distinct purpose: about_auxiliar introduces the service, get_provider details a single provider, get_scorecard shows benchmarks for a verb, and recommend_tools recommends providers. No overlap.
All tool names use a consistent verb_noun pattern in snake_case (about_auxiliar, get_provider, get_scorecard, recommend_tools), making them predictable and easy to understand.
With 4 tools covering onboarding, provider details, benchmarks, and recommendations, the count is well-scoped for an information-focused MCP server about a gateway service.
The tools cover the main user needs: understanding, provider details, benchmarks, and recommendations. A minor gap is the lack of a tool to list all providers or verbs, but recommend_tools partially addresses this by taking a job as input.
API discovery and selection for AI agents connecting to the physical world.
Reliable web access for AI agents: smart HTTP, rotating proxies, and full-browser rendering.
Browser-backed QA with evidence and fix-ready reports for coding agents.
Reproducible benchmarks and reliability evidence for agent tools.
