Run evidence-backed website conversion audits with an AI browsing agent.
Scored across 2 tools
The two tools form a clean start/poll pair: test_conversion initiates an audit and returns an audit_id, while get_conversion_report retrieves its status and result. Descriptions explicitly cross-reference each other and state which one to call when, leaving no realistic chance of misselection.
Both names follow a consistent snake_case verb_noun pattern (test_conversion, get_conversion_report) and share the domain term 'conversion'. The verb choice maps intuitively to the action: test to start, get to retrieve.
Two tools is thin even for a narrow async-audit domain; the surface is minimal but functional. A cancel/abort or list-audits tool would round out the set without bloat, so it sits at the borderline rather than feeling well-scoped.
The core lifecycle is covered: start an audit and poll it to completion, including stale, failed, and withheld-score states, plus rate-limit and error guidance. Missing niceties like cancelling a running audit or listing past audits are minor gaps agents can work around.
Get the status and, once finished, the result of an audit started with test_conversion.
Pass the audit_id returned by test_conversion (a report URL also works). status is "pending" while the agent is still running: wait retry_after_seconds and call again. If stale is true, the audit will never finish; start a new one.
When status is "completed", result contains: task_completed (did the agent reach the goal), score (0-100 Agent Conversion Score, or null when score_withheld because the run did not finish measuring), summary, successes, frictions and recommendations; evidence.steps lists what the agent did. failure_classification explains a non-clean run: "blocked" (safety boundary or site-side access refusal such as a bot challenge — a real finding about the site), "run_did_not_finish", "infrastructure_failure" or "pipeline_error" (Agent Conversion's own problem, not the site's). report_url is a human-readable report to share.
status "failed" means no report will be produced. Calling this tool is free and safe to repeat.
| Name | Required | Description | Default |
|---|---|---|---|
| audit_id | Yes | The audit_id returned by test_conversion, or the report URL. |
| Name | Required | Description |
|---|---|---|
| ok | Yes | |
| url | Yes | |
| task | Yes | |
| stale | Yes | |
| domain | Yes | |
| locale | Yes | |
| result | Yes | |
| status | Yes | |
| message | Yes | |
| audit_id | Yes | |
| evidence | Yes | |
| created_at | Yes | |
| report_url | Yes | |
| updated_at | Yes | |
| is_benchmark | Yes | |
| execution_status | Yes | |
| retry_after_seconds | Yes | |
| failure_classification | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already cover the safety profile (readOnly, idempotent, non-destructive), and the description adds the polling cadence, the retry_after_seconds/stale semantics, and the meaning of failure_classification values. It stops short of describing pagination or rate limits beyond 'free and safe to repeat', but the state machine disclosure is genuinely useful.
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?
Dense but well-organized: status handling first, then completed-result fields, then terminal states. Each sentence carries information, though the enumeration of result fields is long and could be trimmed since an output schema exists.
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?
Although an output schema exists, the description still supplies the semantic meaning of status values, failure_classification categories, and field intent, which schema types alone cannot convey. An agent has everything needed to poll correctly and interpret the outcome.
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 the schema already states the parameter accepts the audit_id or the report URL, so the description's parenthetical merely restates it. No additional format or syntax detail is added, so the baseline 3 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?
Opens with a specific verb+resource: retrieving the status and eventual result of an audit started by test_conversion, and explicitly frames it as the polling counterpart to that sibling. An agent can distinguish it from test_conversion without reading either schema.
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?
Gives concrete when-to-use and what-to-do-next guidance: call after test_conversion, wait retry_after_seconds on 'pending', restart if stale is true, and that 'failed' means no report. Exclusions and state transitions are spelled out rather than implied.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Start a real Agent Conversion audit: an autonomous browsing agent opens the given website and tries to complete the task, then the run is scored 0-100 with problems and recommendations.
Use this when someone wants to know whether an AI agent can find something or complete a commercial step on a specific website, e.g. "Can an agent find the Pro plan's monthly price on example.com?" or "Can an agent start the self-service signup on example.com?".
Returns immediately with an audit_id and status "pending" — it does NOT wait for the result. Call get_conversion_report with the audit_id to get the result.
Rules: url must be a public http(s) URL (no localhost, private IPs, credentials or custom ports). task must be 10-1000 characters and describe the goal in plain language; phrase purchase-like goals as "reach checkout without buying". Each call costs a real browser run and is rate limited: do not call it repeatedly for the same url+task — poll get_conversion_report instead. Error codes rate_limited, daily_limit and at_capacity mean try again later.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | Public website URL to test, including https://, e.g. https://www.example.com or https://www.example.com/pricing. | |
| task | Yes | The commercial task in plain language, e.g. "Find the monthly price of the Pro plan" or "Start the free trial signup and stop before entering payment details". | |
| locale | No | Language for the generated summary and recommendations: "en" (default) or "es". |
| Name | Required | Description |
|---|---|---|
| ok | Yes | |
| url | Yes | |
| task | Yes | |
| domain | Yes | |
| locale | Yes | |
| status | Yes | |
| audit_id | Yes | Pass this to get_conversion_report. |
| next_step | Yes | |
| report_url | Yes | Human-readable report; available once the audit completes (404 before). |
| submitted_at | Yes | |
| retry_after_seconds | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Beyond the annotations (readOnlyHint=false, openWorldHint=true, idempotentHint=false), it discloses the async contract ('returns immediately with audit_id and status "pending" — it does NOT wait'), that each call costs a real browser run, that it is rate limited, and the mapping of error codes to retry-later 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?
Front-loads the purpose, then usage, then the async return contract, then constraints and error semantics, in four short blocks with no filler. Dense but every sentence carries operative 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?
Output schema exists so return values needn't be described, yet the description still covers the async lifecycle, the handoff to get_conversion_report, input validation rules, and rate-limit recovery — everything an agent needs to call it correctly.
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 coverage is 100%, so the baseline is 3, but the description adds real constraints not present in the schema: public http(s) only, task length 10-1000 characters, and the phrasing rule to write purchase-like goals as 'reach checkout without buying'. It also gives worked examples of valid task strings.
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?
Starts with a specific verb+resource ('Start a real Agent Conversion audit') and explains exactly what happens mechanically: an autonomous agent browses the site, attempts the task, and is scored 0-100. It also names the sibling get_conversion_report, so the agent can distinguish the launcher from the result-fetcher.
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?
Gives explicit when-to-use framing with two concrete example questions, plus hard exclusions: no localhost/private IPs/credentials/custom ports, do not call repeatedly for the same url+task, and what the rate_limited/daily_limit/at_capacity codes mean. It routes the agent to get_conversion_report for polling.
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.
get_conversion_reporttest_conversionWebsite checker for AI coding agents. Measures your site in a real browser, with evidence.
Run full technical SEO audits of any website from your AI agent and get structured findings back.
Website intelligence for AI agents: landing page audits, business contact details, tracking checks.
AI website growth audits: SEO, performance, AI readiness (GEO), conversion, a11y, security.
Add one secure layer between your agents and this server.
