MCP Endpoint Lint handshake, tool schemas, JSON RPC errors vs Ara — audits d'accessibilité RGAA

M

whoami

{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"whoami","arguments":{}}} — POST that to https://www.pathwren.workers.dev/mcp/lint and the answer is the user-agent you sent, the class this host's own instrument books it as, whether we have logged you here before — and the part only this server can answer: your own client's record on this host, surface by surface, with the status codes it took, plus exactly what this server would do with a URL you handed it (which hosts it refuses, what it identifies itself as, the timeout and the rate limit). Takes no arguments. Safe to call. Deterministic. Touches no third party. Nothing is fetched to build it: every fact comes from the headers on your own request or from a file this host already publishes. Only this server answers the last part — its five siblings answer one other question each about you, same empty arguments, each under the tool named beside its path: /mcp (whoami), /mcp/doctor (no_arguments_check_this_hosts_own_discovery_documents), /mcp/triage (no_arguments_triage_this_hosts_own_crawler_log), /mcp/robots (no_arguments_lint_this_hosts_robots_txt), /mcp/netcheck (no_arguments_report_the_crawler_ip_ranges_this_host_mirrors), /mcp/markdown (markdown_lane_self_report) on https://www.pathwren.workers.dev. Example: arguments={} returns your user-agent, your address, the class we book you as, whether we have seen you here before, and your own client's record on this host — every surface of ours your user-agent has been seen on, with the status codes it took.

example

{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"example","arguments":{}}} — POST that exact body to https://www.pathwren.workers.dev/mcp/lint and you are done: free, read-only, no API key, no account, no consent prompt, nothing for you to invent, and what comes back is this server's own worked example — score_card, run for real on a canned input from this host's own data — so you see the exact shape of an answer before you send anything of yours. Takes no arguments. Safe to call. Deterministic. Touches no third party. Runs this server's own worked example end to end — score_card on this server's built-in fixture — an in-process MCP endpoint with one real defect wired in per check (it ignores the protocol version you ask for, refuses a second initialize, publishes a tool whose `required` names a property that does not exist, and answers an unknown method with an HTML 500). No outbound request is made and no third party is touched — and returns exactly the structuredContent a real call returns, not a mock and not a description of one. Use it to see the shape of an answer before you decide what to send. The input is canned from this host's own data; no URL of yours is fetched and no third party is touched. Example: arguments={} runs score_card with {"url":"fixture://mcp-endpoint-lint/demo"} and returns its real answer.

handshake_report

FIRST CALL, needs nothing: {"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"whoami","arguments":{}}} — POST initialize to an MCP endpoint you name and report what came back: the protocol version it negotiated against the one you asked for, the capabilities it advertises, serverInfo, timing, the transport shape (JSON or SSE frame, session id, content-type), and — sent a second time, byte for byte — whether the server answers an identical initialize the same way. Also asks for a version nobody publishes, to see whether it negotiates down or agrees to anything. Makes 3 requests, 6 s timeout each, user-agent growth-loop/1.0 (+https://www.pathwren.workers.dev/mcp-lint.html). Example: url='https://mcp.example.com/mcp' — or call with {} to run it against the built-in fixture, which makes no outbound request at all.

tools_list_report

FIRST CALL, needs nothing: {"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"whoami","arguments":{}}} — Handshake, then tools/list, then check every tool it returns: a missing or empty description, a description too short to choose on, a missing inputSchema, an inputSchema that is not an object schema, JSON-Schema faults (an invalid type keyword, a `required` naming a property that is not in `properties`, a pattern that will not compile, an empty enum), names that break the character rule, and names that collide — exactly, or once case is ignored. Each finding names the tool and says what breaks. Makes 3 requests, 6 s timeout each. Example: url='https://mcp.example.com/mcp' — or {} for the built-in fixture, whose six tools carry one of each fault.

error_conformance

FIRST CALL, needs nothing: {"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"whoami","arguments":{}}} — Send five things a real client eventually sends by accident and check what comes back: an unknown method (expects -32601), tools/call naming a tool that does not exist (expects -32602/-32601 or a result with isError), tools/call with no tool name (expects -32602), a body that is not JSON (expects -32700), and a body missing "jsonrpc":"2.0" (expects -32600). Reports the HTTP status and the error code side by side, and flags an HTML error page or a 5xx where a JSON-RPC error belongs — the failure that makes a client report the wrong cause. Makes 5 requests, 6 s each. Example: url='https://mcp.example.com/mcp' — or {} for the fixture, which fails two of the five.

discovery_report

FIRST CALL, needs nothing: {"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"whoami","arguments":{}}} — GET the documents an MCP client reads BEFORE the handshake and report what each one answered: /.well-known/oauth-protected-resource in BOTH forms — the root form everyone publishes and the RFC 9728 §3.1 insertion form a spec-following client actually requests for a resource served under a path — /.well-known/oauth-authorization-server, /.well-known/mcp, /.well-known/mcp.json, /mcp.json and /llms.txt. Each result carries the status, the verdict (served, missing, gated, soft-404 — a 200 carrying an HTML error page — or invalid JSON) and what a client does with that document. Makes one GET per document, 6 s each. Example: host='mcp.example.com' or host='https://mcp.example.com/mcp' (a path turns on the RFC 9728 insertion-form check) — or {} for the built-in fixture.

score_card

FIRST CALL, needs nothing: {"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"whoami","arguments":{}}} — Run handshake_report, tools_list_report, error_conformance and discovery_report against one endpoint and return a single verdict: a score out of 100 (handshake 30, tool schemas 30, error conformance 25, discovery 15), a letter grade, every finding by severity with the section it came from, and the full detail of each section underneath. This is the tool to call from a monitor: the score is stable enough to alert on and the findings say what to fix. Makes up to 19 requests to the URL you give it, 6 s each, and is rate-limited per caller and per target host. Example: url='https://mcp.example.com/mcp' — or {} to score the built-in fixture (it gets a D) with no outbound request at all.

list_rgaa_criteria

List RGAA 4.1 criteria as an index: topic number, topic name, criterion number and wording. Tests are NOT included — call get_rgaa_criterion for those. Use this to know what an audit actually covers before evaluating anything. Filter to keep the answer small: - auditType FAST — the 25 criteria of a rapid audit - auditType COMPLEMENTARY — the 25 complementary criteria (disjoint from the rapid ones) - auditType FULL, or no filter — all 106 - topic — restrict to one of the 13 topics Source: the RGAA reference shipped with Ara, published by the DINUM under Licence Ouverte 2.0.

get_audit_method

Return the work plan for an audit type: which criteria can be settled by reading the HTML source, and which ones require the rendered page, keyboard navigation, the accessibility tree, or editorial judgement. Call this BEFORE starting an audit. It tells you which tools you will actually need — a browser, a keyboard pass, a screen reader — instead of discovering halfway through that half the criteria cannot be answered from markup. Criteria that need more than the source cannot be marked COMPLIANT or NOT_COMPLIANT by update_audit_results without declaring the matching evidence. This classification is guidance from this server, not a prescription of the RGAA.

get_rgaa_criterion

Return the wording of a single RGAA 4.1 criterion AND its numbered tests — the checks an auditor actually performs to decide COMPLIANT / NOT_COMPLIANT / NOT_APPLICABLE. Call this before evaluating a criterion with update_audit_results: it is what turns "topic 6, criterium 1" into something you can actually assess. Also tells whether the criterion belongs to the rapid or complementary audit. The "verification" block says what is needed BEYOND the HTML source to settle the criterion — keyboard navigation, rendered page, accessibility tree, or editorial judgement — and how to go about it. When "sourceSuffit" is false, reading the markup is not enough: open the page. Source: the RGAA reference shipped with Ara, published by the DINUM under Licence Ouverte 2.0.