[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8},"actionData","errors","root",{"_106":107,"_108":109},"routes/embed.x.$handle.$name.$",{"_9":10,"_11":12,"_13":14,"_15":16,"_17":18,"_19":20,"_21":-5},"handle","mcpconformance","name","mcp-client-conformance","path","src/hosted/html.ts","project",{"_22":49,"_11":12,"_35":50,"_51":-5,"_52":44,"_53":54,"_55":54,"_56":-5,"_57":-5,"_58":59,"_60":61,"_62":34,"_63":-5,"_64":-5,"_65":66,"_67":68,"_69":-5,"_70":-5,"_71":-5,"_72":-5,"_73":74,"_75":74,"_76":77,"_78":77,"_79":77,"_80":74,"_81":74,"_82":83,"_84":85,"_86":87,"_88":44},"file",{"_22":23,"_11":24,"_25":26,"_27":28,"_29":30,"_31":32,"_33":34,"_35":36,"_37":38,"_39":40,"_41":42,"_43":44,"_45":-5,"_13":46},"permalink","https://www.val.town/x/mcpconformance/mcp-client-conformance/code/src/hosted/html.ts","extension","id","32615475-c028-48bc-bc96-dc536e999530","html.ts","fileId","d43ddbd6-ad13-11f1-9006-1607ee4eb77e","parentId","11abc952-e3be-41e6-ac33-6929c3af1284","branchId","fdc11199-683b-41f1-a3c5-d5ac53972f86","branchVersion",389,"latest",true,"createdAt",["D",1789379126402],"version",14,"fileType","script","content","/**\n * HTML for the hosted server: the matrix (landing and run/column/cell config\n * pages) and the per-cell check report. Everything interpolated goes through\n * escapeHtml(); JSON embedded for the copy buttons goes through jsonForScript().\n */\n\nimport { ConformanceCheck, CheckStatus } from '../types.ts';\nimport {\n MCP_PATH,\n startableCount,\n type HostedMatrix,\n type MatrixCell\n} from './matrix.ts';\nimport { buildCommit, buildText, type BuildInfo } from './build.ts';\nimport type { CellConfig, CellStatus, RunConfig } from './server.ts';\nimport type { CellRef } from './session.ts';\nimport {\n groupColumn,\n incompleteNote,\n NO_SIGN_IN,\n summarize,\n unavailableScenarios,\n type CellReport,\n type CellState,\n type ColumnReport,\n type RunReport\n} from './report.ts';\nimport type { ClientIdentity } from './identity.ts';\nimport type { Cause } from './findings.ts';\nimport type { SnapshotInfo, StoreRetention } from './store.ts';\nimport type { ShownCheck } from './shown.ts';\nimport {\n BY_LABEL,\n REACHED,\n STATE_GLYPH,\n STATE_LABEL,\n causeNumbers,\n countsLine,\n countsText,\n NO_FINDINGS,\n NOT_TRIED_WHY,\n notTriedHeading,\n passedHeading,\n scoreText,\n showsCounts,\n UNAVAILABLE_HEADING,\n UNAVAILABLE_WHY,\n stopNote,\n utcMinute,\n waitingFor,\n type Unmet\n} from './markdown.ts';\nimport { describeStep, type Step } from '../steps/index.ts';\nimport type { CompositeView } from './composite.ts';\nimport {\n CLIENTS,\n clientConfig,\n compositeName,\n serverName,\n type ServerEntry\n} from './client-config.ts';\nimport { escapeHtml } from './escape.ts';\n\nexport const STATE_STYLE: Record = {\n pass: 'background:#d1fae5;color:#065f46',\n fail: 'background:#fee2e2;color:#991b1b',\n waiting: 'background:#e0e7ff;color:#3730a3',\n stopped: 'background:#fef3c7;color:#92400e',\n 'in-progress': 'background:#dbeafe;color:#1e40af',\n incomplete: 'background:#fef3c7;color:#92400e',\n 'not-tried': 'background:#f3f4f6;color:#6b7280',\n 'not-startable': 'background:#f3f4f6;color:#9ca3af',\n 'n/a': 'background:#f3f4f6;color:#9ca3af'\n};\n\nexport const STATUS_STYLE: Record = {\n SUCCESS: 'background:#d1fae5;color:#065f46',\n FAILURE: 'background:#fee2e2;color:#991b1b',\n WARNING: 'background:#fef3c7;color:#92400e',\n SKIPPED: 'background:#e5e7eb;color:#374151',\n INFO: 'background:#dbeafe;color:#1e40af'\n};\n\nexport const SCORING_STYLE: Record = {\n scored: 'background:#dbeafe;color:#1e40af',\n not_scored: 'background:#ede9fe;color:#5b21b6',\n unlisted: 'background:#f3f4f6;color:#374151',\n 'n/a': 'background:#f3f4f6;color:#9ca3af'\n};\n\nexport const SCORING_LABEL: Record = {\n scored: 'scored',\n not_scored: 'not scored',\n unlisted: 'not in the requirement set',\n 'n/a': 'n/a'\n};\n\nconst css = `\n body{font:14px/1.5 ui-sans-serif,system-ui,sans-serif;max-width:1100px;\n margin:2rem auto;padding:0 1rem;color:#111}\n code,pre{font:12px/1.4 ui-monospace,SFMono-Regular,Menlo,monospace}\n pre{background:#f6f8fa;padding:.75rem;border-radius:6px;overflow:auto;margin:.4rem 0}\n .pill{display:inline-block;padding:2px 8px;border-radius:10px;\n font-size:11px;font-weight:600;white-space:nowrap}\n .check{border:1px solid #e5e7eb;border-radius:6px;padding:.75rem;\n margin:.5rem 0}\n .check h3{margin:0 0 .25rem;font-size:14px}\n details>summary{cursor:pointer;color:#6b7280;font-size:12px}\n ol.steps{margin:.4rem 0;padding-left:1.5rem}\n ol.steps li{margin:.15rem 0}\n table{border-collapse:collapse;width:100%}\n td,th{text-align:left;padding:.4rem .6rem;border-bottom:1px solid #eee;\n vertical-align:top}\n td.cell{min-width:14rem}\n td.na{color:#9ca3af}\n .muted{color:#6b7280;font-size:12px}\n .crumbs{color:#6b7280;margin:0 0 1rem}\n .crumbs a{margin-right:.25rem}\n .actions{margin:.25rem 0}\n button.copy{font:inherit;font-size:11px;padding:1px 8px;border:1px solid #d1d5db;\n border-radius:10px;background:#fff;cursor:pointer}\n button.copy:hover{background:#f3f4f6}\n a{color:#2563eb}\n h1 code,h2 code{font-size:inherit}\n .note{background:#fffbeb;border:1px solid #fde68a;border-radius:6px;\n padding:.5rem .75rem}\n form.inline{display:inline}\n ol.causes li{margin:.5rem 0}\n tr.group td{background:#f9fafb;font-weight:600;color:#374151}\n td.num{white-space:nowrap;font-variant-numeric:tabular-nums}\n td.what div{margin:.1rem 0}\n .client{margin:.6rem 0}\n ol.start li{margin:.3rem 0}\n dl.states{display:grid;grid-template-columns:max-content 1fr;gap:.2rem .75rem;\n margin:.4rem 0}\n dl.states dd{margin:0}\n tr.nottried td{background:#fffbeb}\n tr.nottried td.head{font-weight:600;color:#92400e}\n tr.nottried details>summary{font-size:14px;font-weight:600;color:#92400e}\n ul.nottried{margin:.4rem 0;padding-left:1.25rem}\n ul.nottried li{margin:.3rem 0}\n tr.sub td{font-weight:600;color:#374151}\n details.section{margin:1.5rem 0}\n details.section>summary{font-size:1.2em;font-weight:600;color:#111}\n`;\n\nexport { escapeHtml };\nconst esc = escapeHtml;\n\n/** JSON safe inside a `. */\nexport function jsonForScript(value: unknown): string {\n return JSON.stringify(value).replace(/$1')\n .replace(/`([^`\\n]+)`/g, '$1');\n return breaks ? html.replace(/\\n/g, '
') : html;\n}\n\n/** What a person driving a client by hand should know before starting. */\nconst HAND_NOTES: Record = {\n 'request-metadata':\n 'Your client’s first request is refused once on purpose, with an ' +\n '“Unsupported protocol version” error naming 2026-07-28, to check that ' +\n 'the client retries. A correct client retries at once; if yours shows an ' +\n 'error instead, retry or reconnect.',\n 'http-standard-headers':\n 'Each kind of request your client sends is checked for the headers it ' +\n 'must carry; a kind it never sends is skipped, not failed. To cover more, ' +\n 'list the tools and call one, list and read the resources, and list the ' +\n 'prompts and get one.'\n};\n\nexport function handNote(scenario: string): string {\n const note = HAND_NOTES[scenario];\n return note ? `

${esc(note)}

` : '';\n}\n\n/**\n * Plain steps for an auth cell, which has no generic-client steps: what a\n * person does with a client they drive by hand.\n */\nexport function authSteps(scenario: string): string {\n if (!scenario.startsWith('auth/')) return '';\n const lines = [\n 'add the MCP URL to your client and connect',\n NO_SIGN_IN.test(scenario)\n ? 'there is no sign-in page: your client gets its token itself, with the credentials this scenario gives it'\n : 'approve the sign-in when your client opens it — the test authorization server approves at once, with no account',\n 'once connected, list the tools'\n ];\n return (\n `

Steps

What to do with a client you drive by hand.

` +\n `

${esc(l)}

`\n );\n}\n\nconst CREDENTIAL_KEYS = ['client_id', 'client_secret'] as const;\n\n/** Credentials the scenario gives the client, as fields a person can copy. */\nexport function credentials(cell: CellConfig): string {\n const raw = cell.env.MCP_CONFORMANCE_CONTEXT;\n if (!raw) return '';\n let context: Record;\n try {\n context = JSON.parse(raw) as Record;\n } catch {\n return '';\n }\n const rows = CREDENTIAL_KEYS.flatMap((key) => {\n const value = context[key];\n if (typeof value !== 'string' || !value) return [];\n return [\n `${key}${esc(value)} ` +\n ``\n ];\n });\n if (!rows.length) return '';\n return (\n `

Credentials

Enter these in your client. ` +\n `A scripted client reads them from MCP_CONFORMANCE_CONTEXT.

` +\n `${rows.join('')}

`\n );\n}\n\n/**\n * The page, with a footer naming the server build when `build` is given:\n * the commit linked to its source, and when it was deployed. A report\n * passes its own (a frozen copy's is the build that froze it); `null` says\n * the report was stored before builds were recorded.\n */\nexport function page(\n title: string,\n body: string,\n build?: BuildInfo | null\n): string {\n return `\n${esc(title)}\n${body}${build === undefined ? '' : buildFooter(build)}`;\n}\n\nfunction buildFooter(build: BuildInfo | null): string {\n const text = buildText(build ?? undefined);\n const commit = build ? buildCommit(build) : undefined;\n const shown = commit\n ? esc(text).replace(\n esc(build!.build),\n `${esc(build!.build)}`\n )\n : esc(text);\n return `\\n`;\n}\n\nfunction scoringPill(cell: MatrixCell): string {\n return `${SCORING_LABEL[cell.scoring]}`;\n}\n\n/** One plain line per step: what a person makes a hand-driven client do. */\nfunction stepLines(steps: readonly Step[]): string {\n const items = steps.map((s) => `

${esc(describeStep(s))}

`);\n return `

${items.join('')}`;\n}\n\n/** The same steps as the JSON a generic client reads. */\nfunction stepsJson(steps: readonly Step[]): string {\n return `

${esc(JSON.stringify(steps, null, 1))}

`;\n}\n\n/** Collapsed, for a matrix cell. */\nfunction stepsDetails(cell: Pick): string {\n if (!cell.steps) return '';\n return (\n `

steps (${cell.steps.length})

` +\n `${stepLines(cell.steps)}${stepsJson(cell.steps)}

`\n );\n}\n\n/** Open, for a page about one scenario: the lines, the JSON folded below. */\nexport function stepsOpen(steps: readonly Step[]): string {\n return (\n stepLines(steps) +\n `

as JSON (MCP_CONFORMANCE_CONTEXT.steps)

` +\n `${stepsJson(steps)}

`\n );\n}\n\ninterface TableOptions {\n origin: string;\n /** When set, startable cells link to their page and results. */\n runId?: string;\n /** Column filter. */\n revision?: string;\n /** Row filter. */\n scenario?: string;\n}\n\n/**\n * The matrix as a table: scenarios down, revisions across. Without a run id\n * it is the static overview (what is scored, what can start, what steps a\n * cell wants); with one, every startable cell links into that run.\n */\nexport function renderMatrixTable(\n matrix: HostedMatrix,\n opts: TableOptions\n): string {\n const revisions = matrix.revisions.filter(\n (r) => opts.revision === undefined || r === opts.revision\n );\n const rows = matrix.rows.filter(\n (r) => opts.scenario === undefined || r.scenario === opts.scenario\n );\n const head =\n `scenario` +\n revisions\n .map((r) =>\n opts.runId\n ? `${esc(r)}`\n : `${esc(r)}`\n )\n .join('') +\n '';\n const body = rows\n .map((row) => {\n const cells = row.cells\n .filter((c) => revisions.includes(c.revision))\n .map((c) => renderMatrixCell(c, opts))\n .join('');\n return (\n `${esc(row.scenario)}` +\n `

${prose(row.description)}

${cells}`\n );\n })\n .join('');\n return `${head}${body}

`;\n}\n\nfunction renderMatrixCell(cell: MatrixCell, opts: TableOptions): string {\n if (cell.scoring === 'n/a') {\n return `n/a — ${esc(cell.reason ?? '')}`;\n }\n const key = `${cell.revision}/${cell.scenario}`;\n let lines = `

${scoringPill(cell)}

`;\n if (cell.scoring !== 'scored' && cell.reason) {\n lines += `

${esc(cell.reason)}

`;\n }\n if (!cell.startable) {\n lines += `

not startable: ${esc(cell.startReason ?? '')}

`;\n } else if (opts.runId) {\n const base = `/s/${esc(opts.runId)}/${esc(key)}`;\n const url = `${opts.origin}/s/${opts.runId}/${key}${MCP_PATH}`;\n lines +=\n `

open · ` +\n `results · ` +\n ` ` +\n `

`;\n }\n lines += stepsDetails(cell);\n return `${lines}`;\n}\n\nexport const REPO_URL = 'https://github.com/modelcontextprotocol/conformance';\n\n/** A revision's specification on modelcontextprotocol.io. */\nexport function specUrl(revision: string): string {\n return `https://modelcontextprotocol.io/specification/${revision}`;\n}\n\n/** What the landing page says about how long this deployment keeps a run. */\nexport interface LandingRetention {\n /** SessionManager.ttlMs: a cell with no request for this long leaves memory. */\n idleMs: number;\n /**\n * Present when runs persist to a store (a serverless deployment), with\n * the store's own lifetimes when it states them (RunStore.retention).\n */\n store?: Partial;\n}\n\n/** \"5 minutes\", \"6 hours\", \"30 days\": the largest unit that divides `ms`. */\nexport function duration(ms: number): string {\n const units: [number, string][] = [\n [86_400_000, 'day'],\n [3_600_000, 'hour'],\n [60_000, 'minute'],\n [1_000, 'second']\n ];\n for (const [size, unit] of units) {\n if (ms >= size && ms % size === 0) {\n const n = ms / size;\n return `${n} ${unit}${n === 1 ? '' : 's'}`;\n }\n }\n return `${ms} ms`;\n}\n\n/** Each state a report shows, in one line (see CellState). */\nconst STATE_MEANING: [CellState, string][] = [\n [\n 'pass',\n 'the client did what the scenario tests, and nothing it sent broke a check'\n ],\n [\n 'fail',\n 'something the client sent broke a check; the cell leads with the reason'\n ],\n [\n 'waiting',\n 'the only failures are steps the scenario still expects, such as a sign-in or a form not finished yet'\n ],\n [\n 'stopped',\n 'waiting, but the client has sent nothing for a minute and no sign-in or form is open: it most likely stopped, so re-run it'\n ],\n [\n 'in-progress',\n 'the client reached the cell but has not yet done anything the scenario tests'\n ],\n [\n 'incomplete',\n 'the client reached the cell and stopped: it spoke only another revision, was turned away, and did not retry'\n ],\n ['not-tried', 'nothing has reached the cell yet']\n];\n\nfunction retentionHtml(r: LandingRetention): string {\n const idle = duration(r.idleMs);\n let kept: string;\n if (!r.store) {\n kept =\n `This server keeps runs in memory only: a cell’s recorded traffic is dropped ` +\n `once the cell has had no request for ${idle}, and frozen copies last until ` +\n `the server restarts.`;\n } else {\n const { runMs, snapshotMs } = r.store;\n kept =\n (runMs !== undefined\n ? `A cell’s recorded traffic is kept for ${duration(runMs)} from your ` +\n `client’s first request to it; after that the cell reads as not tried. `\n : `A cell’s recorded traffic is kept in this deployment’s store for as long as it is set to keep it. `) +\n (snapshotMs !== undefined\n ? `A frozen copy is kept for ${duration(snapshotMs)} from when it was ` +\n `frozen, so freeze one to keep a result. `\n : '') +\n `A cell with no request for ${idle} leaves memory and is rebuilt from ` +\n `its stored traffic on its next request, which loses nothing.`;\n }\n return (\n `

How long results last

` +\n `

Anyone with a run’s link can read its results. ${kept} ` +\n `DELETE /results/<run-id> removes a run and its frozen copies at once.

`\n );\n}\n\n/**\n * The landing page: what this is, how to test a client in five minutes,\n * what a score means, what a result is not, how long results last, then\n * the static matrix.\n */\nexport function renderLanding(\n origin: string,\n matrix: HostedMatrix,\n retention: LandingRetention,\n build?: BuildInfo\n): string {\n const cells = matrix.cells();\n const startableCells = cells.filter((c) => c.startable);\n const scored = matrix.revisions\n .map((r) => {\n const set = cells.filter(\n (c) => c.revision === r && c.scoring === 'scored'\n );\n const here = set.filter((c) => c.startable).length;\n return `${esc(r)} scores ${set.length} and can start ${here}`;\n })\n .join('; ');\n const states = STATE_MEANING.map(\n ([state, meaning]) => `

${statePill(state)}

${esc(meaning)}

`\n ).join('');\n const specs = matrix.revisions\n .map((r) => `${esc(r)}`)\n .join(' · ');\n return page(\n 'MCP Conformance — hosted',\n `

MCP Conformance — hosted

This server tests an MCP client that you point at it by hand, such as an\nIDE, a desktop agent or a deployed client that a script cannot drive. Each cell\nof the matrix below is one scenario at one specification revision, with its own\nMCP URL.

\n\n

Test your client in five minutes

Start a new run. The run page holds every URL for that\nrun; keep to that one run.
On the run page, add the one URL per revision that carries most of its\nscenarios without a sign-in. Each auth scenario needs a URL of its own; start with\nauth/metadata-default. Pick your client gives those URLs as one\nblock in your client’s format, for VS Code, Codex, Goose, Copilot CLI and any client\nthat reads mcpServers JSON, and every URL shows whether your client has\nreached it yet.
Make your client do what each cell’s steps say, usually list the tools and call\none. For an auth cell, connect and approve the sign-in: the test authorization\nserver approves at once, with no account.
Read the results at /results/<run-id>, linked from the run\npage, which updates itself while it is open. Freeze a copy to link to gives a\nlink that never changes, and you can copy the report as Markdown or plain text for\nan issue or a chat.

Some clients speak 2026-07-28 only behind an opt-in (Codex CLI:\ncodex --enable mcp_2026_07_28), and many pick up new servers only when\nthey start or open a new chat.

\n\n

What the score means

A report column reads “X of the M scored cells you can run here pass · N are scored\nfor the revision”: X is how many have passed, M is how many of the N cells the\nrevision’s frozen requirement set\nscores this deployment can start, and N is the set’s count, whether or not each\ncan start here: ${scored}. Cells marked not scored or not in the requirement set run and report\nbut never count. Verdicts come only from what your client actually sent to this\nserver, and a cell passes only once your client has spoken that cell’s revision\nthere: a finished sign-in or an older handshake is not enough.

${states}\n\n

Self-reported, and not a tier

Results describe what your client did against this server. Nothing stops\nanyone from testing a modified client, so to offer a result as evidence, share a\nfrozen copy’s link: it is dated, and anyone can check it by pointing the same client\nbuild at a new run. Everything sent to a run’s URLs counts as your client’s, so\ndebug with other tools on a separate run.

These results are not an SDK tier measurement. SDK tiers (SEP-1730) are assessed\nwith tier-check, whose conformance pass rates come from the CLI runner\ndriving the SDK’s own conformance client and server\n(SDK Tier Assessment). This server is\nfor clients that cannot be scripted that way.

\n\n${retentionHtml(retention)}\n\n

Source, CLI and docs: modelcontextprotocol/conformance.\nSpecification: ${specs}. A check looks wrong or a scenario misbehaves?\nOpen an issue with the frozen Markdown report.

\n\n

The matrix

${matrix.rows.length} scenarios × ${matrix.revisions.length} revisions,\n${esc(startableCount(startableCells, matrix.revisions))} here. A cell’s MCP URL is\n${esc(origin)}/s/<run-id>/<revision>/<scenario>${MCP_PATH}, and\nits results sit at the same path under /results. Cells that show\nsteps tell a generic client what to do\n(MCP_CONFORMANCE_CONTEXT.steps). JSON.

\n${renderMatrixTable(matrix, { origin })}`,\n build\n );\n}\n\nfunction crumbs(config: RunConfig): string {\n const parts = [\n `matrix`,\n `run ${esc(config.runId)}`\n ];\n if (config.revision) {\n parts.push(\n `${esc(config.revision)}`\n );\n }\n if (config.scenario) parts.push(`${esc(config.scenario)}`);\n return `

${parts.join(' › ')} · results

`;\n}\n\n/**\n * The copy-to-clipboard script. The config is embedded as JSON in a\n * `;\n\nexport function envPre(cell: CellConfig): string {\n const lines = Object.entries(cell.env).map(\n ([k, v]) => `${k}=${JSON.stringify(v)}`\n );\n return `

${esc(lines.join('\\n'))}

`;\n}\n\n/**\n * Ready-to-paste config per client (see ./client-config.ts), each with a\n * copy button and a line on adding it to a file that already has servers:\n * one cell's or composite's entry, or every entry of a run on the bulk page\n * (./run-page.ts), where `copy` names what the button copies.\n */\nexport function clientBlocks(\n entries: readonly ServerEntry[],\n intro: string,\n copy = 'copy'\n): string {\n if (!entries.length) return '';\n const blocks = CLIENTS.map((client) => {\n const text = clientConfig(client.kind, entries);\n return (\n `

${esc(client.label)} ${esc(client.where)}` +\n `

${esc(client.merge)}

` +\n `

${esc(text)}

`\n );\n });\n return `

Paste into your client

${intro}

${blocks.join('')}`;\n}\n\n/** Config page for a run, a column or a cell. */\nexport function renderConfig(\n origin: string,\n matrix: HostedMatrix,\n config: RunConfig,\n build?: BuildInfo\n): string {\n const title = config.scenario\n ? `${config.scenario} @ ${config.revision} — run ${config.runId}`\n : config.revision\n ? `run ${config.runId} @ ${config.revision}`\n : `run ${config.runId}`;\n const embedded = ``;\n\n let body: string;\n if (config.scenario && config.revision) {\n const cell = config.cells[0];\n const key = `${config.revision}/${config.scenario}`;\n const row = matrix.rows.find((r) => r.scenario === config.scenario);\n body = `

`${esc(config.scenario)}` @ ${esc(\n config.revision\n )}

\n${crumbs(config)}\n

${prose(row?.description ?? '', true)}

\n${handNote(config.scenario)}\n

${scoringPill(matrix.cell(config.scenario, config.revision)!)}${\n cell.reason ? ` ${esc(cell.reason)}` : ''\n }

MCP endpoint

${esc(cell.url)}

\n\n— the URL alone, or an mcpServers entry plus the env the CLI runner would set

\n${clientBlocks(\n [{ name: serverName(cell.revision, cell.scenario), url: cell.url }],\n 'This cell as each client’s config file wants it.'\n)}\n${credentials(cell)}\n${\n cell.steps\n ? `

Steps

What to make the client do here. A generic client reads the same steps from MCP_CONFORMANCE_CONTEXT.steps.

${stepsOpen(\n cell.steps\n )}`\n : authSteps(config.scenario)\n}\n

Environment

\n${envPre(cell)}\n

results for this cell

`;\n } else {\n // A run or a column: renderRunPage() in ./run-page.ts, with live status.\n throw new Error('renderConfig renders one cell; see renderRunPage()');\n }\n return page(title, `${body}\\n${embedded}\\n${copyScript}`, build);\n}\n\n/**\n * A composite's page: the one URL to give the client, then what to make the\n * client do for each scenario behind it, and where each result lands.\n */\nexport function renderComposite(view: CompositeView): string {\n const children = view.children\n .map(\n (c) => `

`${esc(c.scenario)}`

${prose(c.description, true)}

${handNote(c.scenario)}${c.steps ? stepsOpen(c.steps) : ''}\n

results for this scenario

`\n )\n .join('\\n');\n return page(\n `composite @ ${view.revision} — run ${view.runId}`,\n `

${view.children.length} scenarios, one URL @ ${esc(view.revision)}

run ${esc(view.runId)}

Give the client under test this MCP URL. Each scenario below still records\nand scores on its own, so its results page reads exactly as if the client had\nbeen pointed at it directly.

MCP endpoint

${esc(view.url)}

\n${clientBlocks(\n [\n {\n name: compositeName(\n view.revision,\n view.children.map((c) => c.scenario)\n ),\n url: view.url\n }\n ],\n 'This composite as each client’s config file wants it.'\n)}\n

Scenarios behind it

\n${children}\n

results for the whole run at ${esc(view.revision)}

\n${copyScript}`\n );\n}\n\n/** How often a live page checks for news, in seconds. */\nexport const LIVE_SECONDS = 10;\n\n/**\n * What keeps a live page (the run report, a cell's results) current while\n * a person drives a client: every LIVE_SECONDS, while the tab is visible,\n * the page fetches itself and swaps in its #live part if that changed,\n * keeping open whatever was open: a fold by its id when it has one (its\n * summary may carry live counts), else by its summary. A note says so and\n * has a stop button. A frozen copy never has it.\n */\nexport const liveNote =\n `

Updates every ${LIVE_SECONDS} s while this tab is open ` +\n `

`;\n\nexport const liveScript = ``;\n\n/** One line saying where the cell stands, for the cell results page. */\nexport function statusLine(status: CellStatus): string {\n const pill = statePill(status.state);\n const scoring = `${SCORING_LABEL[status.scoring]}`;\n let note = '';\n if (status.verdict === 'n/a') {\n note = `the scenario does not apply to this revision: ${esc(status.reason ?? '')}`;\n } else if (status.startable === false) {\n note = `not startable here: ${esc(status.startReason ?? '')}`;\n } else if (status.verdict === 'incomplete') {\n note = esc(status.note ?? incompleteNote([]));\n } else if (status.state === 'waiting' || status.state === 'stopped') {\n note = esc(status.note ?? '');\n } else if (status.reason) {\n note = esc(status.reason);\n }\n return `

${pill} ${scoring}${note ? ` — ${note}` : ''}

`;\n}\n\n/** Style of a \"not seen\" pill: the scenario's expectation, not a failure. */\nconst NOT_SEEN_STYLE = STATUS_STYLE.SKIPPED;\n\n/**\n * What a check is about, under its reason: its name and description, each\n * said once, and not the \"Expected Check Missing\" placeholder some auth\n * scenarios give an expectation nothing met.\n */\nexport function about(c: ConformanceCheck): string {\n const parts = [c.name, c.description].filter(\n (p, i, all) =>\n p && !p.startsWith('Expected Check Missing') && all.indexOf(p) === i\n );\n return parts.map((p) => esc(p)).join(': ');\n}\n\nexport function renderResults(\n ref: CellRef,\n checks: ShownCheck[],\n status?: CellStatus,\n build?: BuildInfo\n): string {\n const items = checks\n .map((c) => {\n const pill =\n (c.notSeen\n ? `not seen`\n : `${c.status}`) +\n (c.repeats\n ? ` recorded ${c.repeats} times`\n : '');\n const refs = (c.specReferences ?? [])\n .map((r) =>\n r.url\n ? `${esc(r.id)}`\n : `${esc(r.id)}`\n )\n .join(' · ');\n const details =\n c.details || c.errorMessage\n ? `

details

${esc(\n              JSON.stringify(\n                { errorMessage: c.errorMessage, ...c.details },\n                null,\n                2\n              )\n            )}

`\n : '';\n // A failure or warning leads with what went wrong (see ./shown.ts);\n // which check it is and what that check is about follow underneath.\n const said = about(c);\n const head = c.reason\n ? `

${pill} ${esc(c.reason)}

${esc(c.id)}${\n said ? ` — ${said}` : ''\n }

`\n : `

${pill} `${esc(c.id)}` — ${esc(\n c.name\n )}

${esc(c.description)}

`;\n return `

${head}${refs ? `

${refs}

` : ''}${details}

`;\n })\n .join('');\n // Skipped checks are said apart: nothing was checked, so they are\n // neither a pass nor a failure.\n const counts = summarize(checks);\n return page(\n `${ref.scenarioName} @ ${ref.revision} — ${ref.runId}`,\n `

`${esc(ref.scenarioName)}` @ ${esc(ref.revision)}

run ${esc(\n ref.runId\n )} › ${esc(ref.revision)} › ${esc(ref.scenarioName)} · config

\n${liveNote}

${status ? statusLine(status) : ''}

${esc(countsLine(counts))}${\n counts.skipped\n ? ` · ${counts.skipped} skipped: the client did nothing they check`\n : ''\n }

${items}

${liveScript}`,\n build\n );\n}\n\nexport function identityLine(identities: ClientIdentity[]): string {\n if (!identities.length) return 'no client seen yet';\n return identities\n .map((i) => {\n const who = i.name\n ? `${esc(i.name)}${i.version ? ` ${esc(i.version)}` : ''}`\n : 'unnamed client';\n const proto = i.protocolVersions.length\n ? ` · protocol ${i.protocolVersions\n .map((v) => `${esc(v)}`)\n .join(', ')}`\n : '';\n const ua = i.userAgent\n ? ` (${esc(\n i.userAgent.length > 40\n ? i.userAgent.slice(0, 40) + '…'\n : i.userAgent\n )})`\n : '';\n return `${who}${proto}${ua}`;\n })\n .join('
');\n}\n\nexport function statePill(state: CellState): string {\n return ` ${STATE_LABEL[state]}`;\n}\n\nfunction verdictCell(cell: CellReport): string {\n if (cell.verdict === 'n/a') {\n return `n/a — ${esc(cell.reason ?? '')}`;\n }\n let lines = `

${statePill(cell.state)} ${scoringPillFor(cell)}

`;\n const link = (text: string, title = '') =>\n `

${text}

`;\n if (cell.state === 'in-progress') {\n // Its failures are only what the scenario still expects (see\n // incompleteNote()): counting them here would read as a verdict.\n lines += link('reached, nothing tested yet', cell.note);\n } else if (cell.state === 'incomplete') {\n lines += link('reached, stopped short', cell.note);\n } else if (cell.summary) {\n lines += link(countsLine(cell.summary), cell.note);\n } else if (!cell.startable) {\n lines += `

unavailable on this deployment

`;\n } else {\n lines += link('not tried');\n }\n return `${lines}`;\n}\n\n/** \"(cause 2)\" when a finding's or a cell's cause covers other cells too. */\nfunction causeLink(\n key: string | undefined,\n causes: ReadonlyMap,\n numbers: ReadonlyMap\n): string {\n const cause = key ? causes.get(key) : undefined;\n if (!cause || cause.cells.length < 2) return '';\n const n = numbers.get(cause.key)!;\n return ` cause ${n}`;\n}\n\n/** \"the flow has not reached a, …\", or the reason. */\nfunction unmetHtml(u: Unmet): string {\n return u.checks\n ? `the flow has not reached ${u.checks.map((c) => `${esc(c)}`).join(', ')}`\n : esc(u.reason);\n}\n\n/** A reached cell's \"what happened\": its findings, or why it stopped. */\nfunction happenedHtml(\n cell: CellReport,\n causes: ReadonlyMap,\n numbers: ReadonlyMap\n): string {\n const findings = cell.findings ?? [];\n if (cell.state === 'in-progress') {\n return findings.length\n ? `

waiting for:

${waitingFor(cell)\n .map((u) => `

${unmetHtml(u)}

`)\n .join('')}`\n : `

${esc(cell.note ?? '')}

`;\n }\n if (cell.state === 'incomplete') {\n return `

${esc(stopNote(cell))}${causeLink(cell.cause, causes, numbers)}

`;\n }\n // A waiting cell's own expectations are what it waits for; anything the\n // client did (a warning) is listed as on any other row.\n const waiting = cell.state === 'waiting' || cell.state === 'stopped';\n const lines = findings\n .filter((f) => !waiting || f.by === 'client')\n .map(\n (f) =>\n `

${BY_LABEL[f.by]} ${esc(f.check)} ${esc(\n f.reason\n )}${causeLink(f.cause, causes, numbers)}

`\n );\n if (waiting) {\n lines.unshift(\n `

${esc(cell.note ?? '')}; not seen yet:

` +\n waitingFor(cell)\n .map((u) => `

${unmetHtml(u)}

`)\n .join('')\n );\n }\n return lines.length\n ? lines.join('')\n : `${NO_FINDINGS}`;\n}\n\n/** What a not-tried cell wants: the client's steps, and its URL to copy. */\nfunction notTriedHtml(cell: CellReport): string {\n const hint = cell.hint\n ? `

the client must ${esc(cell.hint)}

`\n : '

nothing reached this cell

';\n const url = cell.mcpUrl\n ? `

` +\n `${esc(cell.mcpUrl)}

`\n : '';\n return hint + url;\n}\n\n/**\n * One revision's rows: what needs a look (fail, waiting, in progress,\n * incomplete), then what the client has not tried, then what passed. Cells\n * this deployment cannot start are not here (see unavailableSection()).\n */\nfunction revisionRows(\n col: ColumnReport,\n causes: ReadonlyMap,\n numbers: ReadonlyMap,\n runId?: string\n): string {\n const { problems, notTried, passed } = groupColumn(col);\n if (!problems.length && !passed.length && !notTried.length) return '';\n const link = (cell: CellReport) =>\n `${esc(cell.scenario)}`;\n // What needs a look can be reset from here (the live report only): the\n // client's next request starts a new attempt, which sets the verdict.\n const reset = (cell: CellReport) =>\n runId && cell.startable\n ? ``\n : '';\n const reached = (cell: CellReport, problem = false) => {\n const s = cell.summary;\n const counts =\n s && showsCounts(cell)\n ? `${s.passed} / ${s.failed} / ${s.warnings}`\n : '–';\n const attempt = cell.attempt\n ? ` attempt ${cell.attempt}`\n : '';\n return (\n `${link(cell)}` +\n `${statePill(cell.state)}${attempt}${counts}` +\n `${happenedHtml(cell, causes, numbers)}${\n problem ? reset(cell) : ''\n }`\n );\n };\n let rows = `${esc(col.revision)}`;\n rows += problems.map((cell) => reached(cell, true)).join('');\n if (notTried.length && !problems.length && !passed.length) {\n // Nothing reached at this revision: every startable cell is here, so\n // fold them rather than push the revisions the client did speak down.\n const items = notTried\n .map((cell) => `

${link(cell)}${notTriedHtml(cell)}

`)\n .join('');\n rows +=\n `

${esc(\n notTriedHeading(notTried.length, false)\n )}

${esc(NOT_TRIED_WHY)}

` +\n `

${items}

`;\n } else if (notTried.length) {\n rows +=\n `${esc(\n notTriedHeading(notTried.length, true)\n )}

${esc(NOT_TRIED_WHY)}

` +\n notTried\n .map(\n (cell) =>\n `${link(cell)}${statePill(cell.state)}` +\n `–${notTriedHtml(cell)}`\n )\n .join('');\n if (passed.length) {\n rows += `${esc(passedHeading(passed.length))}`;\n }\n }\n return rows + passed.map((cell) => reached(cell)).join('');\n}\n\n/** Every revision's cells in its groups, one table. */\nfunction revisionsTable(report: RunReport): string {\n const causes = new Map(report.causes.map((c) => [c.key, c]));\n const numbers = causeNumbers(report.causes);\n const groups = report.columns\n .map((col) =>\n revisionRows(\n col,\n causes,\n numbers,\n report.frozenAt ? undefined : report.runId\n )\n )\n .join('');\n const reachedAny = report.columns.some((col) =>\n col.cells.some((c) => REACHED.includes(c.state))\n );\n const lead = reachedAny\n ? ''\n : '

No cell has been reached yet: point the client at a cell’s MCP URL.

';\n if (!groups) return lead;\n return (\n lead +\n `` +\n `${groups}

cell	result	pass / fail / warn	what happened

`\n );\n}\n\n/**\n * The cells this deployment cannot start, once per scenario with its\n * revisions and the reason, folded at the bottom of the page.\n */\nfunction unavailableSection(report: RunReport): string {\n const list = unavailableScenarios(report);\n if (!list.length) return '';\n const items = list\n .map(\n (u) =>\n `

${esc(u.scenario)} ${u.revisions\n .map(esc)\n .join(', ')} — ${esc(u.reason)}

`\n )\n .join('');\n return (\n `

${esc(\n UNAVAILABLE_HEADING\n )} (${list.length})

` +\n `

${esc(UNAVAILABLE_WHY)}

${items}

`\n );\n}\n\n/** Each cause once, with the cells it covers. */\nfunction causesList(report: RunReport): string {\n if (!report.causes.length) return '';\n const items = report.causes.map((c, i) => {\n const cells = c.cells\n .map(\n (key) =>\n `${esc(key)}`\n )\n .join(', ');\n return (\n `

${BY_LABEL[c.by]} ${c.check ? `${esc(c.check)} ` : ''}${esc(c.text)}` +\n `

${c.cells.length} cell${c.cells.length === 1 ? '' : 's'}: ${cells}

`\n );\n });\n return (\n `

What went wrong, by cause

Each cause once, with the cells it covers. ` +\n `client: seen in the client’s traffic. not seen: the scenario’s own ` +\n `expectation that nothing has met yet (the client may not have got that far).

` +\n `

${items.join('')}`\n );\n}\n\nfunction scoringPillFor(cell: CellReport): string {\n return `${SCORING_LABEL[cell.scoring]}`;\n}\n\nexport interface ReportPageOptions {\n /** The report as Markdown (see ./markdown.ts), for the copy button. */\n markdown: string;\n /** The same as plain lines (reportText()), for a chat without tables. */\n text: string;\n /** The live report, which a frozen copy links back to. */\n liveUrl: string;\n /** The run's frozen copies, listed on the live report. */\n snapshots?: readonly SnapshotInfo[];\n}\n\n/** The line under the heading: copy, other formats, freeze or \"frozen\". */\nfunction reportActions(report: RunReport, opts: ReportPageOptions): string {\n const run = esc(report.runId);\n const here = report.snapshotId\n ? `/results/${run}/snapshot/${esc(report.snapshotId)}`\n : `/results/${run}${report.revision ? `/${esc(report.revision)}` : ''}`;\n const formats =\n ` ` +\n ` ` +\n `Markdown · plain text · JSON`;\n if (report.frozenAt) {\n return (\n `

A frozen copy, taken ${esc(utcMinute(report.frozenAt))}: it does not change ` +\n `as more traffic arrives, so it is safe to link from an issue or a chat. ` +\n `The live report has anything since; ` +\n `${\n report.frozenCells\n ? 'each cell your client reached links to its page as frozen with this copy, traffic included.'\n : 'the cell links open the live results.'\n }

${formats}

`\n );\n }\n const earlier = opts.snapshots?.length\n ? `

Frozen copies of this run: ${opts.snapshots\n .map(\n (s) =>\n `${esc(\n utcMinute(new Date(s.createdAt).toISOString())\n )}`\n )\n .join(' · ')}

`\n : '';\n return (\n `

${formats} · ` +\n ` ` +\n `— a permalink to the whole run as it stands now; later traffic will not change it

` +\n earlier\n );\n}\n\n/** Report page for a run or one of its columns, live or frozen. */\nexport function renderReport(\n matrix: HostedMatrix,\n report: RunReport,\n opts: ReportPageOptions\n): string {\n const title = report.frozenAt\n ? `results — run ${report.runId}, frozen ${utcMinute(report.frozenAt)}`\n : report.revision\n ? `results — run ${report.runId} @ ${report.revision}`\n : `results — run ${report.runId}`;\n const head =\n `scenario` +\n report.columns\n .map(\n (col) =>\n `${esc(\n col.revision\n )}

${esc(scoreText(col))}

` +\n `

${identityLine(col.identities)}

`\n )\n .join('') +\n '';\n // A scenario this deployment cannot start at any revision shown is listed\n // once at the bottom (unavailableSection()), not as a greyed-out row.\n const hidden = new Set(['not-startable', 'n/a']);\n const rows = matrix.rows\n .map((row) =>\n report.columns.map(\n (col) => col.cells.find((c) => c.scenario === row.scenario)!\n )\n )\n .filter((cells) => cells.some((c) => !hidden.has(c.state)))\n .map(\n (cells) =>\n `${esc(cells[0].scenario)}${cells\n .map(verdictCell)\n .join('')}`\n )\n .join('');\n const notScored = report.columns\n .map((col) => {\n if (!col.notScored.length) return '';\n const items = col.notScored\n .map(\n (c) =>\n `

${esc(c.scenario)} ${statePill(\n c.state\n )} ${esc(\n SCORING_LABEL[c.scoring]\n )}${c.reason ? ` — ${esc(c.reason)}` : ''} · checks

`\n )\n .join('');\n return `

${esc(col.revision)}: run but not scored

${items}`;\n })\n .join('');\n const run = esc(report.runId);\n const crumbs = [\n `matrix`,\n `run ${run}`\n ];\n if (report.revision) crumbs.push(`${esc(report.revision)}`);\n if (report.frozenAt) {\n crumbs.push(`frozen ${esc(utcMinute(report.frozenAt))}`);\n }\n crumbs.push(\n `config`\n );\n const summary = report.columns\n .map((col) => {\n const counts = countsText(col.counts, [...REACHED, 'not-tried']);\n return (\n `

${esc(col.revision)}: ` +\n `${esc(scoreText(col))}` +\n `${counts ? ` · reached: ${esc(counts)}` : ''}

`\n );\n })\n .join('');\n return page(\n title,\n `

results — run `${run}`${\n report.revision ? ` @ ${esc(report.revision)}` : ''\n }${report.frozenAt ? ' (frozen)' : ''}

${crumbs.join(' › ')}

\n${report.frozenAt ? '' : liveNote}

Client: ${identityLine(report.identities)}

\n${reportActions(report, opts)}\n

Summary

${summary}\n${causesList(report)}\n

Cells by revision

Each revision lists what needs a look first, then the cells your\nclient has not tried, then what passed. Cells this deployment cannot start are\nlisted once, folded, at the bottom of the page.

\n${revisionsTable(report)}\n

as Markdown, to paste into an issue or a chat

${esc(\n      opts.markdown\n    )}

Every cell

A cell passes when checks were recorded and none is a FAILURE.\nA cell the client never reached reads not tried; one it reached where\nnothing its scenario tests has happened yet reads in progress and lists\nwhat it is waiting for; one whose only failures are steps it has not seen yet\n(a sign-in or a form still to finish) reads waiting, though its verdict\nis still a fail until they happen, or stopped once the client has sent nothing\nfor a minute with no sign-in or form open; one where the client stopped short (it spoke only\nanother revision, was turned away, and did not retry) reads incomplete.\nThe score leads with the scored cells you can run here (X of the M scored\ncells you can run here pass) and keeps the requirement set's count\n(N are scored); the other N−M cannot start on this deployment. Not-scored and unlisted cells are listed below the table;\nscenarios this deployment cannot start are left out of it.

\n${head}${rows}

\n${notScored}\n${unavailableSection(report)}

\n${report.frozenAt ? '' : liveScript}\n${copyScript}`,\n report.server ?? null\n );\n}\n","deleted",false,"label",[47,48,24],"src","hosted","7dcb6042-d92f-40c5-9cc8-0f661f96e58b",["D",1789043604380],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","imageUrl","userId","f4518a30-f52b-4b12-891c-a8fffb7240f8","pinnedFileId","e6dd13e2-ad13-11f1-a612-1607ee4eb77e","pinPreview","pinMigrated","screenshotVersion","likeCount",0,"orgId","01a07c12-b73a-76de-b96a-421ce8582af6","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_9":10,"_105":-5},"org",{"_9":10,"_99":100,"_101":-5,"_58":-5,"_102":103,"_104":44},"branch",{"_37":89,"_11":90,"_22":30,"_91":49,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":-5,"_97":12,"_98":10},"isOwned",396,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","displayName","MCP Conformance","avatar","tier","free","canRestrictVals","profileImageUrl","ctx",{"_82":-5,"_139":140,"_141":142,"_143":144,"_145":44,"_146":44,"_147":44,"_148":34,"_149":-5,"_150":-7,"_151":-7,"_152":66,"_153":44,"_154":44,"_155":-5,"_156":44,"_157":44,"_158":44,"_159":-5},"ENV",{"_110":111,"_112":113,"_114":90,"_115":116,"_117":118,"_119":120,"_121":122,"_123":124,"_125":126,"_127":128,"_129":130,"_131":132,"_133":134,"_135":136,"_137":138},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town","orgs",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]

Steps

Credentials

How long results last

MCP Conformance — hosted

Test your client in five minutes

What the score means

Self-reported, and not a tier

The matrix

Paste into your client

${esc(config.scenario)} @ ${esc(\n config.revision\n )}

MCP endpoint

Steps

Environment

${esc(c.scenario)}

${view.children.length} scenarios, one URL @ ${esc(view.revision)}

MCP endpoint

Scenarios behind it

${pill} ${esc(c.reason)}

${pill} ${esc(c.id)} — ${esc(\n c.name\n )}

${esc(ref.scenarioName)} @ ${esc(ref.revision)}

What went wrong, by cause

${esc(col.revision)}: run but not scored

results — run ${run}${\n report.revision ? ` @ ${esc(report.revision)}` : ''\n }${report.frozenAt ? ' (frozen)' : ''}

Summary

Cells by revision

Every cell

`${esc(config.scenario)}` @ ${esc(\n config.revision\n )}

`${esc(c.scenario)}`

${pill} `${esc(c.id)}` — ${esc(\n c.name\n )}

`${esc(ref.scenarioName)}` @ ${esc(ref.revision)}

results — run `${run}`${\n report.revision ? ` @ ${esc(report.revision)}` : ''\n }${report.frozenAt ? ' (frozen)' : ''}