mcpencil.com WebMCP audit

Name: mcpencil.com WebMCP audit
Creator: ora

audited Aug 31, 2026in 42s

Agent-readyEvery tool this page registers declares a handler an agent can call. WebMCP is not on by default in a release browser yet, so reaching them today takes a WebMCP-capable browser or the page's own polyfill - browser rollout, not a gap in this page.

63/ 100100% of the applicable check weight evaluated

10 tools registered. Strongest in tool quality, weakest in task completion.

Shared experience75

Task completion13

Tool quality90

Trust75

https://mcpencil.com

Tools

2 read, 0 write, 8 undeclared

Loading map

get_match_state

Canonical page-exposed WebMCP continuation state. Before joining, it identifies play_mcpencil as the entry action. After joining, read role, mustContinue, completionCondition, exact nextAction, urgency, deadline, and the privatePrompt when you are the agent artist. Between turns, pass afterRevision and waitMs 25000. Guessers should visually inspect the rendered page/canvas as the primary picture; canvasPerception is a fast 32x22 text raster fallback and compact canvasGeometry is only a final cross-check.Answers a question. Declared read only.

pagehttps://mcpencil.com/

implementation

viaimperative

entry pointdocument

registered after363ms

executepresent

api surface

afterRevisionoptional

waitMsoptional

annotations

read onlytrue

untrusted contenttrue

titlenot declared

2 findings

warningSchema quality

weight 4

5 of 10 tool schemas are harder for an agent to use than they need to be: get_match_state description is 509 characters, over Chrome's 500-character guidance.

get_match_state: description is 509 characters, over Chrome's 500-character guidance; get_match_state: declares no required list, so an agent cannot tell which parameters are mandatory; play_mcpencil:

fix

{
  name: "get_match_state",
  inputSchema: {
    type: "object",
    properties: {
      query: { type: "string", description: "What to search for" }
    },
    required: ["query"]
  }
}

warningInjection surface

weight 4

6 of 10 tools carry instruction-shaped text in their own metadata: get_match_state Description instructs agent how to behave and what to read.

get_match_state: Description instructs agent how to behave and what to read; play_mcpencil: Description instructs agent when to call and how to continue; configure_match: Description instructs agent when to call and what values to send; start_match: Description instructs agent when to start

fix

{
  name: "get_match_state",
  // Metadata DESCRIBES the tool - it never addresses the agent reading it.
  // Rewrite any name, description, title, or schema field description that
  // instructs the agent (which tool to prefer, what to output, rules to
  // ignore) so it states what the tool does and what it returns instead.
  description: "Searches the catalog and returns matching items with prices."
}

tool json

{
  "name": "get_match_state",
  "description": "Canonical page-exposed WebMCP continuation state. Before joining, it identifies play_mcpencil as the entry action. After joining, read role, mustContinue, completionCondition, exact nextAction, urgency, deadline, and the privatePrompt when you are the agent artist. Between turns, pass afterRevision and waitMs 25000. Guessers should visually inspect the rendered page/canvas as the primary picture; canvasPerception is a fast 32x22 text raster fallback and compact canvasGeometry is only a final cross-check.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "afterRevision": {
        "type": "integer",
        "minimum": 0,
        "description": "Optional last revision; wait for something newer."
      },
      "waitMs": {
        "type": "integer",
        "minimum": 0,
        "maximum": 25000,
        "description": "Optional wait for a newer revision. Use 25000 between turns."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "readOnlyHint": true,
    "untrustedContentHint": true
  }
}

Findings

Shared experience

75 / 100 · weight 30

warningPage experience

weight 20

2 of 4 page-experience marks came back short: usable The page is a loading screen, so it is not fully functional yet..

usable: weak; clarity: weak

warningHuman parity

weight 10

The page the person sees came back short of the agent surface (2 of 4 page-experience marks came back short: usable The page is a loading screen, so it is not fully functional yet..), so the two are not sharing an equal experience.

fix

// Make the tool's effect visible on the page the person is looking at:
// update the same UI your existing human flow uses when your_tool_name runs,
// so the person co-browsing sees what the agent just did.
execute: async (args, { signal }) => {
  const result = await performAction(args, signal);
  renderResultInPage(result); // the human-visible half of the same action
  return result;
}

Task completion

13 / 100 · weight 25

failTool selection

weight 15

An agent chose an existing tool and built a schema-valid call that held up on 0 of 3 canonical intents; on "Open the document I was working on most recently." no tool on the page served it. Nothing was executed, so this verifies selection, not outcomes.

failCoverage vs. site type

weight 10

The tool set covers 1 of the 3 things an agent needs on a editor site; it cannot edit or write content or save, publish, or share the result.

Tool quality

90 / 100 · weight 25

passRegistration timing

weight 1

Every measured tool registered within 365ms of navigation.

toolsubmit_guesses

passCanonical entry point

weight 3

All 10 tools register on the canonical document.modelContext entry point.

passSchema validity

weight 4

All 10 declared input schemas are structurally valid object schemas.

warningSchema quality

weight 4

5 of 10 tool schemas are harder for an agent to use than they need to be: get_match_state description is 509 characters, over Chrome's 500-character guidance.

toolget_match_state

get_match_state: description is 509 characters, over Chrome's 500-character guidance; get_match_state: declares no required list, so an agent cannot tell which parameters are mandatory; play_mcpencil:

fix

{
  name: "get_match_state",
  inputSchema: {
    type: "object",
    properties: {
      query: { type: "string", description: "What to search for" }
    },
    required: ["query"]
  }
}

warningNaming quality

weight 2

Tool naming makes selection harder than it needs to be: 5 names do not start with a verb (play_mcpencil).

toolplay_mcpencil

fix

{ name: "play_mcpencil" /* short, unique, verb-based */ }

passStub detection

weight 4

All 10 tools declare an execute handler.

passRegistration errors

weight 2

No tool registration threw during the capture.

passDescription quality

weight 5

All 10 rated tool descriptions say what the tool does, when to use it, and what it returns.

Trust

75 / 100 · weight 20

passAnnotations present

weight 6

All 2 read-shaped tools declare readOnlyHint: true - the one declared claim that lets an agent treat a call as safe to make without asking.

passAnnotation mismatch

weight 6

No tool claims to be read-only while its own name or description says it writes (2 declared hints checked).

not applicableUntrusted content hint

weight 4

No tool's contract suggests it returns text written by somebody other than the site, so there is nothing to flag as untrusted.

warningInjection surface

weight 4

6 of 10 tools carry instruction-shaped text in their own metadata: get_match_state Description instructs agent how to behave and what to read.

toolget_match_state

get_match_state: Description instructs agent how to behave and what to read; play_mcpencil: Description instructs agent when to call and how to continue; configure_match: Description instructs agent when to call and what values to send; start_match: Description instructs agent when to start

fix

{
  name: "get_match_state",
  // Metadata DESCRIBES the tool - it never addresses the agent reading it.
  // Rewrite any name, description, title, or schema field description that
  // instructs the agent (which tool to prefer, what to output, rules to
  // ignore) so it states what the tool does and what it returns instead.
  description: "Searches the catalog and returns matching items with prices."
}

Tool selection

0% across 3 intents

Open the document I was working on most recently.missed

choseno toolexpected document openkind not applicable

No tool was picked for this intent.

arguments

{}

Add a short summary paragraph at the top.missed

choseno toolexpected content editkind not applicable

No tool was picked for this intent.

arguments

{}

Save it and give me a link I can share.missed

choseno toolexpected save or sharekind not applicable

No tool was picked for this intent.

arguments

{}

model: gemini-2.5-flash

Add the tools this site is missing

Our scanner reads your website and suggests the right WebMCP tools for it.

The open source webmcp plugin teaches your coding agent to audit a site, implement tools on document.modelContext, and verify them in a real browser. npx @ora-ai/webmcp-verify runs the verification on its own. No signup, no hosted service.

how this was captured

observed via: capture shim (Chromium 148.0.7778.96)
chrome: 148.0.7778.96
capture shim: v1
spec snapshot: 2026-08-26
mode: fast
pages: 1 - entry page only
tools: 10
run time: 42s
finished: Aug 31, 2026

Checked the way in-browser agents discover tools: the top-level document's modelContext registry, read after the page settles.