← WebMCP Directory

Tool tree

sightline-5vu.pages.dev

Developer ToolsTools verified
Website
Sightline is the human-approval layer between an agent and a live page. Through WebMCP, an agent finds accessibility, SEO and performance problems, drafts the fix, and stops until you say yes.
Explore capabilities
    list_packsAnswer

    List the rule packs this page exposes (accessibility, seo, performance): id, label, what each checks, and which fixes require agent-authored content. Call this first to decide which packs to scan.

    View tool JSON
    {
      "name": "list_packs",
      "kind": "answer",
      "impl": "imperative",
      "description": "List the rule packs this page exposes (accessibility, seo, performance): id, label, what each checks, and which fixes require agent-authored content. Call this first to decide which packs to scan.",
      "inputSchema": {
        "type": "object",
        "properties": {},
        "additionalProperties": false
      }
    }
    scan_pageAction

    Run Sightline’s audit over the live page with one or more rule packs (default: all). Returns stable issue IDs with packId, rule, impact, selector, and per-pack plus overall 0–100 score. Start every workflow here.

    View tool JSON
    {
      "name": "scan_page",
      "kind": "act",
      "impl": "imperative",
      "description": "Run Sightline’s audit over the live page with one or more rule packs (default: all). Returns stable issue IDs with packId, rule, impact, selector, and per-pack plus overall 0–100 score. Start every workflow here.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "wcagLevel": {
            "type": "string",
            "enum": [
              "A",
              "AA",
              "AAA"
            ],
            "default": "AA",
            "description": "Requested WCAG conformance target (accessibility pack)."
          },
          "packs": {
            "type": "array",
            "items": {
              "type": "string",
              "enum": [
                "accessibility",
                "seo",
                "performance"
              ]
            },
            "description": "Rule packs to run. Omit to run all."
          }
        },
        "additionalProperties": false
      }
    }
    navigate_nodeAction

    Navigate the shared preview to a CSS selector and return a short HTML/text snapshot. Use it to understand surrounding page context without changing anything.

    View tool JSON
    {
      "name": "navigate_node",
      "kind": "act",
      "impl": "imperative",
      "description": "Navigate the shared preview to a CSS selector and return a short HTML/text snapshot. Use it to understand surrounding page context without changing anything.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "selector": {
            "type": "string",
            "description": "CSS selector inside the audited preview."
          }
        },
        "required": [
          "selector"
        ],
        "additionalProperties": false
      }
    }
    highlight_issueAction

    Move the human-visible focus overlay to one current audit issue. This creates shared visual context before discussing a repair.

    View tool JSON
    {
      "name": "highlight_issue",
      "kind": "act",
      "impl": "imperative",
      "description": "Move the human-visible focus overlay to one current audit issue. This creates shared visual context before discussing a repair.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "issueId": {
            "type": "string",
            "description": "A current ID returned by scan_page."
          }
        },
        "required": [
          "issueId"
        ],
        "additionalProperties": false
      }
    }
    propose_fixAction

    Create a selector-scoped patch for one current issue. Does not mutate the page. Where a fix needs human-facing words the AGENT must author them from the page context — image-alt → altText, label → labelText, aria-roles → role, document-title → title, meta-description → description, link-text → linkText. Call without that field first to receive a needs_input response with requiredField, guidance, html and nearbyText, then call again with it. Contrast, heading level, tabindex, lang, image dimensions and rel=noopener are measured by the engine from the live DOM. The returned patch carries evidence[] and authoredBy; pass its id to apply_fix.

    View tool JSON
    {
      "name": "propose_fix",
      "kind": "act",
      "impl": "imperative",
      "description": "Create a selector-scoped patch for one current issue. Does not mutate the page. Where a fix needs human-facing words the AGENT must author them from the page context — image-alt → altText, label → labelText, aria-roles → role, document-title → title, meta-description → description, link-text → linkText. Call without that field first to receive a needs_input response with requiredField, guidance, html and nearbyText, then call again with it. Contrast, heading level, tabindex, lang, image dimensions and rel=noopener are measured by the engine from the live DOM. The returned patch carries evidence[] and authoredBy; pass its id to apply_fix.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "issueId": {
            "type": "string",
            "description": "A current ID returned by scan_page."
          },
          "altText": {
            "type": "string",
            "description": "image-alt only. Agent-authored description of what the image shows, ≤160 chars, plain text. Empty string marks the image decorative."
          },
          "labelText": {
            "type": "string",
            "description": "label only. Agent-authored visible label for the form field, ≤80 chars, in the page language."
          },
          "role": {
            "type": "string",
            "description": "aria-roles only. The valid ARIA role the control should have, chosen from its behaviour (e.g. button, link)."
          },
          "headingLevel": {
            "type": "integer",
            "minimum": 1,
            "maximum": 6,
            "description": "heading-order only. Optional override; omit to accept the level the engine derives from the document outline."
          },
          "title": {
            "type": "string",
            "description": "seo document-title only. Agent-authored page title, 10–60 chars, plain text, in the page language."
          },
          "description": {
            "type": "string",
            "description": "seo meta-description only. Agent-authored summary, 50–155 chars, plain text."
          },
          "linkText": {
            "type": "string",
            "description": "seo link-text only. Agent-authored descriptive link text (2–6 words) replacing generic text like \"click here\"."
          }
        },
        "required": [
          "issueId"
        ],
        "additionalProperties": false
      }
    }
    re_scanAction

    Re-run the focused audit after page changes. Returns the updated score, remaining issue IDs and next recommended barrier.

    View tool JSON
    {
      "name": "re_scan",
      "kind": "act",
      "impl": "imperative",
      "description": "Re-run the focused audit after page changes. Returns the updated score, remaining issue IDs and next recommended barrier.",
      "inputSchema": {
        "type": "object",
        "properties": {
          "scope": {
            "type": "string",
            "enum": [
              "page",
              "changed"
            ],
            "default": "changed",
            "description": "Label this as a full audit or post-change verification."
          }
        },
        "additionalProperties": false
      }
    }