ide-diagnostics-board

name	ide-diagnostics-board
description	Diagnostic dashboard across the workspace. Calls getDiagnostics, groups results by severity and file, and renders a sortable color-coded HTML table. Opens in the system browser.
argument-hint	[error|warning|all]
effort	low

Generate a visual diagnostics dashboard for the workspace and open it in the browser.

Prerequisites

1. Check if the getToolCapabilities MCP tool is available to you.
   • Available: call it, check extensionConnected → use IDE Path below.
   • Not available (no MCP tool by that name): use CLI Path below.

Arguments

$ARGUMENTS sets the severity filter:

• error — errors only
• warning — warnings and errors
• all (default) — all severities

---

IDE Path

Use this path when bridge MCP tools are available.

Phase 1 — Collect diagnostics

1. Determine the severity filter from $ARGUMENTS (default: all).
2. Call getDiagnostics with no URI filter (full workspace scan). Pass the severity param if supported.
3. If no diagnostics found, report "No diagnostics found — workspace is clean." and stop.

Phase 2 — Summarize

4. Group diagnostics by severity: errors / warnings / info / hints.
5. Group by file: sort files by error count descending.
6. Compute totals: total diagnostic count, count per severity, number of files affected.

Phase 3 — Generate and open HTML

7. Construct the HTML document (see HTML Template section below).
8. Call openInBrowser with the HTML string and filename diagnostics-<timestamp>.html.
9. Report: "Opened diagnostics board — N errors, M warnings across K files. Saved to: "

---

CLI Path

Use this path when bridge MCP tools are NOT available (e.g., remote-control sessions).

Phase 1 — Detect and run linters

1. Determine the severity filter from $ARGUMENTS (default: all).
2. Detect which linters are available by checking for config files using Glob:
   • **/tsconfig.json (not in node_modules) → run via Bash: npx tsc --noEmit 2>&1
   • **/biome.json or **/biome.jsonc → run via Bash: npx biome check . --reporter=json 2>&1
   • **/.eslintrc* or **/eslint.config.* → run via Bash: npx eslint . --format json 2>&1
   • **/pyrightconfig.json or **/*.py → run via Bash: npx pyright --outputjson 2>&1
   • **/ruff.toml or **/pyproject.toml → run via Bash: ruff check . --output-format json 2>&1
   • **/Cargo.toml → run via Bash: cargo check --message-format json 2>&1
3. If no linters detected, report "No linter configuration found in workspace." and stop.

Phase 2 — Parse linter output

4. Parse each linter's output into a unified list of { severity, file, line, col, message, source }:
   • tsc: lines like file(line,col): error TSxxxx: message — parse with regex
   • biome JSON: diagnostics array with severity, location.path, location.span
   • eslint JSON: [{filePath, messages:[{severity,line,column,message,ruleId}]}]
   • pyright JSON: generalDiagnostics array with range.start.line, severity, message
   • ruff JSON: [{filename, location:{row,column}, message, code}]
   • cargo JSON (NDJSON): message.level, message.spans[0].file_name, message.spans[0].line_start
5. Apply the severity filter from $ARGUMENTS.
6. If no diagnostics after filtering, report "No diagnostics found — workspace is clean." and stop.

Phase 3 — Summarize and generate HTML

7. Group diagnostics by severity and by file (same as IDE Path).
8. Compute totals.
9. Construct the HTML document (see HTML Template section below).
10. Write the HTML to diagnostics-<timestamp>.html in the workspace root using the Write tool.
11. Open via Bash: open diagnostics-<timestamp>.html (macOS) or xdg-open diagnostics-<timestamp>.html (Linux).
12. Report: "Opened diagnostics board — N errors, M warnings across K files. Saved to: "

---

HTML Template

Construct a fully self-contained HTML document (no external URLs — must work from file://) containing:

• A summary bar: "N errors · M warnings · K files affected · generated "
• Color-coded severity badges: error=#c0392b, warning=#e67e22, info=#2980b9, hint=#27ae60
• A filter input (plain text, JS filters rows in real time)
• Column header clicks sort the table (vanilla JS, ~60 lines)
• Columns: Severity | File (relative path) | Line:Col | Message | Source
• Row background: #fee2e2 error, #fef9c3 warning, #eff6ff info, #f0fdf4 hint
• A "Copy as Markdown" button that copies a plain Markdown table to clipboard
• Page title: "Diagnostics — "

HTML constraints

• No external URLs or CDN links
• All CSS in <style>, all JS in <script>
• Target 300–500 lines of HTML