Add vardoger

[dstrupl]

Adds io.github-dstrupl-vardoger — a stdio MCP server that personalizes AI coding assistants by analyzing local conversation history from Cursor, Claude Code, Codex, Copilot CLI, Windsurf, Cline, and OpenClaw. Runs entirely on the user's machine; no API keys, no network egress. Source: https://github.com/dstrupl/vardoger (PyPI: vardoger, Apache-2.0).

Metadata

• Transport: stdio, invoking vardoger mcp (installed via pipx install vardoger, Python ≥3.11).
• Auth: none — reads local filesystem paths only (~/.cursor/, ~/.claude/, ~/.codex/, etc.).
• Platforms: all.
• Categories: developer-tools, productivity, ai-ml.
• Optional input: VARDOGER_MCP_PLATFORM (lets users pin the host platform if the server's auto-detect picks the wrong one — not required in the common case).

Local validation

• node scripts/validate.js servers/io.github-dstrupl-vardoger.json → PASS
• node scripts/validate.js --check-conflicts → No conflicts found across 106 server file(s).
• node scripts/validate.js --all → no FAIL or CONFLICT output

DCO signed off in the commit.

[dstrupl]

Thanks for the review, @its-mash! Great suggestion — just pushed 80e6cb2 to the PR branch that switches VARDOGER_MCP_PLATFORM from text to select.

The options mirror vardoger's seven supported platform identifiers plus an explicit "Auto-detect (default)" entry (empty-string value) that maps to the server's existing auto-detect fallback in _env_default_platform():

"options": [
  { "value": "",            "label": "Auto-detect (default)", "description": "Let vardoger pick based on the calling host." },
  { "value": "cursor",      "label": "Cursor" },
  { "value": "claude-code", "label": "Claude Code" },
  { "value": "codex",       "label": "Codex" },
  { "value": "openclaw",    "label": "OpenClaw" },
  { "value": "copilot",     "label": "GitHub Copilot CLI" },
  { "value": "windsurf",    "label": "Windsurf" },
  { "value": "cline",       "label": "Cline" }
]

I also dropped the now-redundant placeholder and trimmed the "Valid values: …" sentence from description since the options make it obvious.

Happy to drop the empty "Auto-detect" option and make the select strictly one-of-7 if you would prefer a cleaner UX — let me know. Validation still passes (npm run validate:all → 106 PASS / 0 FAIL, npm run check-conflicts → no conflicts).