UNPKG

comfyui-mcp

Version:

Local-first, agent-native control plane for ComfyUI — MCP server + autonomous sidebar agent that drives your live graph in natural language on ANY LLM: Claude/ChatGPT/Gemini on your subscription (no API key), free local models via Ollama (fully offline),

117 lines • 5.94 kB
import { z } from "zod"; import { getTracker } from "../services/generation-tracker.js"; import { errorToToolResult } from "../utils/errors.js"; export function registerGenerationTrackerTools(server) { server.tool("suggest_settings", "Recommend concrete, proven sampler/scheduler/steps/CFG (and denoise/shift/LoRA) settings derived from THIS MCP server's local generation-history database (populated as you run workflows; not from ComfyUI). Read-only and works without a running ComfyUI. Narrow results by model_family, lora_hash, or a name search; with no filter it returns the top settings across all history. Returns a ranked list with each combo's reuse count, or a 'no history' message until you have generated images. Use this for ready-to-apply values; use generation_stats for aggregate counts and breakdowns rather than specific suggestions.", { model_family: z .string() .optional() .describe("Model family to query (e.g. 'qwen_image', 'sdxl', 'flux', 'illustrious')"), lora_hash: z .string() .optional() .describe("AutoV2 hash (10 chars) of a specific LoRA to find settings for"), search: z .string() .optional() .describe("Full-text search on model/LoRA filenames (e.g. 'copax', 'lightning')"), limit: z .number() .int() .min(1) .max(50) .optional() .describe("Max results (default 10)"), }, async (args) => { try { const tracker = getTracker(); const limit = args.limit ?? 10; let results; let heading; if (args.lora_hash) { results = tracker.suggestSettingsForLora(args.lora_hash, limit); heading = `Top settings for LoRA ${args.lora_hash}`; } else if (args.search) { results = tracker.searchByName(args.search, limit); heading = `Settings matching "${args.search}"`; } else if (args.model_family) { results = tracker.suggestSettings(args.model_family, limit); heading = `Top settings for ${args.model_family}`; } else { // No filter — show overall top settings const stats = tracker.getStats(); results = stats.topSettings; heading = "Top settings across all models"; } if (results.length === 0) { return { content: [ { type: "text", text: "No generation history found. Run some workflows first to build up settings data.", }, ], }; } const lines = [`## ${heading}\n`]; for (const [i, s] of results.entries()) { const lora = s.loraName ? `LoRA: ${s.loraName} (${s.loraStrength})` : "no LoRA"; lines.push(`${i + 1}. **${s.sampler}/${s.scheduler}** — ${s.steps} steps, CFG ${s.cfg}` + (s.shift != null ? `, shift ${s.shift}` : "") + `, denoise ${s.denoise}` + `\n Model: ${s.modelName ?? s.modelHash} (${s.modelFamily})` + `\n ${lora}` + `\n Used ${s.reuseCount}x` + (s.presetName ? ` | Preset: ${s.presetName}` : "")); } return { content: [{ type: "text", text: lines.join("\n") }] }; } catch (err) { return errorToToolResult(err); } }); server.tool("generation_stats", "Show statistics from this MCP server's local generation-history database (populated as you run workflows; not from ComfyUI itself): total generations, count of unique sampler/scheduler/steps/CFG combos, a per-model-family breakdown, and the most-reused settings. Read-only; works without a running ComfyUI. Returns empty stats until you have generated images. For concrete recommended settings rather than aggregate counts, use suggest_settings.", { model_family: z .string() .optional() .describe("Optional model-family key to scope stats to, e.g. 'sdxl', 'flux', 'qwen_image', 'illustrious'"), }, async (args) => { try { const tracker = getTracker(); const stats = tracker.getStats(args.model_family); const lines = []; if (args.model_family) { lines.push(`## Generation Stats: ${args.model_family}\n`); } else { lines.push("## Generation Stats\n"); } lines.push(`- **Total generations**: ${stats.totalGenerations}`); lines.push(`- **Unique setting combos**: ${stats.uniqueCombos}`); if (stats.modelBreakdown.length > 0) { lines.push("\n### By model family\n"); for (const b of stats.modelBreakdown) { lines.push(`- ${b.modelFamily}: ${b.count} generations`); } } if (stats.topSettings.length > 0) { lines.push("\n### Most-used settings\n"); for (const [i, s] of stats.topSettings.entries()) { const lora = s.loraName ? ` + ${s.loraName}` : ""; lines.push(`${i + 1}. ${s.modelFamily} / ${s.sampler}/${s.scheduler} / ` + `${s.steps} steps / CFG ${s.cfg}${lora} — ${s.reuseCount}x`); } } return { content: [{ type: "text", text: lines.join("\n") }] }; } catch (err) { return errorToToolResult(err); } }); } //# sourceMappingURL=generation-tracker.js.map