@xelth/eck-snapshot 6.5.1 → 6.7.0

package/README.md

@@ -1,4 +1,4 @@
-# 📸 eckSnapshot v6.5.0 (AI-Native Edition)
+# 📸 eckSnapshot v6.7.0 (AI-Native Edition)
 
 A specialized, AI-native CLI tool that creates single-file text snapshots of entire Git repositories and feeds them directly into LLM context windows. Instead of letting AI agents guess which files to read, eckSnapshot force-feeds the complete project into the model's context — giving it a "university degree" in your codebase from the very first prompt.
 
@@ -200,6 +200,7 @@ By analyzing Claude Code's internal architecture, eckSnapshot replaces the old m
 * **📁 The `.eck/` Manifest:** Automatically maintains project context files (`CONTEXT.md`, `ROADMAP.md`, `TECH_DEBT.md`). Dynamic scanning — any `.md` file you add to `.eck/` is automatically included in snapshots.
 * **☠️ Skeleton Mode:** Uses Tree-sitter and Babel to strip function bodies, drastically reducing token count for huge codebases.
 * **📚 NotebookLM Export:** Semantic chunking for Google's NotebookLM with "Brain + Body" architecture (see below).
+* **🧪 ML Model Compatibility:** Smart metadata extraction for `.safetensors`, `.onnx`, `.pt`, `.pth`, `.h5`, `.pb`, `.bin`, `.ckpt`, `.gguf` — reads the first 4KB header instead of loading multi-GB weights into memory.
 * **🧠 Multi-Agent Protocol:** Junior Architect delegation system for multi-agent coding workflows (see below).
 
 ### 🤖 Autonomous AI Protocols

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xelth/eck-snapshot",
-  "version": "6.5.1",
+  "version": "6.7.0",
   "description": "A powerful CLI tool to create and restore single-file text snapshots of Git repositories. Optimized for AI context, LLM workflows, and multi-agent Swarm coordination.",
   "keywords": [
     "ai",

package/setup.json

@@ -164,6 +164,19 @@
         ],
         "priority": 7
       },
+      "esp-idf": {
+        "files": [
+          "sdkconfig.defaults",
+          "partitions.csv"
+        ],
+        "directories": [
+          "managed_components"
+        ],
+        "manifestFiles": [
+          "idf_component.yml"
+        ],
+        "priority": 8
+      },
       "c": {
         "files": [
           "Makefile",
@@ -518,6 +531,27 @@
           "packages/"
         ]
       },
+      "esp-idf": {
+        "filesToIgnore": [
+          "sdkconfig.old",
+          "dependencies.lock"
+        ],
+        "dirsToIgnore": [
+          "managed_components/",
+          "build/",
+          "managed_components/*/examples/",
+          "managed_components/*/test/",
+          "managed_components/*/test_apps/",
+          "managed_components/*/docs/",
+          "managed_components/*/hw/"
+        ],
+        "extensionsToIgnore": [
+          ".bin",
+          ".elf",
+          ".map",
+          ".flash"
+        ]
+      },
       "c": {
         "filesToIgnore": [
           "*.o",

package/src/cli/cli.js

@@ -6,6 +6,7 @@ import chalk from 'chalk';
 import { createRequire } from 'module';
 import os from 'os';
 import crypto from 'crypto';
+import { ensureSnapshotsInGitignore } from '../utils/fileUtils.js';
 
 const __filename = fileURLToPath(import.meta.url);
 const __dirname = path.dirname(__filename);
@@ -48,17 +49,17 @@ const LEGACY_COMMANDS = {
     const base = baseIdx !== -1 && args[baseIdx + 1] ? args[baseIdx + 1] : undefined;
     return { name: 'eck_update_auto', arguments: { fail: args.includes('--fail') || args.includes('-f'), base } };
   },
-  'snapshot':    () => ({ name: 'eck_snapshot', arguments: {} }),
+  'snapshot':    (args) => ({ name: 'eck_snapshot', arguments: { ml: args.includes('--ml') } }),
   'update':      (args) => {
     const baseIdx = args.indexOf('--base');
     const base = baseIdx !== -1 && args[baseIdx + 1] ? args[baseIdx + 1] : undefined;
-    return { name: 'eck_update', arguments: { fail: args.includes('--fail') || args.includes('-f'), base } };
+    return { name: 'eck_update', arguments: { fail: args.includes('--fail') || args.includes('-f'), base, ml: args.includes('--ml') } };
   },
   'setup-mcp':   (args) => ({ name: 'eck_setup_mcp', arguments: { opencode: args.includes('--opencode'), both: args.includes('--both') } }),
   'detect':      () => ({ name: 'eck_detect', arguments: {} }),
   'doctor':      () => ({ name: 'eck_doctor', arguments: {} }),
-  'scout':       (args) => ({ name: 'eck_scout', arguments: { depth: args[0] !== undefined ? parseInt(args[0], 10) : 0 } }),
-  'fetch':       (args) => ({ name: 'eck_fetch', arguments: { patterns: args } }),
+  'scout':       (args) => ({ name: 'eck_scout', arguments: { depth: args[0] !== undefined ? parseInt(args[0], 10) : 0, ml: args.includes('--ml') } }),
+  'fetch':       (args) => ({ name: 'eck_fetch', arguments: { patterns: args.filter(a => a !== '--ml'), ml: args.includes('--ml') } }),
   'link':        (args) => ({ name: 'eck_snapshot', arguments: { isLinkedProject: true, linkDepth: args[0] !== undefined ? parseInt(args[0], 10) : 0 } }),
   'profile':     (args) => args[0] ? ({ name: 'eck_snapshot', arguments: { profile: args.join(',') } }) : ({ name: 'eck_snapshot', arguments: { profile: true } }),
   'booklm':      () => ({ name: 'eck_snapshot', arguments: { notebooklm: 'scout' } }),
@@ -163,6 +164,7 @@ Ranked by frequency of use:
         queue.feedback.push({ type, message: msg, date: new Date().toISOString() });
 
         await fs.mkdir(path.dirname(queuePath), { recursive: true }).catch(() => {});
+        await ensureSnapshotsInGitignore(process.cwd()).catch(() => {});
         await fs.writeFile(queuePath, JSON.stringify(queue, null, 2));
 
         console.log(chalk.green('Feedback saved locally. It will be sent to developers during the next telemetry sync.'));
@@ -218,6 +220,7 @@ Ranked by frequency of use:
             } catch(e) { /* no existing queue */ }
             queue.usage[toolName] = (queue.usage[toolName] || 0) + 1;
             await fs.mkdir(path.dirname(queuePath), { recursive: true }).catch(() => {});
+            await ensureSnapshotsInGitignore(cwd).catch(() => {});
             await fs.writeFile(queuePath, JSON.stringify(queue, null, 2));
           } catch(e) { /* ignore tracking errors */ }
         }
@@ -265,6 +268,7 @@ Ranked by frequency of use:
             } catch(e) { /* no existing queue */ }
             queue.errors.push({ tool: toolName, error: err.message, date: new Date().toISOString() });
             await fs.mkdir(path.dirname(queuePath), { recursive: true }).catch(() => {});
+            await ensureSnapshotsInGitignore(cwd).catch(() => {});
             await fs.writeFile(queuePath, JSON.stringify(queue, null, 2));
           } catch(e) { /* ignore tracking errors */ }
         }

package/src/cli/commands/createSnapshot.js

@@ -3,7 +3,6 @@ import path from 'path';
 import { execa } from 'execa';
 import pLimit from 'p-limit';
 import { SingleBar, Presets } from 'cli-progress';
-import isBinaryPath from 'is-binary-path';
 import zlib from 'zlib';
 import { promisify } from 'util';
 import ora from 'ora';
@@ -15,7 +14,8 @@ import {
   scanDirectoryRecursively, loadGitignore, readFileWithSizeCheck,
   generateDirectoryTree, loadConfig, displayProjectInfo, loadProjectEckManifest,
   ensureSnapshotsInGitignore, initializeEckManifest, generateTimestamp,
-  getShortRepoName, SecretScanner, getProjectFiles
+  getShortRepoName, SecretScanner, getProjectFiles, readMlModelMetadata,
+  isBinaryFile
 } from '../../utils/fileUtils.js';
 import { detectProjectType, getProjectSpecificFiltering, getAllDetectedTypes } from '../../utils/projectDetector.js';
 import { estimateTokensWithPolynomial, generateTrainingCommand } from '../../utils/tokenEstimator.js';
@@ -252,7 +252,7 @@ async function estimateProjectTokens(projectPath, config, projectTypes = null) {
         continue;
       }
 
-      if (isBinaryPath(file)) {
+      if (await isBinaryFile(path.join(projectPath, file))) {
         continue;
       }
 
@@ -394,8 +394,16 @@ async function processProjectFiles(repoPath, options, config, projectTypes = nul
           return null;
         }
 
-        // Check if binary file
-        if (isBinaryPath(filePath)) {
+        const mlExt = path.extname(filePath).toLowerCase();
+        const ML_EXTENSIONS = ['.safetensors', '.onnx', '.pt', '.pth', '.h5', '.pb', '.bin', '.ckpt', '.gguf'];
+        // ML peek is opt-in via `arguments.ml: true`. Default: ML files treated as plain binaries.
+        // This prevents false-positives where `.bin` raw dumps (mitm captures, sniffer output)
+        // get included via readMlModelMetadata when no real model is present.
+        const isMlModel = !!options?.ml && ML_EXTENSIONS.includes(mlExt);
+
+        // Content-aware binary check (catches extensionless ELFs, SQLite DBs, archives).
+        // ML models bypass to allow header metadata extraction below.
+        if (!isMlModel && await isBinaryFile(path.join(repoPath, filePath))) {
           stats.binaryFiles++;
           trackSkippedFile(normalizedPath, 'Binary files');
           return null;
@@ -421,13 +429,18 @@ async function processProjectFiles(repoPath, options, config, projectTypes = nul
         stats.totalSize += fileStats.size;
 
         const maxFileSize = parseSize(config.maxFileSize);
-        if (fileStats.size > maxFileSize) {
-          stats.oversizedFiles++;
-          trackSkippedFile(normalizedPath, `File too large (${formatSize(fileStats.size)} > ${formatSize(maxFileSize)})`);
-          return null;
-        }
+        let content;
 
-        let content = await readFileWithSizeCheck(fullPath, maxFileSize);
+        if (isMlModel) {
+          content = await readMlModelMetadata(fullPath);
+        } else {
+          if (fileStats.size > maxFileSize) {
+            stats.oversizedFiles++;
+            trackSkippedFile(normalizedPath, `File too large (${formatSize(fileStats.size)} > ${formatSize(maxFileSize)})`);
+            return null;
+          }
+          content = await readFileWithSizeCheck(fullPath, maxFileSize);
+        }
 
         // Security scan for secrets
         if (config.security?.scanForSecrets !== false) {