ccsniff 1.0.29 → 1.0.31

package/README.md

@@ -40,8 +40,33 @@ npx ccsniff --since 24h --grep "rs-exec" --limit 50
 npx ccsniff --since 7d --role user --json
 npx ccsniff -f                     # tail new events live
 npx ccsniff --rollup out.ndjson --since 7d
+npx ccsniff --unsloth train.jsonl --since 7d --no-subagents
+npx ccsniff --unsloth train.jsonl --unsloth-format sharegpt --since 7d
 ```
 
+### Unsloth training export
+
+`--unsloth <out>` writes one JSONL line per Claude Code session, ready for
+Unsloth / TRL conversational fine-tuning. All filter flags (`--since`,
+`--project`, `--session`, `--no-subagents`, ...) apply.
+
+Two formats are supported via `--unsloth-format`:
+
+- `messages` (default) — OpenAI / ChatML shape with native tool calling:
+  ```json
+  {"session_id":"...","messages":[
+    {"role":"user","content":"find foobar"},
+    {"role":"assistant","content":null,"tool_calls":[{"id":"tu1","type":"function","function":{"name":"Grep","arguments":"{\"pattern\":\"foobar\"}"}}]},
+    {"role":"tool","tool_call_id":"tu1","content":"hit at line 3"},
+    {"role":"assistant","content":"done"}
+  ]}
+  ```
+- `sharegpt` — `{conversations:[{from:human|gpt|tool, value}]}`, compatible
+  with `standardize_sharegpt`. Tool calls are inlined into the `gpt` turn as
+  `<tool_call>name(json-args)</tool_call>`.
+
+Sessions with no user/assistant turn pair are skipped (no training value).
+
 ## API
 
 ### `watch(projectsDir?)` → `JsonlWatcher`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ccsniff",
-  "version": "1.0.29",
+  "version": "1.0.31",
   "description": "Watch Claude Code JSONL output files and emit structured events as a Node.js EventEmitter",
   "type": "module",
   "main": "./src/index.js",

package/src/cli.js

@@ -1,5 +1,7 @@
 #!/usr/bin/env node
 import { JsonlReplayer, rollup, vault } from './index.js';
+import { toUnslothMessages, toShareGPT } from './unsloth.js';
+import fs from 'fs';
 import path from 'path';
 
 if (process.argv[2] === 'gui') {
@@ -25,7 +27,7 @@ if (process.argv[2] === 'gui') {
 { const r = vault(); if (r.copied > 0) process.stderr.write(`# vault: ${r.copied} copied → ~/.claude/history-backup\n`); }
 
 const FLAGS = {
-  string: ['since', 'until', 'before', 'after', 'grep', 'igrep', 'cwd', 'project', 'role', 'type', 'tool', 'session', 'sid', 'parent', 'rollup', 'format', 'sort'],
+  string: ['since', 'until', 'before', 'after', 'grep', 'igrep', 'cwd', 'project', 'role', 'type', 'tool', 'session', 'sid', 'parent', 'rollup', 'format', 'sort', 'unsloth', 'unsloth-format'],
   multi: ['grep', 'igrep', 'role', 'type', 'tool', 'session', 'sid', 'project', 'cwd'],
   number: ['limit', 'head', 'tail-n', 'ctx', 'truncate'],
   bool: ['json', 'ndjson', 'tail', 'f', 'full', 'reverse', 'invert', 'no-subagents', 'only-subagents', 'no-meta', 'only-meta', 'list-sessions', 'list-projects', 'list-tools', 'stats', 'count', 'help', 'h'],
@@ -99,6 +101,8 @@ OUTPUT
   -f, --tail             live tail after replay
   --rollup <out>         dump filtered events to file
   --format ndjson|sqlite rollup format (default ndjson; sqlite needs better-sqlite3)
+  --unsloth <out>        write Unsloth training JSONL (one conversation per session per line)
+  --unsloth-format <fmt> messages (OpenAI/ChatML, default) | sharegpt
 
 EXAMPLES
   ccsniff --since 24h --grep "rs-exec" --limit 50
@@ -360,6 +364,15 @@ if (opts.count) {
   process.exit(0);
 }
 
+if (opts.unsloth) {
+  const fmt = opts['unsloth-format'] || 'messages';
+  const recs = fmt === 'sharegpt' ? toShareGPT(rows) : toUnslothMessages(rows);
+  const body = recs.map(r => JSON.stringify(r)).join('\n') + (recs.length ? '\n' : '');
+  fs.writeFileSync(opts.unsloth, body);
+  process.stderr.write(`# unsloth(${fmt}): ${recs.length} conversations → ${opts.unsloth}\n`);
+  process.exit(0);
+}
+
 for (const ev of rows) process.stdout.write(formatRow(ev, opts));
 process.stderr.write(`# ${stats.events} events / ${stats.files} files / ${rows.length} matched\n`);
 

package/src/unsloth.js

@@ -0,0 +1,109 @@
+function textOf(b) {
+  if (typeof b.text === 'string') return b.text;
+  if (typeof b.content === 'string') return b.content;
+  if (Array.isArray(b.content)) return b.content.map(c => c?.text || '').join('');
+  return '';
+}
+
+function groupBySession(events) {
+  const m = new Map();
+  for (const ev of events) {
+    const sid = ev.conversation?.id || 'unknown';
+    if (!m.has(sid)) m.set(sid, []);
+    m.get(sid).push(ev);
+  }
+  for (const arr of m.values()) arr.sort((a, b) => (a.timestamp || 0) - (b.timestamp || 0));
+  return m;
+}
+
+function buildMessagesForSession(evs) {
+  const messages = [];
+  let cur = null;
+  const flush = () => { if (cur && (cur.content || (cur.tool_calls && cur.tool_calls.length))) messages.push(cur); cur = null; };
+  for (const ev of evs) {
+    const b = ev.block || {};
+    const t = b.type;
+    if (ev.role === 'user') {
+      if (t === 'tool_result') {
+        flush();
+        const txt = textOf(b);
+        messages.push({ role: 'tool', tool_call_id: b.tool_use_id || '', content: txt });
+        continue;
+      }
+      if (t === 'text' && !b.isMeta) {
+        const txt = textOf(b);
+        if (!txt.trim()) continue;
+        if (cur && cur.role === 'user') cur.content += '\n' + txt;
+        else { flush(); cur = { role: 'user', content: txt }; }
+      }
+      continue;
+    }
+    if (ev.role === 'assistant') {
+      if (!cur || cur.role !== 'assistant') { flush(); cur = { role: 'assistant', content: '' }; }
+      if (t === 'text') {
+        const txt = textOf(b);
+        if (txt) cur.content = cur.content ? cur.content + '\n' + txt : txt;
+      } else if (t === 'thinking') {
+        continue;
+      } else if (t === 'tool_use') {
+        if (!cur.tool_calls) cur.tool_calls = [];
+        cur.tool_calls.push({
+          id: b.id || '',
+          type: 'function',
+          function: { name: b.name || '', arguments: JSON.stringify(b.input || {}) },
+        });
+      }
+      continue;
+    }
+  }
+  flush();
+  for (const m of messages) {
+    if (m.role === 'assistant' && m.tool_calls && !m.content) m.content = null;
+  }
+  return messages;
+}
+
+function hasTrainingValue(messages) {
+  let hasUser = false, hasAsst = false;
+  for (const m of messages) {
+    if (m.role === 'user') hasUser = true;
+    if (m.role === 'assistant') hasAsst = true;
+  }
+  return hasUser && hasAsst;
+}
+
+export function toUnslothMessages(events) {
+  const sessions = groupBySession(events);
+  const out = [];
+  for (const [sid, evs] of sessions) {
+    const messages = buildMessagesForSession(evs);
+    if (!hasTrainingValue(messages)) continue;
+    out.push({ session_id: sid, messages });
+  }
+  return out;
+}
+
+export function toShareGPT(events) {
+  const sessions = groupBySession(events);
+  const out = [];
+  for (const [sid, evs] of sessions) {
+    const messages = buildMessagesForSession(evs);
+    if (!hasTrainingValue(messages)) continue;
+    const conversations = [];
+    for (const m of messages) {
+      if (m.role === 'user') conversations.push({ from: 'human', value: m.content });
+      else if (m.role === 'assistant') {
+        let v = m.content || '';
+        if (m.tool_calls && m.tool_calls.length) {
+          const calls = m.tool_calls.map(c => `<tool_call>${c.function.name}(${c.function.arguments})</tool_call>`).join('\n');
+          v = v ? v + '\n' + calls : calls;
+        }
+        conversations.push({ from: 'gpt', value: v });
+      } else if (m.role === 'tool') {
+        conversations.push({ from: 'tool', value: m.content });
+      }
+    }
+    out.push({ session_id: sid, conversations });
+  }
+  return out;
+}