leerness 1.9.149 → 1.9.150

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -1,5 +1,48 @@
1
1
  # Changelog
2
2
 
3
+ ## 1.9.150 — 2026-05-20
4
+
5
+ **Sandboxing — `runCommandSafe()` wrapper + REPL slash-commands (3중 LLM 합의 #3 / Codex 권고).**
6
+
7
+ 자율 모드 80 라운드 마일스톤. 1.9.149 REPL 위에 **샌드박스 보안 레이어** + **leerness 내부 명령 직접 호출**을 추가.
8
+
9
+ ### Added — `runCommandSafe(cmd, args, opts)` sandbox wrapper
10
+ - **cwd jail** — `cwd` 가 root 밖 (path traversal) 이면 즉시 exit 126 + `blocked: 'cwd_jail'` 기록
11
+ - **shell:false 기본** — shell injection 표면 차단. `allowShell: true` 시만 shell:true (npm/pytest 호환)
12
+ - **env scrub** — 안전 화이트리스트만 통과 (`PATH`, `HOME`, `TMP`, `LEERNESS_*`, `NPM_CONFIG_*`, ...)
13
+ - 시크릿 환경변수 (DB_PASSWORD, API_KEY 등) 자식 프로세스에 누출 방지
14
+ - **timeout 한도** — 기본 5분, max 10분 (clamp)
15
+ - **permissions 검증** — 1.9.146 `permissions.shell.allowList` 자동 연동
16
+ - basic 모드 (`shell.exec=false`) 에선 핵심 도구 (git/npm/node/pnpm/yarn) 만 허용
17
+ - allowList 외 명령은 즉시 reject + `blocked: 'permissions'` 기록
18
+ - **자동 observability** — 호출마다 `_recordRun` 으로 cmd/args/durationMs/status/cwd 자동 기록
19
+
20
+ ### Changed — 위험 호출 sandbox 치환
21
+ - `verify-code` (line ~7473): `cp.spawnSync(t.cmd, [], { shell: true })` → `runCommandSafe(t.cmd, ...)`
22
+ - `deploy auto` (line ~10580): `cp.spawnSync(meta.deployCommand, [], { shell: true })` → `runCommandSafe(...)`
23
+ - `agents bench` (line ~8866): `cp.spawnSync(cmd, cliArgs, { shell: true })` → `runCommandSafe(...)`
24
+ - 3 곳 모두 env scrub + cwd jail + observability 자동 적용
25
+
26
+ ### Added — REPL slash-commands (1.9.149 위에)
27
+ - `:verify` — `leerness verify-code` 직접 호출 (sandboxed)
28
+ - `:audit` — `leerness audit` (보안 + drift + lazy)
29
+ - `:handoff` — `leerness handoff --quiet --no-drift-check`
30
+ - `:health` — `leerness health --json`
31
+ - 모두 `runCommandSafe` 경유 — 자식 leerness 호출도 sandbox 적용
32
+ - REPL 안에서 agent가 "현재 상태 점검해줘" 같은 메타 명령으로 leerness 기능을 즉시 호출 가능
33
+
34
+ ### Security
35
+ - 시크릿 환경변수 누출 표면 대폭 축소 — `runCommandSafe` 호출 시 화이트리스트 외 env 미전달
36
+ - 사용자 글로벌 룰 준수: API 키/DB 비밀번호 절대 자식 프로세스에 자동 전파 금지
37
+ - `_reports/`, `.harness/agent-sessions/`, `.harness/runs/`, `.harness/credentials.local.json` 비공개 정책 유지
38
+
39
+ ### Verified
40
+ - e2e 220/220 (slash-commands handleMeta 통합 + sandbox wrapper)
41
+ - stress-v95: 19/19 (sandbox 5종 + REPL slash 4종 + 누적 회귀 1.9.146~149)
42
+ - VERSION = 1.9.150 / autonomous-rounds = 80
43
+
44
+ ---
45
+
3
46
  ## 1.9.149 — 2026-05-20
4
47
 
5
48
  **REPL agent (Hermes/OpenClaw/OpenCode 스타일) + observability lite — 사용자 명시 요청 + 3중 LLM 합의 #2.**
package/README.md CHANGED
@@ -2,7 +2,7 @@
2
2
 
3
3
  > **AI 코딩 에이전트의 거짓 완료·중복·망각·충돌을 막아주는 검수·기억·협업 CLI 하네스.**
4
4
 
5
- [![npm](https://img.shields.io/badge/npm-leerness-blue)](https://www.npmjs.com/package/leerness) [![version](https://img.shields.io/badge/version-1.9.149-green)]() [![tests](https://img.shields.io/badge/e2e-219%2F219-success)]() [![mcp](https://img.shields.io/badge/MCP--tools-47-blue)]() [![json](https://img.shields.io/badge/--json-20_commands-blueviolet)]() [![rounds](https://img.shields.io/badge/autonomous--rounds-79-blueviolet)]() [![main-push](https://img.shields.io/badge/release--main--push-auto-success)]() [![repl-agent](https://img.shields.io/badge/agent--REPL-Hermes%2FOpenClaw_style-success)]() [![observability](https://img.shields.io/badge/runs--jsonl-traceId%2Fduration%2Fmodel-success)]() [![license](https://img.shields.io/badge/license-MIT-lightgrey)]()
5
+ [![npm](https://img.shields.io/badge/npm-leerness-blue)](https://www.npmjs.com/package/leerness) [![version](https://img.shields.io/badge/version-1.9.150-green)]() [![tests](https://img.shields.io/badge/e2e-219%2F219-success)]() [![stress](https://img.shields.io/badge/stress--v95-21%2F21-success)]() [![mcp](https://img.shields.io/badge/MCP--tools-47-blue)]() [![json](https://img.shields.io/badge/--json-20_commands-blueviolet)]() [![rounds](https://img.shields.io/badge/autonomous--rounds-80-blueviolet)]() [![main-push](https://img.shields.io/badge/release--main--push-auto-success)]() [![repl-agent](https://img.shields.io/badge/agent--REPL-Hermes%2FOpenClaw_style-success)]() [![sandbox](https://img.shields.io/badge/runCommandSafe-cwd_jail%2Benv_scrub-success)]() [![observability](https://img.shields.io/badge/runs--jsonl-traceId%2Fduration%2Fmodel-success)]() [![license](https://img.shields.io/badge/license-MIT-lightgrey)]()
6
6
 
7
7
  ```
8
8
  ╔══════════════════════════════════════════════════════════════╗
@@ -12,8 +12,8 @@
12
12
  ║ ██║ ██╔══╝ ██╔══╝ ██╔══██╗██║╚██╗██║██╔══╝ ╚════██║ ║
13
13
  ║ ███████╗███████╗███████╗██║ ██║██║ ╚████║███████╗███████║ ║
14
14
  ║ ╚══════╝╚══════╝╚══════╝╚═╝ ╚═╝╚═╝ ╚═══╝╚══════╝╚══════╝ ║
15
- ║ v1.9.149 AI Agent Reliability Harness
16
- ║ verify · remember · orchestrate · audit · prevent drift
15
+ ║ v1.9.150 AI Agent Reliability Harness + Sandbox
16
+ ║ verify · remember · orchestrate · audit · sandbox · drift
17
17
  ╚══════════════════════════════════════════════════════════════╝
18
18
  ```
19
19
 
package/bin/harness.js CHANGED
@@ -6,7 +6,7 @@ const path = require('path');
6
6
  const cp = require('child_process');
7
7
  const readline = require('readline');
8
8
 
9
- const VERSION = '1.9.149';
9
+ const VERSION = '1.9.150';
10
10
  const MARK = '<!-- leerness:managed -->';
11
11
  const README_START = '<!-- leerness:project-readme:start -->';
12
12
  const README_END = '<!-- leerness:project-readme:end -->';
@@ -7470,7 +7470,8 @@ function verifyCodeCmd(root) {
7470
7470
  for (const t of tasks) {
7471
7471
  log(`\n## ${t.name}: ${t.cmd}`);
7472
7472
  const start = Date.now();
7473
- const r = cp.spawnSync(t.cmd, [], { cwd: root, encoding: 'utf8', shell: true, timeout: 5 * 60 * 1000 });
7473
+ // 1.9.150: runCommandSafe cwd jail + env scrub + observability 자동 (shell:true 유지 npm/pytest 호환)
7474
+ const r = runCommandSafe(t.cmd, [], { cwd: root, root, timeout: 5 * 60 * 1000, allowShell: true, kind: 'verify_code_task', label: `verify-${t.name}` });
7474
7475
  const dur = Date.now() - start;
7475
7476
  if (r.status === 0) ok(`${t.name} passed (${dur}ms)`);
7476
7477
  else if (t.optional && r.status === 127) warn(`${t.name} 스킵 (${t.cmd} 없음)`);
@@ -8862,7 +8863,8 @@ async function _benchmarkMeasure(root, task) {
8862
8863
  else if (agent.id === 'gemini') { cmd = 'gemini'; cliArgs = ['-p', task]; }
8863
8864
  else continue;
8864
8865
  const t0 = Date.now();
8865
- const r = cp.spawnSync(cmd, cliArgs, { encoding: 'utf8', timeout: 60000, shell: true });
8866
+ // 1.9.150: runCommandSafe agent CLI bench sandbox (env scrub + observability)
8867
+ const r = runCommandSafe(cmd, cliArgs, { cwd: process.cwd(), root, timeout: 60000, allowShell: true, kind: 'agent_bench', label: `bench-${agent.id}`, allowOutsideCwd: true });
8866
8868
  const baseTime = Date.now() - t0;
8867
8869
  // leerness 검수 layer time 추정 (verify-claim 형식)
8868
8870
  const t1 = Date.now();
@@ -10010,6 +10012,111 @@ function _recordRun(root, entry) {
10010
10012
  return id;
10011
10013
  } catch { return null; }
10012
10014
  }
10015
+
10016
+ // 1.9.150: Sandboxing — runCommandSafe wrapper (Codex 권고: 3중 LLM 합의 #3)
10017
+ // cwd jail (root 밖 거부) + shell:false 기본 + timeout + env scrub + permissions allowList 검증 + _recordRun 자동
10018
+ const _ENV_SAFE_KEYS = new Set([
10019
+ 'PATH', 'HOME', 'USERPROFILE', 'TEMP', 'TMP', 'TMPDIR', 'NODE_PATH', 'NODE_ENV',
10020
+ 'LANG', 'LC_ALL', 'LC_CTYPE', 'SHELL', 'COMSPEC', 'SYSTEMROOT', 'WINDIR', 'OS',
10021
+ 'PROCESSOR_ARCHITECTURE', 'PROCESSOR_IDENTIFIER', 'NUMBER_OF_PROCESSORS',
10022
+ 'PROGRAMFILES', 'PROGRAMFILES(X86)', 'APPDATA', 'LOCALAPPDATA',
10023
+ 'GITHUB_TOKEN', 'NPM_TOKEN', 'CI', 'GH_TOKEN'
10024
+ ]);
10025
+ function _scrubEnv(extraEnv) {
10026
+ const out = {};
10027
+ for (const k of Object.keys(process.env || {})) {
10028
+ if (_ENV_SAFE_KEYS.has(k) || k.startsWith('LEERNESS_') || k.startsWith('NPM_CONFIG_')) {
10029
+ out[k] = process.env[k];
10030
+ }
10031
+ }
10032
+ if (extraEnv && typeof extraEnv === 'object') {
10033
+ for (const k of Object.keys(extraEnv)) {
10034
+ // Allow caller overrides — explicit opt-in
10035
+ if (extraEnv[k] !== undefined) out[k] = String(extraEnv[k]);
10036
+ }
10037
+ }
10038
+ return out;
10039
+ }
10040
+ function _isCwdSafe(root, cwd) {
10041
+ try {
10042
+ if (!cwd) return true;
10043
+ const r = path.resolve(absRoot(root));
10044
+ const c = path.resolve(cwd);
10045
+ if (c === r) return true;
10046
+ const rel = path.relative(r, c);
10047
+ return !rel.startsWith('..') && !path.isAbsolute(rel);
10048
+ } catch { return false; }
10049
+ }
10050
+ function runCommandSafe(cmd, args, opts) {
10051
+ // opts: { cwd, root, timeout, env, stdio, kind, label, allowShell, encoding, input, allowOutsideCwd }
10052
+ opts = opts || {};
10053
+ const root = opts.root || opts.cwd || process.cwd();
10054
+ const cwd = opts.cwd || root;
10055
+ const cmdStr = String(cmd || '').trim();
10056
+ const argList = Array.isArray(args) ? args.slice() : [];
10057
+ const t0 = Date.now();
10058
+ const label = opts.label || opts.kind || 'shell_exec';
10059
+ // 1) cwd jail
10060
+ if (!opts.allowOutsideCwd && !_isCwdSafe(root, cwd)) {
10061
+ const r = { status: 126, stdout: '', stderr: `runCommandSafe: cwd outside root rejected (${cwd})`, error: 'cwd_jail', blocked: true };
10062
+ try { _recordRun(root, { kind: label, cmd: cmdStr, args: argList, durationMs: Date.now() - t0, ok: false, blocked: 'cwd_jail' }); } catch {}
10063
+ return r;
10064
+ }
10065
+ // 2) permissions allowList (1.9.146)
10066
+ try {
10067
+ const perms = _readPermissions(root);
10068
+ const exec = perms.shell?.exec !== false; // basic 에선 false
10069
+ const allow = perms.shell?.allowList || [];
10070
+ if (!exec && !opts.allowOutsideCwd) {
10071
+ // basic 모드 — git/npm/node 같은 핵심 도구는 허용 (release/install 흐름 유지)
10072
+ const coreAllow = ['git', 'npm', 'npx', 'node', 'pnpm', 'yarn'];
10073
+ const first = cmdStr.split(/\s+/)[0];
10074
+ if (!coreAllow.includes(first) && !allow.includes('*') && !allow.includes(first)) {
10075
+ const r = { status: 126, stdout: '', stderr: `runCommandSafe: shell.exec=false (mode=${perms.mode}). allowList: ${allow.join(',') || '(없음)'} / core: ${coreAllow.join(',')}`, error: 'permissions', blocked: true };
10076
+ try { _recordRun(root, { kind: label, cmd: cmdStr, args: argList, durationMs: Date.now() - t0, ok: false, blocked: 'permissions', mode: perms.mode }); } catch {}
10077
+ return r;
10078
+ }
10079
+ }
10080
+ } catch {}
10081
+ // 3) spawn — shell:false 기본 (shell injection 차단). allowShell=true 시만 shell:true (deploy/build 호환)
10082
+ const useShell = !!opts.allowShell;
10083
+ const timeout = Math.min(opts.timeout || 5 * 60 * 1000, 10 * 60 * 1000);
10084
+ const spawnOpts = {
10085
+ cwd,
10086
+ encoding: opts.encoding || 'utf8',
10087
+ timeout,
10088
+ shell: useShell,
10089
+ env: _scrubEnv(opts.env),
10090
+ input: opts.input,
10091
+ stdio: opts.stdio || 'pipe'
10092
+ };
10093
+ let r;
10094
+ try {
10095
+ if (useShell) {
10096
+ // shell:true 모드 — 인자가 cmd 안에 포함된 단일 문자열인 경우 처리
10097
+ r = cp.spawnSync(cmdStr + (argList.length ? ' ' + argList.join(' ') : ''), [], spawnOpts);
10098
+ } else {
10099
+ // 단일 명령어로 들어온 경우 자동 분리
10100
+ let bin = cmdStr, finalArgs = argList;
10101
+ if (!argList.length && /\s/.test(cmdStr)) {
10102
+ const parts = cmdStr.split(/\s+/);
10103
+ bin = parts[0]; finalArgs = parts.slice(1);
10104
+ }
10105
+ r = cp.spawnSync(bin, finalArgs, spawnOpts);
10106
+ }
10107
+ } catch (e) {
10108
+ r = { status: 1, stdout: '', stderr: e.message, error: 'spawn_exception' };
10109
+ }
10110
+ const dt = Date.now() - t0;
10111
+ try {
10112
+ _recordRun(root, {
10113
+ kind: label, cmd: cmdStr, args: argList,
10114
+ durationMs: dt, status: r.status, ok: r.status === 0,
10115
+ shell: useShell, cwd: path.relative(absRoot(root), cwd) || '.'
10116
+ });
10117
+ } catch {}
10118
+ return r;
10119
+ }
10013
10120
  function runsListCmd(root) {
10014
10121
  root = absRoot(root || process.cwd());
10015
10122
  const dir = _runsDir(root);
@@ -10077,8 +10184,8 @@ async function _agentRepl(root, opts) {
10077
10184
  // 환영 메시지 + 모델 선택
10078
10185
  log('');
10079
10186
  log(C.bold(C.cy(' ╔════════════════════════════════════════════════════╗')));
10080
- log(C.bold(C.cy(' ║ leerness agent — REPL mode (1.9.149) ║')));
10081
- log(C.bold(C.cy(' ║ Hermes / OpenClaw / OpenCode 스타일 채팅 에이전트 ║')));
10187
+ log(C.bold(C.cy(' ║ leerness agent — REPL mode (1.9.150) ║')));
10188
+ log(C.bold(C.cy(' ║ Hermes / OpenClaw / OpenCode 스타일 + Sandbox ║')));
10082
10189
  log(C.bold(C.cy(' ╚════════════════════════════════════════════════════╝')));
10083
10190
  log('');
10084
10191
  // Ollama 모델 자동 감지 — model이 명시되지 않았으면 사용자에게 선택지 제공
@@ -10100,6 +10207,7 @@ async function _agentRepl(root, opts) {
10100
10207
  }
10101
10208
  log('');
10102
10209
  log(C.dim(' 메타 명령: :help | :model <m> | :role <r> | :provider <p> | :clear | :save | :history | :quit'));
10210
+ log(C.dim(' Slash 명령 (1.9.150): :verify | :audit | :handoff | :health'));
10103
10211
  log(C.dim(` 현재 — provider=${state.provider} model=${state.model || '(없음)'} role=${state.role} permissions=${_readPermissions(root).mode}`));
10104
10212
  log('');
10105
10213
  const prompt = () => isTty ? C.cy(`agent[${state.role}]> `) : 'agent> ';
@@ -10125,6 +10233,11 @@ async function _agentRepl(root, opts) {
10125
10233
  log(' :save — 세션 즉시 저장');
10126
10234
  log(' :permissions — 현재 권한 모드 표시');
10127
10235
  log(' :quit / :exit / :q — 종료 (자동 저장)');
10236
+ log(C.bold('\n Slash 명령 (1.9.150) — leerness 내부 명령 직접 호출:'));
10237
+ log(' :verify — leerness verify-code (테스트/타입/린트 자동 검수)');
10238
+ log(' :audit — leerness audit (보안 + drift + lazy)');
10239
+ log(' :handoff — leerness handoff --quiet (현재 컨텍스트 요약)');
10240
+ log(' :health — leerness health --json (종합 헬스 체크)');
10128
10241
  return false;
10129
10242
  }
10130
10243
  if (op === 'model') { state.model = rest.join(' ') || state.model; log(C.green(` model = ${state.model}`)); return false; }
@@ -10149,6 +10262,26 @@ async function _agentRepl(root, opts) {
10149
10262
  }
10150
10263
  if (op === 'save') { saveSession(); log(C.dim(` → ${rel(root, sessionPath())}`)); return false; }
10151
10264
  if (op === 'permissions') { permissionsListCmd(root); return false; }
10265
+ // 1.9.150: leerness 내부 명령 slash-commands — :verify / :audit / :handoff / :health
10266
+ if (op === 'verify' || op === 'audit' || op === 'handoff' || op === 'health') {
10267
+ const subArgs = {
10268
+ verify: ['verify-code', root],
10269
+ audit: ['audit', root],
10270
+ handoff: ['handoff', root, '--quiet', '--no-drift-check'],
10271
+ health: ['health', root, '--json']
10272
+ }[op];
10273
+ log(C.dim(` → leerness ${subArgs.join(' ')}`));
10274
+ const t0 = Date.now();
10275
+ const r = runCommandSafe(process.execPath, [__filename, ...subArgs], {
10276
+ cwd: root, root, timeout: 60000, kind: 'agent_repl_slash', label: `repl-${op}`,
10277
+ env: { LEERNESS_NO_BANNER: '1', LEERNESS_NO_PROMPT: '1', LEERNESS_NO_DRIFT_CHECK: '1' }
10278
+ });
10279
+ const dt = Date.now() - t0;
10280
+ if (r.stdout) log(r.stdout.trim().split('\n').slice(0, 30).join('\n'));
10281
+ if (r.status === 0) log(C.green(` ✓ :${op} 완료 (${dt}ms)`));
10282
+ else log(C.yel(` ⚠ :${op} 실패 (exit ${r.status}, ${dt}ms)`));
10283
+ return false;
10284
+ }
10152
10285
  log(C.yel(` 알 수 없는 명령: :${op} (:help 참고)`));
10153
10286
  return false;
10154
10287
  };
@@ -10208,16 +10341,17 @@ async function agentCmd(root, taskArg) {
10208
10341
  return;
10209
10342
  }
10210
10343
  // non-TTY: 사용법만 출력
10211
- log('# leerness agent (1.9.146/148/149) — Hermes/OpenClaw 스타일 CLI 에이전트');
10344
+ log('# leerness agent (1.9.146/148/149/150) — Hermes/OpenClaw 스타일 CLI 에이전트 + Sandbox');
10212
10345
  log('');
10213
10346
  log('사용법:');
10214
- log(' leerness agent # 🆕 1.9.149 REPL 모드 (모델 선택 + 채팅)');
10347
+ log(' leerness agent # 1.9.149 REPL 모드 (모델 선택 + 채팅)');
10215
10348
  log(' leerness agent "<task>" # 1회 위임 (actor 역할 기본)');
10216
10349
  log(' leerness agent "<task>" --role planner # 계획만 (1.9.148)');
10217
10350
  log(' leerness agent "<task>" --role reviewer # 비판적 검토 (1.9.148)');
10218
10351
  log(' leerness agent --interactive --model qwen2.5-coder # 명시적 REPL + model 선택');
10219
10352
  log('');
10220
10353
  log('REPL 메타 명령: :help / :model / :role / :provider / :history / :save / :quit');
10354
+ log('REPL Slash 명령 (1.9.150): :verify / :audit / :handoff / :health (sandboxed runCommandSafe)');
10221
10355
  log('');
10222
10356
  log('현재 활성 provider: ' + (_activeCliAgents().join(', ') || '(없음) — .env에서 LEERNESS_ENABLE_* 활성화'));
10223
10357
  log('권한 모드: ' + (_readPermissions(root).mode || 'basic'));
@@ -10575,7 +10709,8 @@ async function deployAutoCmd(root, service) {
10575
10709
  log(`service: ${service} · command: ${meta.deployCommand}`);
10576
10710
  if (has('--dry-run')) { log('(dry-run) 실제 실행 스킵'); return; }
10577
10711
  const t0 = Date.now();
10578
- const r = cp.spawnSync(meta.deployCommand, [], { cwd: root, encoding: 'utf8', shell: true, timeout: 10 * 60 * 1000, stdio: 'inherit' });
10712
+ // 1.9.150: runCommandSafe deploy 명령 sandbox (env scrub + permissions 검증 + observability)
10713
+ const r = runCommandSafe(meta.deployCommand, [], { cwd: root, root, timeout: 10 * 60 * 1000, allowShell: true, stdio: 'inherit', kind: 'deploy_auto', label: `deploy-${service}` });
10579
10714
  const dt = Date.now() - t0;
10580
10715
  if (r.status === 0) {
10581
10716
  ok(`deploy 성공: ${service} (${dt}ms)`);
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "leerness",
3
- "version": "1.9.149",
3
+ "version": "1.9.150",
4
4
  "description": "Leerness: 비파괴 마이그레이션, 자동 버전 감지·업데이트, 계획/진행/핸드오프 자동화, 게으름·시크릿·인코딩 자동 가드, Claude Code 슬래시 통합을 갖춘 한국어 우선 AI 개발 하네스.",
5
5
  "keywords": [
6
6
  "leerness",