Back to skill

Security audit

元忆 yotta-memory

Security checks for vulnerabilities and agentic risk

Overview

ClawScan labels this skill suspicious because it reports serious security weaknesses involving private memory exposure, plaintext network traffic, and unsafe handling of imported data; analysis was partial, so the evidence is incomplete.

Treat this as a review-required skill and weigh the reported risks before installing, especially if you would connect remotely or store private memories. The scan is partial, so the absence of other reported issues would not establish safety; the verdict is suspicious and does not by itself establish malicious intent.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
bin/yotta-memory.js
Finding

Unlocking the viewer authorizes every client of the running server

Content
View full analysis

Impact

Another reachable client can export decrypted private memories, retrieve the recovery key, and invoke privileged viewer operations. Recovery-key disclosure compromises the protection of owner keys available with the encrypted library.

T09 · Insecure Skill Coding Practices

Warning
Location
bin/yotta-memory.js
Finding

Imported public scope bypasses owner validation and permits directory traversal

Content
View full analysis

Impact

The importer creates directories and attacker-content plaintext Markdown files outside the library, within the process's filesystem permissions. For example, ../../outside traverses above the private directory before the prefs and date suffixes are appended. Marking the record public also bypasses the encrypted-private owner-key check. Output filenames and directory suffixes remain constrained, so unrestricted overwrite or code execution is not established.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
Findings (94)

YARA rule 'backdoor_persistence': Backdoor persistence with malicious payloads (shell commands, SSH key injection, hidden root users) [malware]

High
Category
YARA Match
Confidence
75% confidence
Finding

YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).

Content

Scanner excerpt · CHANGELOG.md (reported line 91)May include surrounding context.

md
ss.execPath` 指向 Electron 可执行文件),此前生成的任务 / 兜底 / 定时配置不带 `ELECTRON_RUN_AS_NODE=1`,开机拉起的是桌面程序本体、8787 无监听(新机首例实测)。现在自动检测并统一注入:Windows 主任务与每日备份任务(cmd /c + `set "ELECTRON_RUN_AS_NODE=1"`)、Windows Startup 兜底 .cmd(set 前缀)、Linux systemd 单元 / crontab @reboot / 备份定时器(`Environment=` / 行内前缀)、macOS 备份 LaunchAgent(EnvironmentVariables)。
- `lan status` 新增告警:任务或 Startup 兜底命令指向 Electron 可执行文件且缺少 `ELECTRON_RUN_AS_NODE` 时,提示重新 `lan enable` 修复。
- 非 Electron 宿主行为不变(纯 Node 下生成的命令与旧版一致)。
- `test/lan-windows-autostart.test.js` 新增 8 项(Electron / 非 Electron 双态:任务 XML / 兜底 cmd / 备份任务 / Linux 单元与 crontab / macOS plist / status 告警两例);全量 `npm test` 352/352。

#

YARA rule 'backdoor_persistence': Backdoor persistence with malicious payloads (shell commands, SSH key injection, hidden root users) [malware]

High
Category
YARA Match
Confidence
75% confidence
Finding

YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).

Content

Scanner excerpt · README.md (reported line 36)May include surrounding context.

md
maining v4-compatible; new writes use `yyyy/mm/dd` buckets with `layout plan/apply/rollback`; `doctor` adds four type-pressure indicators; `scan` supports exact rule + file + line-hash + expiry policies, while `--policy-candidates` only generates candidates.
> 🆕 **v0.24.1 (autostart profile)**: `lan enable` writes a per-user `serve-profile.json`; Windows task XML, Startup fallback, systemd and crontab reference that absolute profile. `serve --profile <file>` pins memoryHome / runtimeHome / configDir and fails closed when paths or the Windows user do not match, preventing S4U / no-profile startup from landing on `C:\Users\Default\.yottamemory` and returning remote MCP 401.
> 🆕 **v0.23.2 (ClawHub storefront)**: the ClawHub listing now ships an English-first bilingual summary and description (a dedicated `clawhub-dist/yotta-memory` publish copy), and the README gained the three usage cases above. Commands, options, output and the memory store format are unchanged.
> 🆕 **v0.18.1 (

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · README.md (reported line 133)May include surrounding context.

md
he current page.
- **Optional embedding plugin (v0.9.0)**: `recall --embedding <command>` or `config set embedding_cmd <command>` runs a local subprocess that accepts JSON on stdin and returns vectors on stdout; results are blended into the same ranking. Failures, timeouts, or malformed output automatically fall back to zero-dependency lexical recall.
- **Embedding cache**: vectors are cached under each memory root at `.embed/cache.json`, keyed by `sha256(command + text)`; only vectors are stored, not plaintext.
- The `tokens` field of `index.json` is a Chinese-tokenization term-frequency table (for TF scoring), **not** an access credential; auth tokens live at `.server/tokens.json`.
- Supports keywords, `--type` filter, `--limit` truncation, project-level priority.
- **Root de-duplication (v0.6.5)**: when project and user roots point at the same directory, recall / context uniquify roots so a file shows once.
- Hits accumulate `access_count` / `last_accessed` for lifecycle management.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · README.md (reported line 320)May include surrounding context.

md
_timeout` / `maintain_archived_utility` / `maintain_decay_halflife_<TYPE>` / `consolidate_*`) |
| `yotta-memory whoami` | Show the current agent identity and registration status |
| `yotta-memory iam <id> [--name <name>] [--user <user>] [--relationship <rel>] [--force]` | Register this agent's unique identity and auto-write the self profile (`agents.json`, ID must be unique) |
| `yotta-memory token new --agent <id> [--force]` / `token list` / `token revoke --agent <id>` | Create / list / revoke access tokens for agents (registered at `.server/tokens.json`) |
| `yotta-memory serve [--host 0.0.0.0] [--port 8787] [--no-auth] [--stdio]` | Start the MCP memory engine (streamable HTTP LAN / --stdio local zero-process mode; Bearer token + X-Agent-Id + X-Agent-Key auth) |
| `yotta-memory lan enable [--onstart] / disable / status` | Autostart management (Windows: scheduled task with an S4U principal — runs silently in the background with no console window; `--onstart` starts at boot without a l

Remote setup transmits credentials and memory over unencrypted HTTP without warning users.

High
Category
Output Handling
Confidence
99% confidence
Finding

The remote MCP example sends a bearer token, agent identity, agent key, and memory traffic over plaintext HTTP to another host. A network observer or on-path attacker could capture reusable credentials and read or modify traffic; encryption of stored private files does not protect this transport.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 71)May include surrounding context.

md
t <name>`(或原安装时使用的 `--dir` / `install.sh`);完成后回读 `yotta-memory --version` 与已安装 `SKILL.md` 的 `version:`,两者不一致时不得声称更新成功。

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 71)May include surrounding context.

md
yottameta/yotta-memory yotta-memory-install --agent <name>`(或原安装时使用的 `--dir` / `install.sh`);完成后回读 `yotta-memory --version` 与已安装 `SKILL.md` 的 `version:`,两者不一致时不

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 300)May include surrounding context.

md
4. **Agent Plugin 开箱装配(v0.16.6)**:插件 `mcp.json` 入口是 `bin/plugin-mcp.js`。宿主已替换身份占位符时优先用替换值;未替换时回退到 `YOTTA_MEMORY_AGENT_ID` + `YOTTA_MEMORY_AGENT_KEY_FILE`,再回退到 `

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · SKILL.md (reported line 625)May include surrounding context.

md
| 智能体 | 常见 MCP 配置位置 |
|---|---|
| Claude Code | 项目 `.mcp.json` 或用户级 `~/.claude.json` |
| Codex | `~/.codex/config.toml`(`[mcp_servers]`) |
| Cursor | 项目 `.cursor/mcp.json` 或用户级 |
| 其它(Trae / Qwen / Comate / Kimi 等) | 各自 MCP 配置 |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 685)May include surrounding context.

md
- 协议细节、目录结构与类型规则见 `references/protocol.md`,需要时读取,不要每次全读。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
85% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · USER_GUIDE.md (reported line 75)May include surrounding context.

md
> **v0.17.0 升级提示**:升级后请用 0.17.0 引擎执行一次 `yotta-memory reindex`。0.16.7 及更早引擎不识别年/月分层目录,旧引擎重建索引会漏掉分层条目。`doctor` 会报告平铺 / 分层同序号:内容相同的两份只索引一次;内容不同的两份都保留可读并给出路径告警;新写入不复用已占用的序号。

**卸载:** `npm rm -g @yottameta/yotta-memory`,并从智能体的 skills 目录删除整个 `yotta-memory` 文件夹。

## 3. 本机单机使用

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
85% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · USER_GUIDE.md (reported line 75)May include surrounding context.

md
> **v0.17.0 升级提示**:升级后请用 0.17.0 引擎执行一次 `yotta-memory reindex`。0.16.7 及更早引擎不识别年/月分层目录,旧引擎重建索引会漏掉分层条目。`doctor` 会报告平铺 / 分层同序号:内容相同的两份只索引一次;内容不同的两份都保留可读并给出路径告警;新写入不复用已占用的序号。

**卸载:** `npm rm -g @yottameta/yotta-memory`,并从智能体的 skills 目录删除整个 `yotta-memory` 文件夹。

## 3. 本机单机使用

Remote MCP setup transmits credentials and memory data over HTTP without warning that transport is unencrypted.

High
Category
Output Handling
Confidence
99% confidence
Finding

The remote MCP example sends the bearer token, agent key, and memory traffic over plain HTTP. An attacker able to observe or modify LAN traffic could steal credentials, impersonate the agent, and access or alter authorized memories; encryption at rest does not protect this transport.

Content

No source excerpt is available for this finding.

Private distillation reports are written in plaintext even in encrypted memory stores.

High
Category
Excessive Agency
Confidence
100% confidence
Finding

distillCore creates private report filenames ending in .md even in encrypted stores, while writeMemoryText encrypts only paths ending in .enc. Reports containing decrypted private statements and psychological summaries therefore persist in plaintext, defeating the store's confidentiality guarantee and potentially spreading into backups.

Content

No source excerpt is available for this finding.

The --keep-memories option retains encrypted files but deletes the keys needed to recover their contents.

High
Category
Excessive Agency
Confidence
100% confidence
Finding

identityRemoveCore deletes both password-wrapped and recovery-wrapped owner keys, then removes agent bindings, even when --keep-memories preserves the encrypted files. This makes the retained memories unreadable from the remaining live-store credentials; the transaction snapshot may permit recovery, but the option does not provide the preservation its name and help imply.

Content

No source excerpt is available for this finding.

LAN setup instructs clients to send authentication secrets over plain HTTP without warning about interception.

High
Category
Output Handling
Confidence
99% confidence
Finding

The FAQ explicitly pairs an HTTP URL with a Bearer token and agent key, exposing reusable authentication secrets and memory traffic to network interception. Advising secure file transfer for initial key delivery does not protect the key when it is subsequently transmitted in plaintext HTTP headers.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
75% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · CHANGELOG.md (reported line 94)May include surrounding context.

md
- **修复 Electron 宿主(YottaCode shim)下开机自启拉起 GUI 而非服务**:宿主运行时为 Electron 时(`process.versions.electron`;`process.execPath` 指向 Electron 可执行文件),此前生成的任务 / 兜底 / 定时配置不带 `ELECTRON_RUN_AS_NODE=1`,开机拉起的是桌面程序本体、8787 无监听(新机首例实测)。现在自动检测并统一注入:Windows 主任务与每日备份任务(cmd /c + `set "ELECTRON_RUN_AS_NODE=1"`)、Windows Startup 兜底 .cmd(set 前缀)、Linux systemd 单元 / crontab @reboot / 备份定时器(`Environment=` / 行内前缀)、macOS 备份 LaunchAgent(EnvironmentVariables)。
- `lan status` 新增告警:任务或 Startup 兜底命令指向 Electron 可执行文件且缺少 `ELECTRON_RUN_AS_NODE` 时,提示重新 `lan enable` 修复。
- 非 Electron 宿主行为不变(纯 Node 下生成的命令与旧版一致)。
- `test/lan-windows-autostart.test.js` 新增 8 项(Electron / 非 Electron 双态:任务 XML / 兜底 cmd / 备份任务 / Linux 单元与 crontab / macOS plist / status 告警两例);全量 `npm test` 352/352。

## v0.22.4 (2026-10-07)

Session Persistence

Medium
Category
Rogue Agent
Confidence
75% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · bin/yotta-memory.js (reported line 3832)May include surrounding context.

js
}
    if (platform === 'darwin') {
      const dir = opts.launchAgentDir || path.join(os.homedir(), 'Library', 'LaunchAgents');
      const plistPath = path.join(dir, BACKUP_LAUNCHD_LABEL + '.plist');
      fs.mkdirSync(dir, { recursive: true });
      fs.writeFileSync(plistPath, backupLaunchdPlist({
        time: time,

Session Persistence

Medium
Category
Rogue Agent
Confidence
75% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · bin/yotta-memory.js (reported line 3832)May include surrounding context.

js
}
    if (platform === 'darwin') {
      const dir = opts.launchAgentDir || path.join(os.homedir(), 'Library', 'LaunchAgents');
      const plistPath = path.join(dir, BACKUP_LAUNCHD_LABEL + '.plist');
      fs.mkdirSync(dir, { recursive: true });
      fs.writeFileSync(plistPath, backupLaunchdPlist({
        time: time,

Session Persistence

Medium
Category
Rogue Agent
Confidence
75% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · bin/yotta-memory.js (reported line 3834)May include surrounding context.

js
const dir = opts.launchAgentDir || path.join(os.homedir(), 'Library', 'LaunchAgents');
      const plistPath = path.join(dir, BACKUP_LAUNCHD_LABEL + '.plist');
      fs.mkdirSync(dir, { recursive: true });
      fs.writeFileSync(plistPath, backupLaunchdPlist({
        time: time,
        nodePath: nodePath,
        scriptPath: scriptPath,

Session Persistence

Medium
Category
Rogue Agent
Confidence
75% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · bin/yotta-memory.js (reported line 3842)May include surrounding context.

js
electronHost: opts.electronHost,
      }), 'utf8');
      const uid = typeof process.getuid === 'function' ? process.getuid() : '';
      const bootstrap = schedulerTry(opts, 'launchctl', ['bootstrap', 'gui/' + uid, plistPath]);
      if (!bootstrap.ok) schedulerTry(opts, 'launchctl', ['load', '-w', plistPath]);
      return { error: false, scheduler: 'macos-launchd', plist: plistPath, text: '已注册 macOS LaunchAgent 每日备份。' };
    }

Session Persistence

Medium
Category
Rogue Agent
Confidence
75% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · bin/yotta-memory.js (reported line 3843)May include surrounding context.

js
}), 'utf8');
      const uid = typeof process.getuid === 'function' ? process.getuid() : '';
      const bootstrap = schedulerTry(opts, 'launchctl', ['bootstrap', 'gui/' + uid, plistPath]);
      if (!bootstrap.ok) schedulerTry(opts, 'launchctl', ['load', '-w', plistPath]);
      return { error: false, scheduler: 'macos-launchd', plist: plistPath, text: '已注册 macOS LaunchAgent 每日备份。' };
    }
  } catch (error) {

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-memory'.

Medium
Category
MCP Rug Pull
Confidence
85% confidence
Finding

The documented npx command leaves the package version unspecified and uses -y, allowing execution of a changing registry release without an installation prompt. This creates a supply-chain reproducibility risk if copied, although its placement in a historical changelog rather than an active MCP configuration reduces immediate exposure.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y --package @yottameta/yotta-memory'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The recommended installer command resolves an unpinned npm package and automatically accepts installation. A compromised future release could execute with the invoking user's permissions and modify installed skill files; the changelog itself does not execute the command.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-memory'.

Medium
Category
MCP Rug Pull
Confidence
84% confidence
Finding

This historical installation example executes an npm package without fixing its version, so the code obtained later need not match the release described. It is a conditional supply-chain risk when copied, not evidence of an actively configured or malicious MCP server.

Content

No source excerpt is available for this finding.