Back to skill

Security audit

awam-todo

Security checks for vulnerabilities and agentic risk

Overview

This todo skill needs review because confirmed flaws can expose private tasks, overwrite files outside its storage folder, and stop unrelated services.

Prefer waiting for fixes before using private or important records. For evaluation, use isolated test data, leave the web service off, avoid the Windows launcher and batch-file editor commands, and back up migration destinations. Moving storage does not relocate the task index. Source review and in-memory checks confirmed material defects; live browser and Windows behavior were not tested. The evidence supports Review, without establishing intentional malicious behavior.

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 (3)

T09 · Insecure Skill Coding Practices

Warning
Location
web/server.py
Finding

Untrusted websites can read and modify the local task database

Content
View full analysis

Impact

Disclosure of task text, notes, workspace paths, and storage metadata; unauthorized creation, alteration, or deletion of tasks. The external page can transmit disclosed responses elsewhere using its own network access.

T09 · Insecure Skill Coding Practices

Error
Location
scripts/todo.py
Finding

Unvalidated task dates permit writes outside the configured storage directory

Content
View full analysis

Impact

Creation or destructive overwrite of writable Markdown files outside storage, including the installed skill manual under the default layout. Task text and notes influence the serialized contents. This is a Markdown-scoped write primitive, not demonstrated arbitrary-extension writing or code execution.

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/start-awam-todo.ps1
Finding

Startup cleanup forcefully terminates services outside this installation

Content
View full analysis

Impact

Abrupt shutdown of other Python web services, interrupting requests and active work. The first pass is not limited to the configured port.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (43)

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 44)May include surrounding context.

md
command is `scripts/todo.py` (Python 3).

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 116)May include surrounding context.

md
| CLI / API output (`todo.py`, `server.py`) | **English, fixed** | All user-facing CLI output, argparse help, the tkinter dialog, web API error/message fields a

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 239)May include surrounding context.

md
constant in `todo.py`; to add another editor, add one entry.

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 266)May include surrounding context.

md
- This config only affects the check/ask side in `todo.py`; the web service's

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 368)May include surrounding context.

md
constants declared together at the top of `scripts/todo.py`

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 849)May include surrounding context.

md
- The backend reuses `todo.py`'s parse / validate / index logic; when the CLI

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 889)May include surrounding context.

md
restarting. Web changes and CLI `todo.py` share the same

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 79)May include surrounding context.

md
Capture the clipboard with a global hotkey (no filtering) | Start the web UI (`server.py`); press the hotkey (default `Ctrl+Alt+T`); clipboard text feeds the t

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 116)May include surrounding context.

md
| CLI / API output (`todo.py`, `server.py`) | **English, fixed** | All user-facing CLI output, argparse help, the tkinter dialog, web API error/message fields a

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 256)May include surrounding context.

md
| `ui.port` | Web service port, default `8796` (matches `web/server.py` default) |

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 535)May include surrounding context.

md
content. The service runs **inside the web UI process** (`web/server.py`), so

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 792)May include surrounding context.

md
`web/server.py` in the background and opens the browser; on "skip", it

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 800)May include surrounding context.

md
the `server.py` start command.

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 117)May include surrounding context.

md
| Web UI (`web/index.html`) | **i18n: English default, switchable to Chinese** | Inline i18n word tables (zh/en) with **no external libraries**, offline-safe. P

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 731)May include surrounding context.

md
When editing `web/index.html`, always follow:

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 750)May include surrounding context.

md
`node scripts/smoke-web.js` (front-end render smoke, minimal DOM stubs).

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% 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 · SKILL.md (reported line 865)May include surrounding context.

md
dependency/subtask guards)
- `PATCH  /api/todos/<id>/status`   status only (conflict → 409 + conflicts;
  `force` supported)
- `DELETE /api/todos/<id>`          delete (also cleans other tasks'
  references to it: dependencies / parent)
- `POST   /api/todos/apply-patches` apply a patch batch and persist once
  (web delayed-save entry)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% 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 · SKILL.zh-CN.md (reported line 685)May include surrounding context.

md
- `POST   /api/todos`               新增(重复检测命中返回 409 + duplicates,可 `force` / `update_id`)
- `PUT    /api/todos/<id>`          更新字段(含状态,带依赖/子任务守卫)
- `PATCH  /api/todos/<id>/status`   仅改状态(冲突返回 409 + conflicts,可 `force`)
- `DELETE /api/todos/<id>`          删除(同时清理其他任务对它的依赖 / 父任务引用)
- `POST   /api/todos/apply-patches` 批量应用 patch 一次性落盘(网页延迟保存入口)
  body: `{"revision": <GET /api/todos 返回的指纹>, "patches": [{"op":"create|update|status|delete", ...}]}`;
  保存前对比 revision 与当前文件指纹,不一致时把 patch 应用到最新文件(合并语义),响应含

Env Variable Harvesting

High
Category
Data Exfiltration
Confidence
60% confidence
Finding

Code enumerates, copies, or searches environment variables for secrets. Bulk environment access can collect credentials unrelated to the skill's stated purpose.

Content

Scanner excerpt · scripts/selftest.py (reported line 20)May include surrounding context.

python
def run(tmp, *args, timeout=60, env=None):
    """跑一条 CLI 命令。带超时保护:卡住的一律当成失败,不让自测整体挂死。
    默认注入 AWAM_TODO_UI_PROMPT=off:自测里不弹「是否启动网页 UI」的对话框。"""
    e = dict(os.environ)
    e.setdefault("AWAM_TODO_UI_PROMPT", "off")
    if env:
        e.update(env)

Storage migration can overwrite existing destination data despite its preservation guarantee.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

Migration uses copy2 and copytree with dirs_exist_ok=True without rejecting destination collisions, so existing daily files and archive files can be overwritten. Rollback removes only newly created daily files and does not restore overwritten content, while count-based verification cannot detect these losses; this is particularly damaging for a personal-record storage tool that promises preservation.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% 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 · web/server.py (reported line 25)May include surrounding context.

python
POST   /api/todos                       新增任务(重复检测命中时返回 409 + duplicates)
  PUT    /api/todos/<id>                  更新任务字段(含状态,带依赖/子任务守卫)
  PATCH  /api/todos/<id>/status           仅改状态(冲突返回 409 + conflicts)
  DELETE /api/todos/<id>                  删除任务(同时清理其他任务对它的引用)
  POST   /api/todos/apply-patches         批量应用 patch 并一次性落盘(网页端延迟保存入口)
  POST   /api/workspace/open              用系统文件管理器打开目录
  POST   /api/workspace/editor            用 env.json 配置的编辑器打开目录

Batch updates can persist a blocked status change even when reported as rejected, bypassing the promised dependency and subtask guards.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

Batch updates mutate the shared task object through _apply_fields before dependency and child-task guards run, and rejection does not restore the original object. A successful patch affecting the same date can cause the rejected status to be saved, while later patches can also observe that rejected state and bypass dependency guards.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

CLI/API output and capture dialogs are forced to English without user opt-in.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

The language policy fixes all user-facing CLI output, help, system dialogs, and API messages to English, regardless of the user's language. Although conversation follows the user's language and the web UI offers a selector, neither provides a language choice for these fixed-English interfaces, and no domain-specific justification is given.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 277)May include surrounding context.

powershell
python ...\todo.py env --set capture.enabled=false          # stop listening with the UI
python ...\todo.py env --set capture.hotkey="Ctrl+Shift+X" # change the hotkey (re-registers on UI start)
python ...\todo.py env --set capture.confirm_delete=false   # delete without asking (not recommended)
python ...\todo.py env --set capture.toast=false            # no corner toast
python ...\todo.py env --set capture.triage_types.delete=false
                                                             # disable delete triage (falls back to Add + ask)