Back to skill

Security audit

Workspace Bootstrap

Security checks for vulnerabilities and agentic risk

Overview

This workspace setup skill is generally purpose-aligned, but users should review it because it can overwrite existing workspace files and its documentation may lead users to store or publish sensitive workspace data.

Install only if you are comfortable reviewing generated files before use. Run bootstrap or wizard in a new empty directory, back up any existing workspace first, avoid storing plaintext secrets in TOOLS.md or user-data/apis, and do not use git add . or copy user-data until you have reviewed and excluded private data.

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

Warning
Location
templates/WORKSPACE-TEMPLATE.md:32
Finding

Workspace Templates Encourage Plaintext Credential Storage

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/bootstrap.sh:36
Finding

Bootstrap Script Silently Overwrites Existing Core Workspace Files

Content
View full analysis
Remediation
View remediation
&2 exit 1 fi ``` 6. Canonicalize and validate the workspace root before writing. 7. Use a staging directory and atomic renames so partial failures do not leave the workspace inconsistent. 8. Add automated tests confirming that existing files remain unchanged unless an explicit force option is supplied. ]]>
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (35)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill explicitly advertises one-command workspace creation and automatic generation of directories and core files, but it does not warn users that it will modify the filesystem or potentially overwrite existing content. In a bootstrap skill, omission of scope, overwrite behavior, and confirmation requirements can lead to unintended data loss or corruption if run in the wrong directory or against an existing workspace.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The migration/check scenario presents the skill as suitable for upgrading an existing workspace, but it does not clarify whether the check operation is read-only or whether migration-related use may alter files and structure. This ambiguity is dangerous because users may invoke the skill on valuable existing workspaces assuming it is safe, leading to accidental modification or overwrite of user data.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The quickstart explicitly tells users to initialize a Git repository and push the generated workspace, but the workspace contains files like SOUL.md, USER.md, MEMORY.md, and user-data that may hold personal, operational, or sensitive configuration details. Without any warning, review step, or .gitignore guidance, users may unintentionally publish private data to a remote repository.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The guide recommends copying USER.md and the entire user-data directory between machines via scp, which can transfer personal information and other sensitive workspace state without any data-classification warning or minimization guidance. Although scp is encrypted in transit, the risk is inadvertent propagation of sensitive files to unmanaged hosts, overbroad copying, and poor handling of private user content.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The USER.md example states that all time display, dialogue, and scheduled tasks must use local time, and the setup example hard-codes Asia/Shanghai. This imposes a locale/timezone convention in natural language without indicating that users may choose a different locale or opt in to that behavior.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The guide instructs users to git add . and push the entire workspace to a remote repository without warning that the workspace may contain sensitive material such as USER.md, memory files, reports, temp data, or other private workspace content. In an agent workspace context, this increases the chance of unintentional data exfiltration to shared or public remotes.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The README is entirely written in Chinese and defines the skill/persona configuration in Chinese without any indication that users may choose another language or that the skill is intended only for a Chinese-language context. This creates a natural-language locale policy concern because the file appears to prescribe a single language experience without opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The README presents the skill's identity, instructions, and configuration exclusively in Chinese, which can amount to forcing a specific language without user opt-in. The policy allows locale constraints only when explicitly offered as a choice or clearly justified as region-specific, neither of which is stated here.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The script copies template files directly into the target workspace using fixed destination names and does not check whether those files already exist. If run against an existing workspace, it can silently overwrite user-maintained documentation or configuration, causing data loss or corruption of the workspace state.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This shell script contains user-facing natural language in Chinese, starting with the header comment and continuing throughout its status messages. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale constraint is clearly documented and justified, which is not present here.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script unconditionally writes SOUL.md and USER.md using shell redirection, which will overwrite any existing files at those paths without prompting or creating backups. Because WORKSPACE_ROOT is user-controllable and the wizard is intended to run in an existing workspace, this can cause accidental destruction of prior configuration or user-authored content.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The generated text states that all time display, dialogue, and scheduled tasks must use local time and must not use UTC, using mandatory language. This imposes a locale policy unconditionally rather than offering a user choice or documenting a justified region-specific constraint.

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 · templates/AGENTS-template.md (reported line 59)May include surrounding context.

md
## Red Lines

- Don't exfiltrate private data. Ever.
- Don't run destructive commands without asking.
- `trash` > `rm` (recoverable beats gone forever)
- When in doubt, ask.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The template authorizes periodic deletion of files in temp/ older than 7 days without describing confirmation requirements, scope restrictions, dry-run behavior, or failure safeguards. In an agent skill context, undocumented autonomous deletion increases the risk of unintended data loss, especially if temp/ is mis-scoped, symlinked, or contains user-important files.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

Lines L31-L34 describe the heartbeat as limited to self-check behavior and exception-only notifications. However, elsewhere the same file assigns the heartbeat active business and maintenance operations, including immediate handling/pushing of urgent matters, calendar reminders, memory updates, and temp-file cleanup. This is an intent-level contradiction within the documentation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The entire template content, headings, and instructions are written in Chinese, which imposes a specific language/locale on skill users without offering any choice or opt-in. The policy allows locale constraints only when explicitly justified or when users are given a language option, neither of which appears here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This markdown template uses Chinese throughout headings, examples, and instructions, which can impose a specific language on users without any opt-in or stated regional justification. The policy requires flagging language or locale constraints when the skill does not explicitly offer a language choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The template instructs that all displayed times, conversations, and scheduled tasks must use Beijing time (UTC+8) and explicitly forbids UTC. This is a natural-language locale/timezone policy constraint applied universally, without offering user choice or framing it as a justified region-specific requirement.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

This markdown file contains all user-facing instructions and examples in Chinese, but does not provide an opt-in, alternative language, or justification for the locale constraint. Under the policy, forcing a specific language without user choice can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The template explicitly suggests a TOOLS.md file for API keys, SSH, and related configuration but provides no warning against storing plaintext secrets in the workspace. In a shared or synced workspace, this can lead to accidental credential exposure through version control, backups, logs, or cross-agent access.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
85% 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 · templates/WORKSPACE-TEMPLATE.md (reported line 283)May include surrounding context.

配置方法:

bash
# 编辑 crontab
crontab -e

# 添加任务
0 3 * * 0 cd /path/to/workspace && python3 scripts/validate-index-paths.py

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script's comments, user-facing output, and interactive test inputs are all fixed in Chinese, and it also hard-codes the timezone value "Asia/Shanghai". This imposes a specific language/locale without offering a choice or documenting a justified region-specific constraint, which matches the policy-violation criteria.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This shell script contains exclusively Chinese comments, echo output, and here-doc input values, which imposes a specific language/locale on users and test execution. Under the policy, forcing a language without user opt-in or a clearly documented region-specific justification is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This shell script contains user-facing output and instructional comments only in Chinese, which imposes a language choice on users without opt-in. The policy requires flagging language or locale constraints unless the file offers a language choice or clearly documents a justified region-specific limitation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This shell script's comments and all user-facing echo messages are in Chinese, which imposes a specific language on users and operators. Under the policy, language constraints should either be optional/opt-in or clearly justified as region-specific; neither is present in this file.

Content

No source excerpt is available for this finding.