ClawHub
Back to skill

Security audit

Git Helper

Security checks for vulnerabilities and agentic risk

Overview

This is a Git help skill that gives command explanations and safety warnings, with no executable code, persistence, or hidden data access.

This skill is reasonable to install if you want Git command help. Still review any suggested destructive Git commands carefully, especially `reset --hard`, `clean`, rebase, amend, or force-push operations, because those can lose local work or rewrite shared history if used in the wrong repository state.

A.I.G

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (12)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
70% 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 · README.md (reported line 8)May include surrounding context.

md
- 理解特定 Git 命令的作用（如 `git rebase -i`, `git fetch` vs `git pull` 等）。
- 根据任务目标提供确切的 Git 操作步骤（如“撤销上一次提交但保留代码”、“合并最近的三次提交”等）。
- 提供应对复杂情况的最佳实践（如处理合并冲突、使用 `git reflog` 恢复误删代码）。
- 对破坏性操作（如 `git reset --hard`、`git push --force`）进行安全警告并提供更安全的替代方案。

支持中英文双语输出，会根据您的提问语言自动适配。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
65% 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 · README.md (reported line 8)May include surrounding context.

md
- 理解特定 Git 命令的作用（如 `git rebase -i`, `git fetch` vs `git pull` 等）。
- 根据任务目标提供确切的 Git 操作步骤（如“撤销上一次提交但保留代码”、“合并最近的三次提交”等）。
- 提供应对复杂情况的最佳实践（如处理合并冲突、使用 `git reflog` 恢复误删代码）。
- 对破坏性操作（如 `git reset --hard`、`git push --force`）进行安全警告并提供更安全的替代方案。

支持中英文双语输出，会根据您的提问语言自动适配。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
65% 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 · README.md (reported line 22)May include surrounding context.

md
- 理解特定 Git 命令的作用（如 `git rebase -i`, `git fetch` vs `git pull` 等）。
- 根据任务目标提供确切的 Git 操作步骤（如“撤销上一次提交但保留代码”、“合并最近的三次提交”等）。
- 提供应对复杂情况的最佳实践（如处理合并冲突、使用 `git reflog` 恢复误删代码）。
- 对破坏性操作（如 `git reset --hard`、`git push --force`）进行安全警告并提供更安全的替代方案。

支持中英文双语输出，会根据您的提问语言自动适配。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
65% 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 22)May include surrounding context.

md
- 理解特定 Git 命令的作用（如 `git rebase -i`, `git fetch` vs `git pull` 等）。
- 根据任务目标提供确切的 Git 操作步骤（如“撤销上一次提交但保留代码”、“合并最近的三次提交”等）。
- 提供应对复杂情况的最佳实践（如处理合并冲突、使用 `git reflog` 恢复误删代码）。
- 对破坏性操作（如 `git reset --hard`、`git push --force`）进行安全警告并提供更安全的替代方案。

支持中英文双语输出，会根据您的提问语言自动适配。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
65% 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 78)May include surrounding context.

md
- 理解特定 Git 命令的作用（如 `git rebase -i`, `git fetch` vs `git pull` 等）。
- 根据任务目标提供确切的 Git 操作步骤（如“撤销上一次提交但保留代码”、“合并最近的三次提交”等）。
- 提供应对复杂情况的最佳实践（如处理合并冲突、使用 `git reflog` 恢复误删代码）。
- 对破坏性操作（如 `git reset --hard`、`git push --force`）进行安全警告并提供更安全的替代方案。

支持中英文双语输出，会根据您的提问语言自动适配。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
65% 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 · evals/evals.json (reported line 18)May include surrounding context.

json
- 理解特定 Git 命令的作用（如 `git rebase -i`, `git fetch` vs `git pull` 等）。
- 根据任务目标提供确切的 Git 操作步骤（如“撤销上一次提交但保留代码”、“合并最近的三次提交”等）。
- 提供应对复杂情况的最佳实践（如处理合并冲突、使用 `git reflog` 恢复误删代码）。
- 对破坏性操作（如 `git reset --hard`、`git push --force`）进行安全警告并提供更安全的替代方案。

支持中英文双语输出，会根据您的提问语言自动适配。

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
70% 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 22)May include surrounding context.

md
3. **Task to Command Translation:** If the user wants to achieve a specific goal, provide the exact commands needed. Address advanced topics like Submodules, complex Merge Conflicts, and `.gitignore` troubleshooting when necessary.
4. **Workflow Guidance:** Provide best practices for common workflows like feature branching, resolving merge conflicts, syncing with upstream, or squashing commits.
5. **Safety First & Risk Mitigation:** 
   - For destructive commands (`git reset --hard`, `git push --force`, `git clean -fd`), ALWAYS include a strong warning about potential data loss.
   - **Uncommitted Code Warning:** Explicitly state that uncommitted changes destroyed by `reset --hard` or `clean` CANNOT be recovered using `git reflog`.
   - **Public History Warning:** Strongly warn against rewriting history (`rebase`, `commit --amend`) on shared/public branches (like `main` or `develop`).
   - **Sensitive Data Leaks:** If the user accidentally committed passwords or API keys, advise them to immediately revoke the credentials and suggest tools like `git filter-repo` or BFG, rather than just using `git rm`.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The trigger description is very broad and can activate on a large range of normal development questions, increasing the chance this skill is invoked in situations where its guidance is unnecessary or unsafe. Broad auto-invocation expands the attack surface for prompt-controlled command generation, especially because the skill is empowered to recommend potentially destructive Git operations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This manifest-like JSON explicitly requires 'A Chinese explanation' for one prompt and 'An English explanation' for another, creating a hard language constraint in the skill's expected behavior. The file does not indicate that language choice is user-configurable or justified by a region-specific requirement, which matches the locale-policy violation criteria.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The expected output specifies 'An English explanation' as a required behavior. Enforcing a specific language in natural-language instructions or config without opt-in is a policy concern under the language/locale rule.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This expected output requires 'A Chinese guide', which hard-codes a locale in the skill evaluation. Because no opt-in or justified locale limitation is present, this 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
95% confidence
Finding

The expected output requires 'An English step-by-step guide', which is a fixed language requirement embedded in the file. The policy says not to force a language unless the user is given a choice or the restriction is clearly justified.

Content

No source excerpt is available for this finding.