Back to skill

Security audit

spec-stateflow

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed software planning workflow that writes spec documents and task trackers, with no hidden execution, credential use, or network behavior found.

Install this if you want a formal spec-driven development workflow. Expect it to create and update requirements, design, and task-tracking files, and to guide code changes and commits during execution. Use continuous operation only when you are comfortable with the agent advancing through approved tasks without per-task review.

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
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (8)

YARA rule 'agent_skill_prompt_injection_hidden_instructions': Prompt injection or hidden instructions embedded in AI agent skill text [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 9)May include surrounding context.

md
re, complex architecture, multi-module integration, database/UI design, or any task requiring structured planning and execution. Trigger words: new feature, architecture design, requirement implementation, spec workflow, state recovery, task tracking, 新功能, 复杂架构, 多模块, 需求实现, spec工作流, 状态恢复, 任务追踪"
alwaysApply: false
---

# Spec Stateflow

**Important: You must follow these rules. Each phase must be confirmed by the user before proceeding to the next phase.**

## When to Use This Skill

Use this skill for **structured development workflow** when you need to:

- Develop new features from scratch
- Design complex architecture
- Integrate multiple modules
- Ensure high-quality requirement analysis and acceptance criteria

**Do NOT use for:**
- Simple requirements
- Simple bug fixes (single file / single method / obvious root cause)
- Documentation updates
- Configuration changes

> **Fix vs Simple bug fix boundary**: A "Fix" task (use Spec) inv

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
85% confidence
Finding

Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.

Content

Scanner excerpt · SKILL.md (reported line 425)May include surrounding context.

md
**Step 5: Use programmatic verification, not manual reading**
- For "does it exist", "is it aligned", "does it match" — prefer Bash tools (grep, javap, diff) for programmatic verification
- Never judge interface/impl alignment, method existence, or parameter matching by reading alone

**Step 6: Update `tasks.md` before executing**
- Before making any code changes, update `{SPEC_PATH}/tasks.md` first

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger word list contains ambiguous everyday engineering phrases in English and Chinese, including 'new feature' and '复杂架构', which may match many benign requests. In an agent environment, such ambiguity can cause unintended skill invocation and let the skill's internal instructions dominate normal handling of routine tasks.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger word list contains ambiguous everyday engineering phrases in English and Chinese, including 'new feature' and '复杂架构', which may match many benign requests. In an agent environment, such ambiguity can cause unintended skill invocation and let the skill's internal instructions dominate normal handling of routine tasks.

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 250)May include surrounding context.

text

**Session compressed?** Jump to [Compression Recovery](#compression-recovery).
**Continuous batch authorized by the user (e.g. "you may continue implementing" / "complete the rest without asking"), or running under driver/monitor automation?** Jump to [Continuous Operation Mode](#continuous-operation-mode).

### State Navigation

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 323)May include surrounding context.

text

**Session compressed?** Jump to [Compression Recovery](#compression-recovery).
**Continuous batch authorized by the user (e.g. "you may continue implementing" / "complete the rest without asking"), or running under driver/monitor automation?** Jump to [Continuous Operation Mode](#continuous-operation-mode).

### State Navigation

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

In this JSON test prompt, the activation-style phrase '帮我实现这个功能' ('help me implement this feature') is very general and overlaps with common everyday requests. The surrounding text does not define narrow trigger constraints, exclusions, or a specific invocation context, so it could ambiguously map many unrelated requests into the spec-stateflow path.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The phrase '这个bug需要修一下' ('this bug needs fixing') is a broad natural-language trigger that could match many ordinary conversations. The file does not provide explicit boundaries for when such wording should invoke the Fix/spec-stateflow process versus when it should be treated as a simple support request.

Content

No source excerpt is available for this finding.