Ae1
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md - `SKILL.md` - complete instructions (no reference to this meta skill needed)
Security audit
Financial Financial Analysis
Security checks for vulnerabilities and agentic risk
Overview
This financial-analysis skill is broadly disclosed and purpose-aligned, though it includes adjacent PowerPoint, Excel, and skill-packaging workflows that users should understand before installing.
Install only if you are comfortable with a skill that can read and write local Excel/PowerPoint artifacts and, when explicitly asked, create packaged reusable skills from PowerPoint templates. Review any generated skill before installing it, especially if the template or instructions came from someone else.
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
- Rogue AgentSelf-Modification, Session Persistence
- Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
- MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (9)
Self-Modification
High
- Category
- Rogue Agent
- Confidence
- 97% confidence
- Finding
The skill explicitly instructs creation and packaging of a new skill by writing a SKILL.md and bundling assets, which is a direct self-modification or capability-proliferation pathway. In an adversarial setting, user-supplied template content or embedded instructions could be repackaged into reusable agent behavior, enabling persistence of unsafe prompts or expansion beyond the original trust boundary.
- Content
md 2. **Analyze template** - extract layouts, placeholders, dimensions 3. **Initialize skill** - use the `skill-creator` skill to set up the skill structure 4. **Add template** - copy .pptx to `assets/template.pptx` 5. **Write SKILL.md** - follow template below with PPT-specific details 6. **Create example** - generate sample presentation to validate 7. **Package** - use the `skill-creator` skill to package into a .skill file
Context-Inappropriate Capability
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The top-level financial-analysis skill exposes a PPT template skill-packaging workflow that is outside the stated scope of financial modeling and analysis. That scope expansion creates an unexpected capability to transform user-supplied files into new skills, which increases the attack surface and can be abused to pivot from business analysis into artifact generation and indirect code/content propagation.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The trigger list for
audit-xlsincludes phrases like "sanity check this" and "something's off in my model," which are broad, conversational expressions that can occur in ordinary discussion and may cause unintended activation. The description does not provide exclusion conditions or tighter scoping to spreadsheet-specific contexts.- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The description says the skill should be used for requests like "strategic review" or "any request to systematically evaluate competitive dynamics across an industry," which is expansive and not tightly bounded. Because there are no explicit exclusion conditions, this wording could overlap with many non-deck or non-skill-specific business conversations.
- Content
Context-Inappropriate Capability
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
Embedding a general-purpose skill-creation capability inside a financial-analysis package gives the agent a meta-programming pathway unrelated to the declared purpose. In practice this enables creation or modification of new skills and instructions, which can be leveraged to persist unsafe behavior, repackage adversarial content, or bypass expected task boundaries.
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 88% confidence
- Finding
The file presents section headings and explanatory text in Chinese (for example, "来源" and "底层技能") while most of the rest of the skill is in English, with no indication that the user can choose their preferred language. This can impose a language/locale choice implicitly rather than by user opt-in.
- Content
Intent-Code Divergence
Low
- Category
- Not specified by scanner
- Confidence
- 78% confidence
- Finding
The description and heading emphasize 'Creates self-contained PPT template SKILLS (not presentations)' and 'This skill creates SKILLS, not presentations.' However, the workflow explicitly includes 'Create example - generate sample presentation to validate,' which contradicts the strict claim that it does not create presentations.
- Content
Context-Inappropriate Capability
Low
- Category
- Not specified by scanner
- Confidence
- 80% confidence
- Finding
The manifest emphasizes financial modeling, valuation, competitive analysis, and deck QC, but 'pptx-author' is a generic capability for programmatically generating PowerPoint files on disk via Python scripts. While adjacent to presentation work, it is broader than analysis/QC and functions as a reusable document-generation tool.
- Content