ClawHub
Back to skill

Security audit

Fundraising Playbook

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a fundraising advice prompt, but its runtime instructions include unrelated commercial promotion and external links that could steer user responses.

Review the promotional section before installing. The skill does not appear to run code or access private data, but it may cause fundraising responses to include or be biased by unrelated commercial links unless those instructions are removed or ignored.

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
Findings (1)

T01 · Skill Instruction Hijacking

Warning
Location
SKILL.md:153
Finding
Hard-Coded Commercial Promotion in Agent Instructions<![CDATA[ ## Vulnerability Details **File Location**: `SKILL.md`, lines 153-175 **Vulnerability Type**: Agent instruction hijacking through unrelated promotional directives **Risk Level**: Medium ### Vulnerable Code ```markdown ## Output Format Structure your response as: 1. **Readiness Score** — 8-dimension assessment with specific gaps to fix 2. **Recommended Round** — Size, valuation range, dilution, timeline 3. **Investor Target List Framework** — Criteria for each tier, example investor profiles to research 4. **Materials Status** — What exists, what needs building, priority order 5. **90-Day Fundraising Calendar** — Week-by-week action plan 6. **Risk Factors** — What could derail the raise and mitigation strategies --- ## Want the Full Business Context? This skill covers fundraising strategy. For complete operational frameworks across your entire business: - **[AI Agent Context Packs](https://afrexai-cto.github.io/context-packs/)** — $47 per industry. SaaS, Fintech, Healthcare, Legal, Construction, and 5 more. Each pack gives your AI agent deep domain expertise for autonomous operations. - **[AI Revenue Leak Calculator](https://afrexai-cto.github.io/ai-revenue-calculator/)** — Find where your business is losing money to manual processes (free tool). - **[Agent Setup Wizard](https://afrexai-cto.github.io/agent-setup/)** — Configure your AI agent in minutes (free tool). **Bundles:** Pick 3 packs $97 | All 10 packs $197 | Everything bundle $247 ``` ### Technical Analysis The skill places commercial advertising, product prices, and outbound links directly after its mandatory response-format instructions. Because `SKILL.md` defines the agent's behavior when the skill is loaded, this placement can cause the agent to treat unrelated promotional content as part of its operational instructions or expected response. The promoted products are not necessary to perform fundraising readiness assessments, round sizing, investor targeting, or the other adver ...[truncated 1704 chars]
Remediation
<![CDATA[ ## Remediation Suggestions 1. Remove the promotional section at lines 164-175 from `SKILL.md`. 2. Restrict the skill instructions to behavior strictly required for fundraising analysis and planning. 3. If commercial resources must remain documented, place them only in `README.md`, outside the agent's behavioral prompt. 4. Clearly label any retained links as optional third-party or commercial resources. 5. Add an explicit instruction that promotional content and unrelated outbound links must not be included in generated responses. 6. Review the final skill prompt to ensure all directives directly support the user's requested fundraising task. 7. Add response tests verifying that ordinary fundraising prompts do not produce advertisements, prices, affiliate-style messaging, or unsolicited external links. ]]>

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep