Back to skill

Security audit

KibiBot Skill

Security checks for vulnerabilities and agentic risk

Overview

The skill matches KibiBot's stated purpose, but it gives an agent financially consequential actions without clear per-action confirmation guidance.

Review before installing. Use the least-privilege API key you can, avoid enabling Agent Reload unless you intentionally want the agent to move USDC/USDT from your trading wallet, and require explicit confirmation before token creation, reload, or reload-disable actions.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (9)

Session Persistence

Medium
Category
Rogue Agent
Confidence
78% confidence
Finding

The front-matter description broadly instructs the agent to use this skill for token creation, credit reloads, and LLM gateway actions whenever related requests appear, but it does not state any requirement for confirmation on state-changing or fund-moving operations. In agent ecosystems, such broad activation text can contribute to unsafe automatic execution or persistent use of a capability beyond what the user expected.

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: kibibot
description: Create tokens on-chain, check fee earnings, check Kibi Credit balance, trigger agent credit reload, and interact with KibiBot's Agent API and Kibi LLM Gateway. Use when asked to create a token via KibiBot, check fee earnings across chains and platforms, check KibiBot Kibi Credit balance, check daily token creation quota, reload credits from trading wallet, or make LLM calls through KibiBot's gateway.
---

# KibiBot Skill

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill enables real token deployment on public chains but does not place a clear, prominent warning near the token-creation capability that these actions are irreversible and may consume sponsored quota or wallet funds once free quota is exhausted. In an agent setting, that omission increases the chance of a user triggering financially consequential blockchain actions without informed consent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The credit reload feature can move USDC/USDT from the trading wallet into Kibi Credits, yet the documentation frames it as a convenience feature without a sufficiently prominent warning about real fund movement and spending-limit effects. An autonomous or semi-autonomous agent could therefore trigger balance transfers the user did not fully anticipate.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 704)May include surrounding context.

Compatible with Claude models via Anthropic Messages API format.

bash
curl https://llm.kibi.bot/v1/messages \
  -H "Content-Type: application/json" \
  -H "X-Api-Key: YOUR_KB_API_KEY" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 295)May include surrounding context.

md
# KibiBot Agent API — Quick Reference

**Base URL:** `https://api.kibi.bot/agent/v1`  
**Auth:** `X-Api-Key: kb_...` header on all requests

---

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/api.md (reported line 3)May include surrounding context.

md
# KibiBot Agent API — Quick Reference

**Base URL:** `https://api.kibi.bot/agent/v1`  
**Auth:** `X-Api-Key: kb_...` header on all requests

---

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The documentation exposes an irreversible POST /balance/credits/reload/disable operation with only a brief parenthetical note and no prominent warning, confirmation guidance, or operational safeguards. In an agent skill context, destructive financial-control actions are higher risk because an LLM-driven agent may invoke them from ambiguous prompts or automation flows, causing permanent loss of reload functionality for the user.

Content

No source excerpt is available for this finding.

Excessive Permissions

Low
Category
Privilege Escalation
Confidence
82% confidence
Finding

The setup instructions encourage enabling optional permissions for LLM access and especially Agent Reload, which authorizes fund movement from a trading wallet. While described as optional, combining broad capabilities in a single key increases blast radius if the key is misused by the agent, leaked, or used in an unexpected context.

Content

Scanner excerpt · SKILL.md (reported line 22)May include surrounding context.

md
### Step 1 — Get your API key
Go to [kibi.bot/settings/api-keys](https://kibi.bot/settings/api-keys) → Create API Key → copy the `kb_...` key.

> **Permissions:** Base API is always included. Enable **Kibi LLM Gateway** if you want to use AI models. Enable **Agent Reload** if you want the agent to top up your Kibi Credits automatically from your trading wallet.

### Step 2 — Add Kibi Credits (for AI model access)
Go to [kibi.bot/credits](https://kibi.bot/credits) → Add Credit.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The reference documents /me and /balance/wallet as returning wallet addresses and profile-related data without a privacy or sensitive-data handling warning. In an agent setting, this increases the chance that an integrator or downstream agent exposes addresses in logs, chat transcripts, or third-party tools, enabling unnecessary data disclosure and profiling.

Content

No source excerpt is available for this finding.