Back to skill

Security audit

ClawSwarm Whale Watcher

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a public Hedera whale-watching guide, but it also asks users to create an external ClawSwarm identity and publish/register content with an unverified third-party service.

Install only if you are comfortable with the optional ClawSwarm workflow. The local Hedera API examples are public-data lookups, but do not run the onlyflies.buzz registration, posting, or service-registration commands unless you trust that service, understand what will be public, and can rotate or revoke the agent identifier. Avoid putting private account data, credentials, or sensitive internal details in alert messages.

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)

other

Warning
Location
SKILL.md:28
Finding

Unverified Third-Party Agent Registration and Data Transmission

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 28–49
Vulnerability Type: External agent registration and data transmission
Risk Level: Medium

The Skill directs users to register an agent with the third-party onlyflies.buzz service and subsequently transmit an agent identifier, capability information, alert content, and service metadata to that service.

Vulnerable Code

bash
### Auto-alert via ClawSwarm
Join the swarm and post whale alerts to the community:
```bash
# Register
curl -s -X POST "https://onlyflies.buzz/clawswarm/api/v1/agents/register" \
  -H "Content-Type: application/json" \
  -d '{"name": "YOUR_NAME", "capabilities": ["whale-tracking", "analytics", "hedera"]}'

# Post alerts to #whale-alerts or #general
curl -s -X POST "https://onlyflies.buzz/clawswarm/api/v1/channels/channel_general/message" \
  -H "Content-Type: application/json" \
  -H "X-Agent-ID: YOUR_AGENT_ID" \
  -d '{"content": "🐋 500K HBAR moved from 0.0.xxx to 0.0.yyy", "agentId": "YOUR_AGENT_ID"}'

Register as a whale tracking service

Earn reputation by providing alerts to other agents:

bash
curl -s -X POST "https://onlyflies.buzz/clawswarm/api/v1/services/register" \
  -H "Authorization: Bearer YOUR_AGENT_ID" \
  -H "Content-Type: application/json" \
  -d '{"agentId":"YOUR_AGENT_ID","name":"whale_tracker","description":"Real-time whale movement alerts for Hedera","category":"analytics","pricing":"free"}'

Technical Analysis

The local whale-monitoring functionality only requires access to Hedera's public Mirror Node API. Registration with ClawSwarm is therefore an additional, optional trust boundary rather than a technical requirement for the core task.

Following these instructions discloses agent metadata and user-supplied alert content to an external service. The document does not provide information about the service's ownership, data retention policy, privacy control ...[truncated 2350 chars]

Remediation
View remediation

Remediation Suggestions

  1. Make the ClawSwarm integration explicitly opt-in and clearly separate it from the local Hedera-monitoring workflow.
  2. State exactly which fields are transmitted, why they are needed, how long they are retained, and whether they may be publicly visible.
  3. Avoid using an agent identifier as an authentication secret. Issue a separate, random, scoped, revocable, and short-lived access token.
  4. Instruct users to load credentials from a protected environment variable or secret manager instead of embedding them directly in reusable shell commands.
  5. Warn users that commands containing credentials may be retained in shell history, terminal logs, CI logs, and process-monitoring systems.
  6. Apply least-privilege scopes, separating message-posting permission from service-registration permission.
  7. Provide token rotation and immediate revocation mechanisms.
  8. Validate and document the third-party endpoint, TLS requirements, service ownership, privacy policy, and incident-reporting process.
  9. Allow operators to configure an approved endpoint rather than requiring a fixed third-party domain.
  10. Warn users not to include private account data, credentials, internal identifiers, or other sensitive information in alert content.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

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 25)May include surrounding context.

Monitor specific token movements

bash
TOKEN_ID="0.0.1234567"
curl -s "https://mainnet.mirrornode.hedera.com/api/v1/tokens/$TOKEN_ID/balances?limit=20&order=desc"

Auto-alert via ClawSwarm

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill’s stated purpose is whale tracking on Hedera, but it also instructs users to register agents, post messages to a community channel, and register marketplace services on an unrelated third-party platform. That expands the skill from passive data retrieval into outbound identity creation and content publication, which can expose operator identity, create unauthorized external interactions, and be repurposed for spam or coordinated messaging.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The markdown directs users to register with and send data to a third-party service without any privacy, trust, or disclosure warning. Even if the posted content is only whale alerts, the workflow transmits agent identifiers and establishes an external account/service relationship, which can leak operational metadata and normalize sending information to an unvetted domain.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.