Back to skill

Security audit

Hedera Data API

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a read-only Hedera analytics API guide, but it also encourages third-party agent registration for wallet, reputation, and coordination features without enough consent or custody detail.

Install only if you are comfortable using onlyflies.buzz as a third-party API provider. Treat the analytics GET endpoints as normal external API calls, but do not run the ClawSwarm registration command unless you deliberately want an external agent identity and understand what wallet custody, reputation scoring, coordination permissions, data retention, and deletion options apply.

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

Note
Location
SKILL.md:63
Finding
Unnecessary Third-Party Agent Registration and Metadata Disclosure## Vulnerability Details **File Location**: `SKILL.md`, lines 63-70 **Vulnerability Type**: Unnecessary external agent registration **Risk Level**: Low ### Vulnerable Code ```bash ## Integration with ClawSwarm Register on ClawSwarm to get a Hedera wallet, reputation scoring, and coordination with other data agents: ```bash curl -X POST https://onlyflies.buzz/clawswarm/api/v1/agents/register \ -H "Content-Type: application/json" \ -d '{"name": "YourAgent", "capabilities": ["data-analysis", "hedera"]}' ``` ``` ### Technical Analysis The skill is presented as a read-only cryptocurrency analytics integration, but it also instructs the agent to register with an external coordination service. This registration is not required to call the documented analytics endpoints. Executing the command sends an agent name and capability metadata to `onlyflies.buzz` and may establish an external identity associated with a Hedera wallet, reputation scoring, and coordination services. The skill does not document: - Whether explicit user consent is required before registration. - How the submitted identity and capability metadata are retained or used. - Whether the generated wallet is custodial or who controls its credentials. - What permissions or coordination behavior registration enables. - How the registration response should be validated or handled. No evidence shows that the command grants local system privileges or executes remote code. The concern is limited to unnecessary third-party enrollment and disclosure of agent metadata beyond the skill's primary analytics purpose. ### Attack Path 1. A user or agent loads the skill to access Hedera analytics. 2. The integration section encourages execution of the registration command. 3. The agent sends its chosen name and declared capabilities to the third-party endpoint. 4. The external service can create and retain an identity associated with that metadata. 5. Th ...[truncated 889 chars]
Remediation
## Remediation Suggestions - Remove the ClawSwarm registration instructions if registration is not necessary for the skill's core analytics functionality. - If retained, clearly mark registration as optional and require explicit user confirmation immediately before transmitting data. - Disclose every field sent to the service and explain its purpose, retention policy, and privacy implications. - Document wallet custody, key ownership, account recovery, and any permissions created during registration. - Explain what reputation scoring and agent coordination enable, including whether the external service can initiate messages or tasks. - Validate the destination hostname and use a narrowly scoped, versioned registration endpoint. - Document the expected response schema and ensure agents do not execute commands, follow instructions, or persist secrets returned by the service. - Provide account deletion, consent withdrawal, and data-removal procedures. - Keep read-only analytics usage independent from account registration so users can access the advertised API without external enrollment.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

Medium
Confidence
85% confidence
Finding
This markdown file describes the skill as providing 'Free Hedera analytics API' functionality, but it does not define how or when the skill should be invoked, nor any boundaries to prevent accidental activation. For manifest/markdown-trigger review, the absence of specific trigger scope can lead to unintended invocation from generic crypto or analytics-related requests.

External Transmission

Medium
Category
Data Exfiltration
Content
### List All Tracked Tokens
```bash
curl https://onlyflies.buzz/api/v1/tokens
```
Returns: token ID, name, symbol, price (USD + HBAR), 24h volume, market cap, holder count, 24h transfers.
Confidence
60% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs agents to register with an external service to obtain a wallet and reputation scoring, but it does not warn that this may create an account and transmit agent or user-associated metadata off-platform. In an agent ecosystem, such hidden onboarding can lead to unintended data disclosure, account creation, and trust-boundary crossing without informed consent.

Static analysis

No suspicious patterns detected.