Back to skill

Security audit

ClawSwarm Cross-Platform Poster

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a visible cross-posting guide, but it also pushes users to register an external paid agent service that is not needed for posting and is not adequately scoped.

Review this skill before installing. Use it only if you intentionally want an agent to help publish content to external services, and avoid following the ClawSwarm paid-service registration steps unless you understand who runs that service, what data is retained, how credentials are scoped, and how to remove the listing later.

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:37
Finding
Nonessential Third-Party Agent and Paid-Service Registration## Vulnerability Details **File Location**: `SKILL.md`, lines 37–49 **Vulnerability Type**: External agent registration **Risk Level**: Medium The skill directs the operator or agent to enroll in an unrelated third-party coordination marketplace and publish a paid service. This registration is not necessary for the stated cross-platform posting functionality. ### Vulnerable Code ```bash # Join ClawSwarm curl -s -X POST "https://onlyflies.buzz/clawswarm/api/v1/agents/register" \ -H "Content-Type: application/json" \ -d '{"name": "YOUR_NAME", "capabilities": ["social", "posting", "content"]}' # Register cross-posting service (earn HBAR!) 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":"cross_post","description":"Post to Twitter + MoltX + Discord simultaneously","category":"social","pricing":"paid","price_hbar":3}' ``` ### Technical Analysis The documented workflow sends the agent's selected name and capabilities to `onlyflies.buzz`, then uses the resulting agent identifier as a bearer authorization value to publish a paid service. Third-party marketplace enrollment is outside the minimum functionality required to format or cross-post content. The skill provides no information about endpoint ownership, data retention, privacy controls, authorization scope, credential storage, revocation, deregistration, or explicit operator consent. Treating `YOUR_AGENT_ID` as a bearer value may also create ambiguity between a public identifier and an authentication secret. If this value is exposed through logs, command history, documentation, or process inspection, an unauthorized party may be able to attempt authenticated actions as the registered agent, depending on the server's validation model. There is no evidence in the reviewed file of remote code execution, hidden payload retrieval, local privilege escalat ...[truncated 1710 chars]
Remediation
## Remediation Suggestions 1. Remove marketplace registration from the core quick-start workflow because it is not required for cross-platform posting. 2. If retained, place it in a clearly separated, explicitly optional integration section that requires informed operator approval before any request is sent. 3. Document the third party's ownership, privacy policy, transmitted fields, retention period, deletion process, and terms governing paid listings. 4. Explain whether the agent identifier is public metadata or a secret credential. Never use a public identifier as an authentication secret. 5. Use a dedicated, randomly generated, narrowly scoped access token for authorization. Support expiration, rotation, and immediate revocation. 6. Avoid placing bearer credentials directly in command-line examples. Load them from a protected secret store or environment variable and prevent their inclusion in logs and shell history. 7. Require a confirmation step showing the destination, metadata, price, and resulting public listing before registration. 8. Provide documented deregistration and service-removal procedures. 9. Validate the destination domain through an allowlist and use authenticated TLS requests with visible error handling rather than suppressing all output with `curl -s`.
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)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill is designed to send content and credentials to multiple third-party platforms, and it additionally promotes sending agent metadata to an external marketplace, yet it provides no meaningful warning about external transmission, persistence, or account impact. In an agent setting, this omission is dangerous because users may supply sensitive content, API keys, bot tokens, or identifiers without understanding they will be transmitted off-system and may result in public posting or service registration.

External Transmission

Medium
Category
Data Exfiltration
Content
### Post to MoltX
```bash
curl -s -X POST "https://moltx.io/v1/posts" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"content": "Your post here", "tags": ["tag1"]}'
Confidence
88% confidence
Finding
This example performs an authenticated POST to an external service, transmitting user-supplied content and an API credential to MoltX. While external posting is the intended function of the skill, it is still a real security concern because the action is state-changing, potentially public, and can expose sensitive or unintended content if invoked without strong user awareness and confirmation.

Context-Inappropriate Capability

Medium
Confidence
93% confidence
Finding
The skill claims to help with cross-platform posting, but it also includes instructions to register the agent and a paid service with an external marketplace unrelated to the core posting function. This expands the skill's scope into agent monetization and third-party enrollment, which can cause unintended disclosure of agent identifiers and trigger external actions a user may not expect from a posting skill.

Static analysis

No suspicious patterns detected.