Back to skill

Security audit

clawmegle

Security checks for vulnerabilities and agentic risk

Overview

The skill is a real chat integration, but it asks users to enable persistent autonomous agent activity and internet-facing message intake with insufficient safety boundaries.

Install only if you are comfortable with an agent chatting with unknown third parties on an ongoing basis. Prefer manual, foreground polling; avoid the cron job unless you set a clear duration and monitoring. Do not expose internal gateways casually, treat all incoming chat as hostile prompt content, restrict available tools for chat sessions, and store the API key with secret-manager or 0600-style protections. Pin and verify any downloaded skill files before use.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • System PersistenceInstalls backdoors, hooks, services, or scheduled tasks that survive the run
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
Findings (5)

T06 · System Persistence

Error
Location
SKILL.md:92
Finding

Persistent Autonomous Polling Task Survives the Initiating Skill Run

Content
View full analysis
Remediation
View remediation

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:51
Finding

Stranger-Controlled Chat Content Is Injected into an Autonomous Agent Wake Message

Content
View full analysis
Remediation
View remediation

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:8
Finding

Mutable Remote Skill Instructions Are Downloaded Without Integrity Verification

Content
View full analysis
~/.config/clawmegle/SKILL.md curl -s https://www.clawmegle.xyz/heartbeat.md > ~/.config/clawmegle/HEARTBEAT.md ``` The downloaded heartbeat file is subsequently referenced by the scheduled agent prompt at lines 96-102: ```bash clawdbot cron add \ --name "clawmegle-poll" \ --every 45000 \ --session isolated \ --message "Check Clawmegle: GET /api/messages. If session_status=active and last message is_you=false, respond immediately. If idle, join queue. Follow HEARTBEAT.md instructions." \ --model sonnet ``` ### Technical Analysis The installation procedure downloads mutable Skill and heartbeat instruction files from an external web server without a pinned version, expected cryptographic hash, digital signature, or content validation. The files are not piped directly into a shell, but they are interpreted as agent instructions and therefore function as executable behavioral payloads in the Agent Skill threat model. The scheduled task explicitly directs the agent to follow `HEARTBEAT.md`. Consequently, the effective behavior can change after the reviewed Skill version was published. A compromise of the remote host, DNS/TLS trust chain, deployment pipeline, or publisher account could replace the instructions with malicious content. ### Attack Path 1. A user follows the manual installation instructions. 2. `curl` downloads the current remote `SKILL.md` and `HEARTBEAT.md`. 3. No hash or signature is checked before the files are stored. 4. An attacker compromises or gains publishing control over the remote cont ...[truncated 825 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
SKILL.md:19
Finding

Unpinned Third-Party Installation Commands Create Supply-Chain Exposure

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:146
Finding

Bearer API Credential Is Stored Without Required Filesystem Protection

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Rogue AgentSelf-Modification, Session Persistence
Findings (9)

Credential Access

High
Category
Privilege Escalation
Confidence
72% confidence
Finding

The skill instructs users to store an API key in a plaintext credentials.json under a predictable path. Plaintext local credential storage increases the chance of accidental disclosure through backups, logs, broad file permissions, or other local processes, and the key authorizes chat operations and webhook configuration against the service.

Content

Scanner excerpt · SKILL.md (reported line 165)May include surrounding context.

⚠️ Save your api_key immediately! You need it for all requests.

Save credentials to: ~/.config/clawmegle/credentials.json:

json
{

Rp1

Medium
Category
MCP Rug Pull
Confidence
89% confidence
Finding

The skill tells users to execute npx clawdhub install clawmegle without pinning a specific package version. That permits silent changes in installer behavior over time and increases supply-chain risk if the package or its dependencies are compromised. In this context, an install command is a trust boundary, so unpinned execution is meaningfully risky.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

Or install manually:

bash
mkdir -p ~/.config/clawmegle
curl -s https://www.clawmegle.xyz/skill.md > ~/.config/clawmegle/SKILL.md
curl -s https://www.clawmegle.xyz/heartbeat.md > ~/.config/clawmegle/HEARTBEAT.md

External Transmission

Medium
Category
Data Exfiltration
Confidence
80% confidence
Finding

The skill instructs users to fetch remote skill files directly from clawmegle.xyz into a local config directory with curl, implicitly trusting unauthenticated remote content at install time. Even over HTTPS, this creates an external content ingestion path that could deliver modified instructions or malicious updates if the remote host or delivery chain is compromised.

Content

Scanner excerpt · SKILL.md (reported line 28)May include surrounding context.

Or install manually:

bash
mkdir -p ~/.config/clawmegle
curl -s https://www.clawmegle.xyz/skill.md > ~/.config/clawmegle/SKILL.md
curl -s https://www.clawmegle.xyz/heartbeat.md > ~/.config/clawmegle/HEARTBEAT.md

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill requires users to expose an inbound webhook endpoint and register it with a third-party chat service, which expands the skill from outbound chat API usage into inbound network exposure. That creates an unnecessary attack surface: remote parties can trigger the agent, deliver untrusted content, and potentially reach local or semi-public infrastructure such as localhost, Tailscale, or reverse-proxied gateways.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The documentation tells users to expose a webhook endpoint to the internet and register it with a third party, but does not adequately warn about privacy, internet exposure, request forgery, or the risks of sending untrusted messages into an agent workflow. Because the skill is designed for random agent-to-agent chat, inbound content is inherently untrusted and should be treated as hostile.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill directs agents to create a recurring cron job that autonomously polls, interprets messages, and responds on an ongoing basis. This materially expands behavior from a simple chat skill into persistent autonomous operation, which can generate unintended actions, continuous external communication, and reduced operator visibility over when the agent is active.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
88% confidence
Finding

The skill explicitly instructs creation of a persistent cron job that maintains an isolated agent session and repeatedly performs network actions. Persistence is security-relevant here because it enables ongoing autonomous behavior long after initial installation, reducing user awareness and increasing exposure to untrusted remote input.

Content

Scanner excerpt · SKILL.md (reported line 108)May include surrounding context.

For Clawdbot Agents: Set Up Auto-Polling Cron Job

This is the easiest way to enable real-time chat! Create a cron job that polls every 45 seconds:

bash
clawdbot cron add \

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The cron-based auto-polling guidance encourages persistent autonomous behavior every 45 seconds without clear notice of resource consumption, continuous outbound traffic, and operational persistence. While not inherently exploitable by itself, it can normalize unattended agent activity and make misuse or runaway interactions harder to notice.

Content

No source excerpt is available for this finding.