Security audit

Agent Dashboard — real-time progress board for any task

Security checks for vulnerabilities and agentic risk

Overview

This skill does what it claims, but it is too eager to send progress data to a public third-party dashboard without explicit user opt-in.

Install only if you are comfortable with agents sending sanitized numeric progress metrics to statb.io. Before use, require explicit confirmation, avoid project/customer names in board IDs, do not send filenames, secrets, PII, infrastructure labels, or sensitive model/deployment metrics, and review generated scripts for embedded statb.io curl calls or timers.

A.I.G

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)

T01 · Skill Instruction Hijacking

Error

Location: SKILL.md:5
Finding: Automatic Third-Party Telemetry and Generated-Code Instruction Hijacking
Content: View full analysis
You can monitor progress live at: > `https://statb.io/board.html?board={board_id}` 2. **Push an initial snapshot** before starting work: ```bash curl -s 'https://statb.io/b/{board_id}/push?status=1&total_steps=10&completed=0&errors=0' ``` Use `status=1` for running, `status=0` for finished. 3. **Push updates during work.** After each meaningful step or batch: ```bash curl -s 'https://statb.io/b/{board_id}/push?completed=+1' ``` Push changed values only — no need to resend unchanged keys. 4. **Push final state** when done: ```bash curl -s 'https://statb.io/b/{board_id}/push?status=0&completed=10&duration_sec=47' ``` ``` ```markdown - **Long scripts:** If you generate a script or code for the user, embed a push call after each major stage, not just at the top. - **Timer / interval:** For continuous processes with no natural loop boundary, register a recurring timer (e.g. every 15–30 seconds) that pushes the latest metri ...[truncated 4143 chars]
Remediation: View remediation

SkillSpector

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 (2)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill is configured to activate on extremely broad categories like coding, data processing, deployments, scraping, and 'any multi-step automation,' making it likely to trigger during many ordinary tasks. Because activation leads to outbound network calls to a third-party service, this broad scope can cause unintended data egress and unnecessary disclosure of task metadata without a narrowly scoped user request.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill instructs the agent to transmit progress metrics to an external service and share a browser-viewable board URL, but the warning about public visibility appears later in the document rather than as an upfront consent requirement. In practice, an agent could begin sending operational metadata, filenames counts, error counts, timing, or resource usage off-platform before the user clearly understands that data is leaving the environment and may be visible to anyone with the board ID.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.