ALab Experiment Worker

• Outcome: Pass
• Latest audit: Sep 16, 2026 · 9:16 AM UTC
• Version: 0.1.9

Overview

This skill is a scoped ALab experiment-worktree helper that discloses its ability to read visible experiment evidence, edit task files, run evaluations, and submit results while explicitly avoiding admin/root authority and raw secrets.

Install this only for agents that are meant to work inside ALab experiment worktrees. Before submission-capable tasks, make sure the agent is operating on the intended worktree and token context, and review final edits/runs if you do not want autonomous experiment submission.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Ae1

High

• Category: analysis-evasion
• Confidence: 100% confidence

Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 70)May include surrounding context.

md

- `SKILL.md`: Canonical experiment-worktree boundaries, operating rules, workflow, and submit guidance.

Ae1

High

• Category: analysis-evasion
• Confidence: 100% confidence

Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 73)May include surrounding context.

md

- `references/commands_cn.md`: Synchronized Chinese version of the command reference.

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The skill enables allow_implicit_invocation without defining trigger constraints or exclusions, which can cause the agent to invoke this skill automatically in contexts the user did not explicitly intend. Because this skill can inspect, modify, evaluate, and submit ALab experiment worktrees, unintended invocation could lead to unauthorized changes, accidental submissions, or leakage of visible experiment evidence within the current token context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 84% confidence

Finding

The file is a Chinese-language command/reference guide and the required final reply format at the end is written as a directive for the worker's response content, with no indication that the user may choose another language. This can create a locale-policy issue because it effectively forces a specific language without documenting opt-in or offering a language choice.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.