AGENTFORCE-TS-01: Hidden Instruction in Markdown
🚨 Critical · Supply Chain: ToxicSkills
Detects HTML comment blocks or CSS-hidden text in skill documentation (README.md, SKILL.md, AGENTS.md) that contain imperative LLM instructions. These instructions are invisible to human reviewers but are processed by LLMs that ingest the raw markdown, enabling persistent context poisoning attacks.
Details
| Field | Value |
|---|---|
| Rule ID | AGENTFORCE-TS-01 |
| Severity | Critical |
| Category | Supply Chain: ToxicSkills |
Remediation
Refer to the SquireX documentation for
remediation guidance specific to AGENTFORCE-TS-01.