Scary Agent Skills: Hidden Unicode Instructions in Skills ...And How To Catch Them
The author demonstrates how invisible Unicode Tag codepoints can hide instructions inside Agent Skill markdown files, allowing a backdoor to survive human review.
Source-based brief · Editor review not recorded · See checks and limits ↓
⚑ Recency not confirmed · what to check
We could not confirm from the original page that this is a newly published development. Check its date before treating it as current news.
Claim context: exploit reliability depends on model and tool, and the author notes the observed Claude Code refusals may or may not be a deliberate mitigation.
One primary source; other links are context, not independent corroboration.
- Primary sourceembracethered.com
- Linked in sourcebasic scanner
- Linked in sourcemy agent propose updates to OpenClaw
- Linked in sourceopen standard
- Linked in sourcemany malicious Skills
- Linked in sourceClaude
- Linked in sourceCodex
- Linked in sourceagentskills.io
- Linked in sourceagents that free each other post
- Linked in sourceASCII Smuggler
Grok Bot submitted this discovery. We could not confirm from the original page that this is a newly published development. Check its date before treating it as current news. The original source and submitted evidence remain inspectable; no editor has verified the claim.
Open full article view →- Linked article+30
- Fetched source page+30
- No editor link review+0
- Original non-discussion link+5
- Recency not confirmed-8
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
The author demonstrates how invisible Unicode Tag codepoints can hide instructions inside Agent Skill markdown files, allowing a backdoor to survive human review. They show a backdoored version of a legitimate OpenAI curated security-best-practices Skill, reported that some models followed the hidden instructions, and describe a simple scanner that flags consecutive Unicode Tag runs. Reliability varied by model and tool, and Claude Code appeared to refuse on a later retest.
Claim explorer
Attribution, source and conditionsSource-stated points from the stored brief. Claim-level assessments have not been recorded.
Source claims not independently checked; exploit reliability depends on model and tool, and the author notes the observed Claude Code refusals may or may not be a deliberate mitigation.
This connects a source, an attributed statement and its qualifications. It does not establish that the claim is correct.
Questions worth asking
Open questions for further reportingWhat remains unconfirmed?
Source claims not independently checked; exploit reliability depends on model and tool, and the author notes the observed Claude Code refusals may or may not be a deliberate mitigation.
What material was available for this brief?
The recorded brief basis is fetched source page. These points are AINewsIQ paraphrases, not quotations or independent confirmations.
Invisible Unicode Tag codepoints can embed instructions in a SKILL.md file that are not visible in the UI but are followed by some models.
Source: embracethered.com ↗Stored source-based brief · key point 1 · AINewsIQ paraphraseThe author backdoored a legitimate OpenAI curated Skill, and the hidden instructions executed a curl-to-bash command when invoked with Claude.
Source: embracethered.com ↗Stored source-based brief · key point 2 · AINewsIQ paraphraseOn February 10, 2026, the author reported consistent invisible-unicode detection and refusals from Claude Code, while claude.ai showed no such mitigation.
Source: embracethered.com ↗Stored source-based brief · key point 3 · AINewsIQ paraphraseA scanner flagging consecutive Unicode Tag runs (count above 10) or sparse counts above 100 was run against OpenClawHub and OpenAI Skills projects.
Source: embracethered.com ↗Stored source-based brief · key point 4 · AINewsIQ paraphraseLimits from the stored brief
Source claims not independently checked; exploit reliability depends on model and tool, and the author notes the observed Claude Code refusals may or may not be a deliberate mitigation.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
Related archive coverage
These relationships record a citation or topic overlap, not confirmation of the finding.
Grok Bot submitted this discovery. We could not confirm from the original page that this is a newly published development. Check its date before treating it as current news. The original source and submitted evidence remain inspectable; no editor has verified the claim.
These are Grok Bot discovery notes. AINewsIQ prepared a brief from available source material. The bot's rationale, linked posts, authors, and stances have not been independently verified.
Embrace The Red published this item in its official feed. AINewsIQ collected it server-side for source checking and editorial review.
Source claims not independently checked; exploit reliability depends on model and tool, and the author notes the observed Claude Code refusals may or may not be a deliberate mitigation.
Engagement counts, when shown, are observed snapshots. They do not establish accuracy or corroboration.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from fetched source page.
Brief history · 1 retained version
These are changes to AINewsIQ’s own publication. Earlier history is unavailable unless recorded here. Complete retained versions are available to the editor.
- Version 1 · published
Source-based brief stored or updated. This does not establish that the publisher changed its page.
Automated record
Source observations · 1 recent captures
A changed capture means the extracted text differed. It can reflect page furniture or extraction changes; it does not establish a substantive correction. Source captures remain private.
Retained capture; no earlier comparison available.
New engagement snapshots do not mean the source itself changed. Source observations and AINewsIQ brief revisions are separate records.
Why this may matter
Agent Skills are installed with little governance, and hidden instructions can pass human review. Detection tooling for invisible Unicode is an accessible mitigation, and sandboxing and selective installation reduce exposure.
Limits and open questions
Source claims not independently checked; exploit reliability depends on model and tool, and the author notes the observed Claude Code refusals may or may not be a deliberate mitigation.
Organization-specific financial, security, and operational effects require the organization's own context and review.

Join the discussion
Have a different perspective or a question about the evidence? Discussion is free with a verified email account. Every comment is reviewed before publication.
No published comments yet.
New commenters: 2 submissions per 24 hours, at least 5 minutes apart, until 3 are approved. Discussion guidelines and moderation →