AGENT SKILL

block-invisible-unicode

The mechanizable slice of prompt-injection defense, enforced at two points: at commit (the git hook, over staged changes) and at agent tool-use (over a tool call's arguments, as the agent writes) -- invisible and direction-override Unicode. Bidi controls make code read differently than it parses (Trojan Source, CVE-2021-42574); Unicode tag-block characters smuggle instructions that are invisible to a human reviewer but fully legible to the agent reading the file. Zero-width joiners and bidi marks (ZWJ/ZWNJ/LRM/RLM) are deliberately NOT matched -- they are legitimate in emoji sequences and in Persian, Arabic and Indic text -- so ordinary internationalised content passes; only the override/embed/isolate controls and the tag block, which have no honest use in a source tree, are blocked. Escape: 'pragma: allowlist invisible-unicode' on the same line.

License 许可
Apache-2.0
最近更新
2026-11-15
Stars
★ 10
Forks
1

01 前言 Front Matter

name
block-invisible-unicode
metadata
{ "chock.artifact": "hook", "chock.enforcement": "block", "chock.coverage_without_chock": "advisory" }
description
The mechanizable slice of prompt-injection defense, enforced at two points: at commit (the git hook, over staged changes) and at agent tool-use (over a tool call's arguments, as the agent writes) -- invisible and direction-override Unicode. Bidi controls make code read differently than it parses (Trojan Source, CVE-2021-42574); Unicode tag-block characters smuggle instructions that are invisible to a human reviewer but fully legible to the agent reading the file. Zero-width joiners and bidi marks (ZWJ/ZWNJ/LRM/RLM) are deliberately NOT matched -- they are legitimate in emoji sequences and in Persian, Arabic and Indic text -- so ordinary internationalised content passes; only the override/embed/isolate controls and the tag block, which have no honest use in a source tree, are blocked. Escape: 'pragma: allowlist invisible-unicode' on the same line.

03 SKILL.md 正文

Block Invisible Unicode

The mechanizable slice of prompt-injection defense, enforced at two points: at commit (the git hook, over staged changes) and at agent tool-use (over a tool call's arguments, as the agent writes) -- invisible and direction-override Unicode. Bidi controls make code read differently than it parses (Trojan Source, CVE-2021-42574); Unicode tag-block characters smuggle instructions that are invisible to a human reviewer but fully legible to the agent reading the file. Zero-width joiners and bidi marks (ZWJ/ZWNJ/LRM/RLM) are deliberately NOT matched -- they are legitimate in emoji sequences and in Persian, Arabic and Indic text -- so ordinary internationalised content passes; only the override/embed/isolate controls and the tag block, which have no honest use in a source tree, are blocked. Escape: 'pragma: allowlist invisible-unicode' on the same line.

on(commit|tool_use): block(content_regex) scan=added_lines allowlist_pragma=pragma:\s*allowlist\s+invisible-unicode ...
Invisible or direction-override Unicode detected in this change. These characters change how code reads to a human or hide instructions an agent will still obey. Remove them. 'pragma: allowlist invisible-unicode' on the same line marks a documented exception; in the agent (tool use, the turn's end) it counts only when that exact line is already committed in HEAD, so an agent asks a person rather than writing the pragma itself.

This skill is advisory: the client reading it has no mechanism to enforce it. The same policy compiled by chock blocks at commit, on an agent's file writes and at turn end. See https://github.com/open-coder-ai/chock

04 同仓库更多 Skills