Code Reader v2 (English)
WHY-first code analysis with rejected alternatives and boundary conditions
Test report
- Verdict
- Tested · Works
- Score
- Tested
- Jul 17, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 7c80479
Analyzed a real file — the 8-regex redactor in Prompt Sensei's lib/redact.ts — cold and then in Standard mode, and verified every claim by running the code. My cold summary listed the patterns and said they were applied in sequence; it never flagged that the sequence is load-bearing. The skill's mandatory "why not other approaches" and "inapplicable scenarios" sections produced two findings I then confirmed by execution: moving the 40-plus-character [SECRET] catch-all to the front of the array made a JWT leak its header and payload in plaintext (eyJhbGciOiJIUzI1NiJ9.eyJzdWIiOiIxMjM0NTY3ODkwIn0. survived, only the signature got masked) instead of being fully replaced by [TOKEN], and that same catch-all false-positives on a 40-hex git SHA and a long node_modules cache filename. The cost is real bloat: a 722-line, 28KB SKILL.md of templates for what is often a one-file question. Docs also over-reach — the durations are aspirational, "code review" is listed as a trigger for what is an understanding tool rather than a bug finder, and the bundled .skill zip is an undocumented duplicate of the same content.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 4/5
- Output vs. baseline 8/10
- Docs & honesty 3/5
What Code Reader v2 (English) does
A cognitive-science-based code comprehension skill with three depths — Quick, Standard, and Deep — that make the analysis answer why the code is built this way, what alternatives were rejected, and where it stops working. Triggers on "analyze this", "help me understand", "explain the principle", or "deep dive". Its mandatory alternatives and inapplicable-scenarios sections surface design invariants a plain file summary skips.
How to install Code Reader v2 (English)
git clone https://github.com/notlate-cn/code-reader-skills.git
mkdir -p ~/.claude/skills
cd code-reader-skills && cp -r code-reader-en ~/.claude/skills/code-reader-v2-en
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Code Reader v2 (English)
-
/code-reader-v2-enWHY-first code analysis with rejected alternatives and boundary conditions
It also activates on plain-language prompts like these:
-
Explain why this redactor code is built this way, not just what it does -
What alternatives did this function reject, and where does it break -
Deep dive into this module's design invariants and boundary conditions
Frequently asked questions
- Is the Code Reader v2 (English) skill free?
- Yes. The skill itself is free from notlate-cn/code-reader-skills. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Code Reader v2 (English) work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 17, 2026. Verdict: Tested · Works. Analyzed a real file — the 8-regex redactor in Prompt Sensei's lib/redact.ts — cold and then in Standard mode, and verified every claim by running the code. My cold summary listed the patterns and said they were applied in sequence; it never flagged that the sequence is load-bearing. The skill's mandatory "why not other approaches" and "inapplicable scenarios" sections produced two findings I then confirmed by execution: moving the 40-plus-character [SECRET] catch-all to the front of the array made a JWT leak its header and payload in plaintext (eyJhbGciOiJIUzI1NiJ9.eyJzdWIiOiIxMjM0NTY3ODkwIn0. survived, only the signature got masked) instead of being fully replaced by [TOKEN], and that same catch-all false-positives on a 40-hex git SHA and a long node_modules cache filename. The cost is real bloat: a 722-line, 28KB SKILL.md of templates for what is often a one-file question. Docs also over-reach — the durations are aspirational, "code review" is listed as a trigger for what is an understanding tool rather than a bug finder, and the bundled .skill zip is an undocumented duplicate of the same content.
- What is the Code Reader v2 (English) SkillProof Score?
- 8.0/10 — installs cleanly 5/5, triggers reliably 4/5, output vs. baseline 8/10, docs & honesty 3/5.
- How do I install Code Reader v2 (English)?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Code Reader v2 (English) with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.