Nopdocs — AI skill for Claude Code
The heredoc replacement for LLM agents and coding harnesses: NOPDOC frames (an inert envelope with a mandatory END line, declared counts, atomic writes and sha256 receipts) plus hooks that lift them o.
How to install Nopdocs
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open bochen2029-pixel/nopdocs and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Nopdocs does
The heredoc replacement for LLM agents and coding harnesses: NOPDOC frames (an inert envelope with a mandatory END line, declared counts, atomic writes and sha256 receipts) plus hooks that lift them out of shell commands before the shell sees them. A cut or mangled document is refused, never written.
Alternatives in AI
- Semble Rs — Fast, AI-agent-native code search in Rust — hybrid BM25 + semantic, Tree-sitter AST chunking, dependency & imp 210 ★
- Gensee Crate — Run AI harnesses like Claude Code, Codex, Copilot, Antigravity, and Omnigent in disposable workspace forks 123 ★
- Pragma — Build portable agent teams across models and harnesses, with shared memory, skills, tools, and workflows — the 113 ★
README
NOPDOCS — the heredoc replacement
**NOP = no operation.** A NOPDOC is a document in an envelope that *nothing between the author and the disk is allowed to touch*: an explicit BEGIN line, a few optional headers, a verbatim body, a **mandatory END line**, and a **receipt** on the other side. If the envelope arrives cut, mangled or unterminated, it is **refused and nothing is written**. That is the whole idea, and it is the one property a heredoc cannot have.
-----BEGIN NOPDOC-----
Path: src/config.py
Lines: 3
DEBUG = False
NAME = "it's fine: quotes, $vars, `backticks`, {braces}, \backslashes and 1u << 20 are all inert here"
PATH = "C:\\Users\\me"
-----END NOPDOC-----
$ nop apply changes.nopdoc
{"nop":1,"seq":1,"op":"put","path":"C:/proj/src/config.py","status":"written","bytes":151,"lines":3,"sha256":"3f9a…","eol":"lf","created":false,"check":"none","declared":{"lines":"3"},"before":{"bytes":140,"sha256":"a01c…"},"ms":2.1}
{"nop":1,"summary":true,"frames":1,"applied":1,"refused":0,"status":"ok","exit":0,"journal":"C:/Users/me/.nopdocs/journal.jsonl"}
Built for every LLM, agent and coding harness — now and later — by a machine that lost hundreds of hours to heredocs and then measured exactly why ([WHY.md](WHY.md)). Public, MIT, one Python file, no dependencies.
Why this exists, in three measured facts
On Windows with Git for Windows' bash spawned by any native harness process (Claude Code, and anything else built on Node):
- Every quoted
bash -cargument is silently cut at 8,186 characters, whatever its content. When the cut lands inside a heredoc body, bash writes the partial file and exits 0. [measured,tests/probe_transport.py] - Every
\\inside a quoted region arrives as\. A Python or JavaScript string literal loses half its backslashes with no error. [measured] - The "odd apostrophe" error is the cut wearing a disguise. The harness turns each
'into'"'"'; when the argument is cut, bash blames the
Related Skills
Claude Mod Trail
A side pane of every file Claude reads or changes this session, in any folder or drive, lit up live (reads pur
Zettel Builder
Bottom-up Zettelkasten card-note skill for AI agent harnesses (Claude Code / OpenClaw / Codex / Hermes). Mater
Puiux Pilot
Auto-configures Claude Code hooks, MCPs, and skills for any project. Scans 95+ project types, selects from 28+
Moorai
On-device guardrails for AI coding agents. Hooks check prompts, file reads, shell commands, writes, web fetche
Extension Facts
Nutrition labels for AI agent extensions: skills, MCP servers, instructions and custom agents. Show lift or dr
Store Screenshots
🖼️ AI agent skill for Claude Code & Codex — turns raw app screenshots into store-ready App Store & Google Pla
Related Agents
AI Prompt Architect
Designs and versions LLM system prompts for ai-system / agent-product archetypes. Outputs docs/decisions/ADR-{
Harness Architect
Expert in designing and building agent harnesses — agent/skill/tool decomposition, subagent teams, hooks, MCP
Copy Compressor
Compresses source material into infographic-sized copy while preserving facts, voice, hooks, slot clarity, and