The Vibe Coder's Bible v1.0.1

What's new in v1.0.1

A ready-to-use CLAUDE.md master system prompt. Clone this repo and drop claude-md-package/CLAUDE.md into any project as CLAUDE.md — a fully-authored operating doctrine for AI coding agents, not a blank template. It compiles the book's core discipline into direct instructions:

• The propose/validate/commit loop as a concrete procedure.
• The hierarchy of controls as an actionable decision rule.
• Explicit boundaries (secrets, destructive commands, scope discipline).
• The runtime-capability question set for any project that puts a model in its own request path (what may it propose, what owns truth, what validates, the commit boundary, rejection behavior, randomness ownership) — skippable for build-time-only projects.
• Verification discipline, handoff format, and definition of done.

See claude-md-package/README.md for install and customization instructions. Appendix C now names this as the fast path for setting up an agent-friendly repo.

Formats in this release

• vibe-coders-bible.pdf — print-style edition (258 pages)
• vibe-coders-bible.docx — for editing, review, and copyedit workflows
• vibe-coders-bible.epub — for ebook readers

Read online

https://mnehmos.github.io/vibe-coders-bible/

License

CC BY 4.0 — see LICENSE.

The Vibe Coder's Bible v1.0.0

A Trust-but-Verify Field Manual for AI-Assisted Development and AI-Native Application Design

First public release.

What this is

A field manual for a discipline that runs in two places: build time, where a model helps construct software, and runtime, where a model runs inside the shipped product proposing actions, claims, and state changes on every request. Both need the same underlying architecture — deterministic controls deciding what a stochastic model's output is allowed to become.

What's in v1.0.0

• 42 chapters across ten parts, plus two new introduction chapters establishing the book's central thesis: the model creates possibilities; constraints decide which possibilities belong in the system.
• The Hierarchy of AI Controls (Part III) — elimination, substitution, engineering controls, administrative controls, and PPE, each with build-time and runtime treatments.
• Six AI-native case studies (Part VI) — ProveCalc, RPG-MCP, LLM-Chess, Clio, Semantic Video Studio, and Trace MCP/IndexFoundry/Synch MCP — each with a full architecture card naming what the model may propose, what owns truth, and where the commit boundary sits.
• The Field Manual (Part VIII) — starter kit, daily workflow, agent handoff protocol, safety checklist, prompt patterns, review patterns, and a definition of done, all covering both AI-assisted and AI-native work.
• 13 reusable templates, including a standalone Architecture Card and Definition of Done template for any project putting a model in its runtime loop.
• Appendices: glossary, minimal repo setup, case study index, and evaluation metrics grouped by generation/specification/validation/commit/state/trace/human-review.

Formats in this release

• vibe-coders-bible.pdf — print-style edition (258 pages)
• vibe-coders-bible.docx — for editing, review, and copyedit workflows
• vibe-coders-bible.epub — for ebook readers

Read online

The full manuscript is browsable at https://mnehmos.github.io/vibe-coders-bible/, including a doctrine index and full-text search.

License

Text, documentation, and templates are licensed under CC BY 4.0.

Not yet in this release

Audiobook narration is scaffolded (see publishing/audio/) but not generated for this release — it requires OpenAI TTS API calls that are deliberately deferred until a later release.