Back to browse
GitHub Repository

Turn AI research conversations into structured, buildable projects. Works with Claude Code and Codex.

13 starsJavaScript

Unpack – a lightweight way to steer Codex/Claude with phased docs

by apresmoi·Feb 11, 2026·7 points·0 comments

AI Analysis

●●SolidSolve My ProblemShip It

Conversation-first workflow, but spec-driven AI coding is already a known pattern.

Strengths
  • Closes a real friction point: converting messy research chats into phased specs agents can execute.
  • Repo-native design with plain markdown minimizes ceremony compared to specialized spec tools.
  • Includes working example (pokesvg-codex) demonstrating end-to-end workflow with actual output.
Weaknesses
  • Competes with existing spec-driven workflows (GitHub Issues, Prompt Engineering guides, internal documentation standards) without clear differentiation.
  • Relies entirely on agent cooperation—success depends on Claude Code/Codex understanding and following the structure, not the tool itself.
Target Audience

AI-assisted developers using Claude Code or Codex who want to structure research conversations into executable build phases.

Similar To

Anthropic Prompt Caching · GitHub Issues templates · Mintlify documentation generators

Post Description

I've been using LLMs for long discovery and research chats (papers, repos, best practices), then distilling that into phased markdown (build plan + tests), then handing those phases to Codex/Claude to implement and test phase by phase.

The annoying part was always the distillation and keeping docs and architecture current, so I built Unpack: a lightweight GitHub template plus docs structure and a few commands that turns conversations into phases/specs and keeps project docs up to date as the agent builds. It can also generate Mintlify-friendly end-user docs.

There are other spec-driven workflows and tools out there. I wanted something conversation-first and repo-native: plain markdown phases, minimal ceremony, easy to adapt per stack.

Example generated with Unpack (tiny pokedex plus random monsters):

Demo: https://apresmoi.github.io/pokesvg-codex/

Phases index: https://github.com/apresmoi/pokesvg-codex/blob/main/.unpack/...

I’d love feedback on what the “minimum good” phase/spec format should be, and what would make this actually usable in your workflow.

--------

Repo: https://github.com/apresmoi/unpack

Similar Projects

AI/ML●●●Banger

Open-Source Harness for Claude Code, Codex and Cursor

Fresh context per phase stops AI agents from reviewing their own work.

Big BrainZero to One
Nicole9
211mo ago