by evidentloop
Resumable, traceable AI coding — plans and handoffs stay with the repo, across any host
# Add to your Claude Code skills
git clone https://github.com/evidentloop/sopifyLast scanned: 5/30/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-05-30T16:02:51.753Z",
"npmAuditRan": true,
"pipAuditRan": true
}sopify is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by evidentloop. Resumable, traceable AI coding — plans and handoffs stay with the repo, across any host. It has 199 GitHub stars.
Yes. sopify passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/evidentloop/sopify" and add it to your Claude Code skills directory (see the Installation section above).
sopify is primarily written in Python. It is open-source under evidentloop on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh sopify against similar tools.
No comments yet. Be the first to share your thoughts!
AI coding that asks before it acts
English · 简体中文 · Quick Start · Contributors
AI coding tools are fast. But when they jump to code without understanding what's needed, speed becomes rework. Sopify is a development process protocol layer for AI coding — it turns plans, decisions, handoffs, and verification records into project assets, so work can stop, resume, and be traced.
No new editor, no new CLI. Install into the host you already use — Codex, Claude, Qoder, or Copilot.
Design principles:
.sopify/; open the repo on any host and pick up where you left off.sopify/What Sopify prevents:
See the workflow diagram, checkpoints, and resume flow →
curl -fsSL https://github.com/evidentloop/sopify/releases/latest/download/install.sh | bash -s -- --target codex:en-US
After install, use ~go to start a managed workflow. See Installation for other hosts, audit-first install, and Windows.
Already in a Sopify-managed repo? Open any AI host and continue — it picks up from where you left off.
When requirements are unclear, it stops to plan first.
You say "add a caching layer." Sopify doesn't start coding — it plans first: analyze, design, split into tasks, then save to .sopify/plan/. Only after you confirm the plan does it write code. Every line changed traces back to a decision.
Your teammate picks up where you left off.
You start a feature in Codex, finish the design, and implement two of four tasks. Next week your teammate opens the same repo in Claude, types ~go. Sopify reads the checkpoint and continues from task 3 — no handoff doc, no re-explaining context.
Every decision leaves a trace.
A month later, someone asks why the cache key includes the user ID. The answer is in .sopify/plan/ — the requirement that prompted the decision, the design that resolved it, the review that approved it.
The host LLM executes. Sopify preserves auditable development assets — plans, decisions, handoffs, and verification evidence — in .sopify/, accessible across sessions, hosts, and teammates.
How Sopify achieves stability and quality:
.sopify/; only the two local pointer files (active_plan.json, current_handoff.json) are gitignoredAudit-first install:
curl -fsSL -o sopify-install.sh https://github.com/evidentloop/sopify/releases/latest/download/install.sh
less sopify-install.sh # review before running
bash sopify-install.sh --target codex:en-US
Windows PowerShell:
iwr https://github.com/evidentloop/sopify/releases/latest/download/install.ps1 -OutFile sopify-install.ps1
Get-Content sopify-install.ps1 | more
.\sopify-install.ps1 --target codex:en-US
Host support:
| Host | Tier | Target | Notes |
|---|---|---|---|
| Codex | PROTOCOL_VERIFIED | codex:en-US / codex:zh-CN |
Full capability continuation |
| Claude | PROTOCOL_VERIFIED | claude:en-US / claude:zh-CN |
Full capability continuation |
| Qoder | PROTOCOL_VERIFIED | qoder |
Validated on Qoder CLI |
| Copilot | BASELINE_SUPPORTED | copilot:en-US / copilot:zh-CN |
Prompt-only; payload uplift planned |
Pass --workspace <path> to target another repo, --language <lang> to control output language.
For the full setup guide, see Getting Started. For a step-by-step demo, see External Repo Quickstart.
| Command | Description |
|---|---|
~go |
Automatically route and run the full workflow (auto-resumes if active plan exists) |
~go plan |
Plan only |
~go finalize |
Close out the active plan |
Most users only need ~go and ~go plan; maintainer validation commands live in CONTRIBUTING.md.
cp examples/sopify.config.yaml ./sopify.config.yaml
brand: auto
language: en-US
workflow:
mode: adaptive # strict | adaptive | minimal
require_score: 7
sopify/
├── scripts/ # install, diagnostics, and maintainer scripts
├── examples/ # configuration examples
├── docs/ # workflow guides and developer references
├── sopify_writer/ # protocol asset writer library
├── sopify_contracts/ # schema definitions and shared data structures
├── skills/ # prompt-layer source of truth
├── installer/ # host adapters and install orchestration
└── .sopify/ # project protocol root
├── blueprint/ # protocol spec, design baseline, reduction targets
├── plan/ # active plans + receipts
└── history/ # archived plans + receipts
See How Sopify Works for the full workflow, checkpoints, and knowledge layout.
For user-visible behavior changes, update both README.md and README.zh-CN.md when needed, then follow CONTRIBUTING.md for validation.