mirror of
https://github.com/JuliusBrussee/cavekit.git
synced 2026-09-14 16:32:40 +08:00
070ed06bb9521e5b1edd53bc52d82b9dbecc4373
SDD — Spec-Driven Development
Claude Code plugin + tmux dashboard for spec-driven development with automated iteration loops.
Install
git clone https://github.com/JuliusBrussee/sdd-os.git ~/.sdd
~/.sdd/install.sh
This registers the Claude Code plugin, installs the sdd CLI, and makes all scripts executable.
Terminal: tmux monitor
sdd --monitor # launch dashboard + claude + build loop
sdd --monitor --adversarial # with Codex adversarial review
sdd --monitor --filter v2 # scope to v2 specs
sdd --kill # tear down the session
Opens a tmux session with three panes:
- Left (70%) — Claude Code running
/sdd-execute - Top-right — live progress: tasks done, tiers, progress bar
- Bottom-right — live activity: iteration log, git commits
Claude commands
/sdd-brainstorm → write specs (the WHAT)
/sdd-plan → generate frontier (the ORDER)
/sdd-execute → ralph loop (the BUILD)
/sdd-review → gap analysis + adversarial review (the CHECK)
Brainstorm — write specs
/sdd-brainstorm # interactive — asks what to build
/sdd-brainstorm context/refs/ # from PRDs, API docs, research
/sdd-brainstorm --from-code # from existing codebase
Plan — generate frontier
/sdd-plan # all specs
/sdd-plan --filter v2 # only v2 specs
Execute — run the loop
/sdd-execute # implement everything
/sdd-execute --adversarial # add Codex (GPT-5.4) review
/sdd-execute --max-iterations 30
Review — post-loop check
/sdd-review # gap analysis + adversarial review
All commands
| Command | Description |
|---|---|
/sdd-brainstorm |
Write specs |
/sdd-plan |
Generate feature frontier |
/sdd-execute |
Implementation loop |
/sdd-review |
Gap analysis + adversarial review |
/sdd-progress |
Check frontier progress |
/sdd-gap-analysis |
Compare built vs intended |
/sdd-back-propagate |
Trace manual fixes to specs |
/sdd-help |
Show usage |
Update
cd ~/.sdd && git pull
License
MIT
Description
caveman: Caveman encoding for SPEC.md and spec-adjacent writes. Loaded by /spec, /build,
/check. Cuts tokens ~75% vs prose while staying precise. Triggers on any…; build: Plan-then-execute implementation against SPEC.md. Native single-thread
loop, no sub-agents. On test or build failure, auto-invokes the backprop
skill before…; spec: Create, amend, or backprop bugs into SPEC.md at repo root. Sole mutator
of the project spec. Triggers when the user asks to write a spec, start
a new spec,…; check…
Readme
MIT
4.1 MiB
Languages
Markdown
100%