░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
░ ◆ ◆ ◆ ░
░ ╱ ╲ ░
░ ╱ CROWN ╲ ░
░ ◆═══════◆ ░
░ ░
░ ▓▓▓▓▓▓▓ ▓▓▓▓▓▓▓╗▓▓▓▓▓▓▓▓╗ ▓▓▓▓▓╗ ▓▓▓▓▓▓╗▓▓╗ ▓▓╗ ░
░ ▓▓ ▓▓ ▓▓╔════╝╚══▓▓╔══╝▓▓╔══▓▓╗▓▓╔════╝▓▓║ ▓▓╔╝ ░
░ ▓▓ ▓▓╗▓▓▓▓▓▓▓╗ ▓▓║ ▓▓▓▓▓▓▓║▓▓║ ▓▓▓▓╔╝ ░
░ ▓▓ ▓▓╚════ ▓▓║ ▓▓║ ▓▓╔══▓▓║▓▓║ ▓▓╔═▓▓╗ ░
░ ╚▓▓▓▓▓╔╝▓▓▓▓▓▓▓║ ▓▓║ ▓▓║ ▓▓║╚▓▓▓▓▓▓╗▓▓║ ▓▓╗ ░
░ ╚═════╝╚══════╝ ╚═╝ ╚═╝ ╚═╝ ╚═════╝╚═╝ ╚═╝ ░
░ ░
░ ┌─────┐ ┌─────┐ ┌─────┐ ┌─────┐ ┌─────┐ ░
░ │█████│ │█████│ │█████│ │█████│ │█████│ ░
░ │█ 1 █│ │█ 2 █│ │█ 3 █│ │█ 4 █│ │█ N █│ ● ● ● ● ░
░ │█████│ │█████│ │█████│ │█████│ │█████│ ░
░ └──┬──┘ └──┬──┘ └──┬──┘ └──┬──┘ └──┬──┘ ░
░ ═══╧══════╧══════╧══════╧══════╧═══ ░
░ ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ EMPIRE ░
░ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓ FOUNDATION ░
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
Built by Oussama Ammar (@daedalium) — solo capitalist and venture builder operating as a "Holding Company of One" from Dubai. Early investor in Stripe, Algolia, PayFit, Trainline. Previously co-founded The Family, a European startup accelerator whose portfolio raised $1.5B+.
Standing on the shoulders of a giant.
ostack is a fork of gstack by Garry Tan — Y Combinator President and legendary builder. Garry open-sourced gstack as an AI-powered engineering team for Claude Code, and made it free and MIT-licensed for everyone. That generosity is rare at his level, and it gave people like me a foundation that would have taken months to build from scratch. All credit for the original architecture, skills system, browser engine, and sprint philosophy goes to Garry and the gstack contributors.
ostack takes that foundation and injects a different soul — my own. Different philosophy, different mental models, different decision frameworks. Same incredible bones.
ostack turns Claude Code into a virtual engineering team. Not a copilot that completes your lines — a full team that thinks, plans, reviews, tests, and ships.
You get a CEO who rethinks the product, an eng manager who locks architecture, a designer who catches AI slop, a staff engineer who finds production bugs, a QA lead who opens a real browser and clicks through your app, a security officer who runs OWASP + STRIDE audits, and a release engineer who ships the PR.
Specialist skills. All slash commands. All Markdown. All free.
- Solo founders — one person, institutional output. The org chart is gone.
- Technical CEOs — who still want to ship, not just manage.
- First-time Claude Code users — structured roles instead of a blank prompt.
- Staff engineers — rigorous review, QA, and release automation on every PR.
ostack is not a cosmetic rebrand. It's a philosophical fork.
| Dimension | gstack | ostack |
|---|---|---|
| Philosophy | Engineering-first | Operator-first — the Sovereign Builder worldview |
| Mental models | Standard best practices | Hybrid: SHU-HA-RI, Power Law concentration, Contrarian by Construction |
| Decision framework | Process-driven | Taste-driven — "Context Over Principle" as the meta-rule |
| Communication | Comprehensive | Compressed — no hedging, no filler, conviction first |
| CEO review | Strategic review | Merged cognitive patterns — thinks like a founder who builds |
| Builder ethos | Implicit | Explicit — ETHOS.md codifies the worldview |
| Code quality | High | Garry's engineering rigor cranked to 11 |
| Identity | gstack | ostack — "Ouss's stack" |
The core idea: AI collapses the org chart. A single operator with the right tools achieves institutional output — not because the tools replace judgment, but because they multiply it.
# 1. Install (30 seconds)
git clone https://github.com/mr-daedalium/ostack.git ~/.claude/skills/ostack
cd ~/.claude/skills/ostack && ./setup
# 2. Try it
/office-hours # Describe what you're building
/plan-ceo-review # Challenge the scope
/review # Find bugs in your branch
/qa https://your.app # QA with a real browser
/ship # Ship the PRStop after /qa. You'll know if this is for you.
You: I want to build a daily briefing app for my calendar.
You: /office-hours
Claude: [asks about the pain — specific examples, not hypotheticals]
You: Multiple Google calendars, events with stale info, wrong locations.
Prep takes forever and the results aren't good enough...
Claude: I'm going to push back on the framing. You said "daily briefing
app." But what you actually described is a personal chief of
staff AI.
[extracts 5 capabilities you didn't realize you were describing]
[challenges 4 premises — you agree, disagree, or adjust]
[generates 3 implementation approaches with effort estimates]
RECOMMENDATION: Ship the narrowest wedge tomorrow, learn from
real usage. The full vision is a 3-month project — start with
the daily briefing that actually works.
[writes design doc -> feeds into downstream skills automatically]
You: /plan-ceo-review
[reads the design doc, challenges scope, runs 10-section review]
You: /plan-eng-review
[ASCII diagrams for data flow, state machines, error paths]
[test matrix, failure modes, security concerns]
You: Approve plan. Exit plan mode.
[writes 2,400 lines across 11 files. ~8 minutes.]
You: /review
[AUTO-FIXED] 2 issues. [ASK] Race condition -> you approve fix.
You: /qa https://staging.myapp.com
[opens real browser, clicks through flows, finds and fixes a bug]
You: /ship
Tests: 42 -> 51 (+9 new). PR: github.com/you/app/pull/42
You said "daily briefing app." The agent said "you're building a chief of staff AI" — because it listened to your pain, not your feature request. Eight commands, end to end. That is not a copilot. That is a team.
ostack is a process, not a collection of tools. The skills run in the order a sprint runs:
Think -> Plan -> Build -> Review -> Test -> Ship -> Reflect
Each skill feeds into the next. /office-hours writes a design doc that /plan-ceo-review reads. /plan-eng-review writes a test plan that /qa picks up. /review catches bugs that /ship verifies are fixed. Nothing falls through the cracks because every step knows what came before it.
| Phase | Skill | Specialist | What they do |
|---|---|---|---|
| Think | /office-hours |
YC Office Hours | Six forcing questions that reframe your product before code. Challenges premises, generates alternatives. Design doc feeds downstream. |
| Plan | /plan-ceo-review |
CEO / Founder | Find the 10-star product. Four modes: Expansion, Selective Expansion, Hold Scope, Reduction. |
/plan-eng-review |
Eng Manager | Lock architecture, data flow, edge cases, and tests. Forces hidden assumptions into the open. | |
/plan-design-review |
Senior Designer | Rates each design dimension 0-10. Explains what a 10 looks like. Fixes the plan. AI Slop detection. | |
/design-consultation |
Design Partner | Build a complete design system from scratch. Typography, color, layout, spacing, motion. | |
/autoplan |
Review Pipeline | One command, fully reviewed plan. CEO + design + eng review with encoded decision principles. | |
| Build | /browse |
QA Engineer | Real Chromium browser, real clicks, real screenshots. ~100ms per command. |
/investigate |
Debugger | Systematic root-cause debugging. Iron Law: no fixes without investigation. | |
| Review | /review |
Staff Engineer | Find bugs that pass tests but blow up in production. Auto-fixes the obvious ones. |
/codex |
Second Opinion | Independent review from OpenAI Codex CLI. Cross-model analysis. | |
/design-review |
Designer Who Codes | Live-site visual audit + fix loop. Atomic commits, before/after screenshots. | |
/cso |
Chief Security Officer | OWASP Top 10 + STRIDE. Zero-noise: 8/10+ confidence gate, exploit scenarios. | |
| Test | /qa |
QA Lead | Real browser testing. Find bugs, fix them, re-verify. Auto-generates regression tests. |
/qa-only |
QA Reporter | Same methodology, report only. No code changes. | |
/benchmark |
Performance Engineer | Baseline page loads, Core Web Vitals, resource sizes. Before/after on every PR. | |
| Ship | /ship |
Release Engineer | Sync main, run tests, audit coverage, push, open PR. |
/land-and-deploy |
Release Engineer | Merge, verify the deploy, and check production health. One command. | |
/canary |
SRE | Post-deploy monitoring. Console errors, perf regressions, page failures. | |
| Reflect | /retro |
Eng Manager | Team-aware weekly retro. Per-person breakdowns, shipping streaks, growth areas. /retro global across all projects. |
/document-release |
Technical Writer | Update all project docs to match what shipped. Catches stale READMEs. |
| Skill | What it does |
|---|---|
/careful |
Warns before destructive commands (rm -rf, DROP TABLE, force-push). Override any warning. |
/freeze |
Restrict file edits to one directory. Prevents accidental changes while debugging. |
/guard |
/careful + /freeze in one command. Maximum safety for prod work. |
/unfreeze |
Remove the /freeze boundary. |
/setup-deploy |
One-time deploy config for /land-and-deploy. Detects your platform automatically. |
/setup-browser-cookies |
Import cookies from Chrome, Arc, Brave, or Edge into the headless session. |
/ostack-upgrade |
Self-updater. Detects global vs vendored install, syncs both, shows what changed. |
Deep dives with examples for every skill ->
This is why ostack exists. The ratio between human-team time and AI-assisted time:
| Task type | Human team | With ostack | Compression |
|---|---|---|---|
| Boilerplate / scaffolding | 2 days | 15 min | ~100x |
| Test writing | 1 day | 15 min | ~50x |
| Feature implementation | 1 week | 30 min | ~30x |
| Bug fix + regression test | 4 hours | 15 min | ~20x |
| Architecture / design | 2 days | 4 hours | ~5x |
| Research / exploration | 1 day | 3 hours | ~3x |
The last 10% of completeness that teams used to skip? It costs seconds now. Do the full thing. Every time.
Requirements: Claude Code or Codex, Git, Bun v1.0+, Node.js (Windows only)
Global install:
git clone https://github.com/mr-daedalium/ostack.git ~/.claude/skills/ostack
cd ~/.claude/skills/ostack && ./setupPer-project install:
cp -Rf ~/.claude/skills/ostack .claude/skills/ostack
rm -rf .claude/skills/ostack/.git
cd .claude/skills/ostack && ./setup./setup prints the short CLAUDE.md snippet to add. Use that instead of pasting a long hand-maintained block from the README.
Per-repo install:
git clone https://github.com/mr-daedalium/ostack.git .agents/skills/ostack
cd .agents/skills/ostack && ./setup --host codexGlobal install:
git clone https://github.com/mr-daedalium/ostack.git ~/ostack
cd ~/ostack && ./setup --host codex./setup --host codex prints the short AGENTS.md snippet to add. Codex support is intentionally first-class; other hosts are out of scope.
ostack works well with one sprint. It gets interesting with ten running at once.
Conductor runs multiple Claude Code sessions in parallel — each in its own isolated workspace. One session on /office-hours, another on /review, a third implementing a feature, a fourth running /qa. All at the same time. The sprint structure is what makes parallelism work — without a process, ten agents is ten sources of chaos. With a process, each agent knows exactly what to do and when to stop.
ostack is built on seven principles. The full manifesto is in ETHOS.md. The short version:
- The Sovereign Builder — AI collapses the org chart. One operator with compound tools achieves institutional output.
- Often Wrong, Never in Doubt — Conviction expressed as speed. Ship, learn, course-correct.
- Context Over Principle — The meta-rule. When a principle becomes dogma, it becomes the enemy.
- Search Before Building — Four layers of knowledge: solved, tried-and-true, new-and-popular, first principles.
- Leverage Over Effort — Code, media, capital. Master at least two.
- Taste as Quality Bar — AI makes shipping easy. That makes taste more important, not less.
- SHU-HA-RI — Obey the rules, break them deliberately, transcend them entirely.
"The worst outcome is a complete, well-engineered product that nobody needed. The best outcome is a product born from conviction, built with leverage, filtered through taste, and shipped at a speed that would have been impossible two years ago — by one person who knew when to follow the rules and when to throw them away."
ostack/
├── browse/ # Headless Chromium CLI (Playwright, ~100ms/cmd)
│ ├── src/ # CLI + persistent server + 50+ commands
│ └── dist/ # Compiled binary
├── [skill-name]/ # skill directories, each with SKILL.md.tmpl
├── scripts/ # Build tooling: template generator, health checks
├── test/ # 3-tier test suite: validation, E2E, LLM-as-judge
├── bin/ # CLI utilities (config, update, repo-mode detection)
├── ETHOS.md # Builder philosophy
├── SKILL.md.tmpl # Root skill template (generates SKILL.md)
└── setup # One-command installer for all agent platforms
Tech stack: TypeScript, Bun, Playwright, Anthropic SDK
How skills work: Each skill is a SKILL.md.tmpl template that gets compiled into SKILL.md — a prompt that Claude Code or Codex reads and executes. Templates use shared partials (preamble, browse setup, effort tables) so the skill set stays consistent. Skills are just Markdown. No plugins, no APIs, no lock-in.
How browse works: A persistent headless Chromium server that starts on first use (~3s) and stays alive for 30 minutes. Each command takes ~100ms. 50+ commands: navigate, click, fill, screenshot, snapshot (accessibility tree with clickable refs), responsive testing, cookie import, network/console inspection. The agent treats it like a real browser because it is one.
| Doc | What it covers |
|---|---|
| Skill Deep Dives | Philosophy, examples, and workflow for every skill |
| Builder Ethos | The Sovereign Builder philosophy |
| Architecture | Design decisions and system internals |
| Browser Reference | Full command reference for /browse |
| Contributing | Dev setup, testing, contributor mode |
| Changelog | What's new in every version |
| Problem | Fix |
|---|---|
| Skill not showing up | cd ~/.claude/skills/ostack && ./setup |
/browse fails |
cd ~/.claude/skills/ostack && bun install && bun run build |
| Stale install | Run /ostack-upgrade or set auto_upgrade: true in ~/.ostack/config.yaml |
| Windows issues | Node.js required alongside Bun (bun#4253). Both bun and node must be on PATH. |
| Claude can't see skills | Add an ostack section to your project's CLAUDE.md listing the available skills |
See CONTRIBUTING.md for dev setup, the 3-tier test suite, and contributor mode.
bun install # install dependencies
bun run test:fast # quick local checks
bun test # full free suite
bun run test:evals # paid evals (~$4/run, needs ANTHROPIC_API_KEY)
bun run skill:check # health dashboard for all skillsMIT. Forked from gstack by Garry Tan.
Go build something.