English | 한국어 | 日本語 | 中文 | Deutsch | Français

Codex CLI Looking for Codex CLI? → my-codex — same orchestration in native TOML format

---

---

Installation

For Humans

git clone --depth 1 https://github.com/sehoon787/my-claude.git /tmp/my-claude
bash /tmp/my-claude/install.sh
rm -rf /tmp/my-claude

Or install as a Claude Code plugin first, then run the companion installer:

# Inside a Claude Code session:
/plugin marketplace add sehoon787/my-claude
/plugin install my-claude@my-claude

# Then install companion tools:
git clone --depth 1 https://github.com/sehoon787/my-claude.git /tmp/my-claude
bash /tmp/my-claude/install.sh
rm -rf /tmp/my-claude

For AI Agents

Read https://raw.githubusercontent.com/sehoon787/my-claude/main/AI-INSTALL.md and follow every step.

---

Open-Source Tools Used

1. Oh My Claude Code (OMC)

An agent harness dedicated to Claude Code. 19 specialist agents (architect, debugger, code reviewer, security reviewer, etc.) divide work by role, and magic keywords like autopilot: activate automatic parallel execution.

2. Oh My OpenAgent (omo)

A multi-platform agent harness. Bridges to the Claude Code ecosystem via claude-code-agent-loader and claude-code-plugin-loader. Automatically routes across 8 providers (Claude, GPT, Gemini, etc.) by category. The 9 agents in this repository are adaptations of omo agents in Claude Code standalone .md format.

3. Andrej Karpathy Skills

The 4 AI coding behavioral guidelines proposed by Andrej Karpathy (Think Before Coding, Simplicity First, Surgical Changes, Goal-Driven Execution). Included in CLAUDE.md and always active across all sessions.

4. Everything Claude Code (ECC)

A comprehensive framework of 278 skills + 67 agents + 94 commands + language-specific rules. my-claude installs a curated 79-skill subset (stack patterns, AI/agent engineering, codebase tooling) plus 9 rule sets. Automates repetitive development patterns with slash commands like /tdd, /plan, /code-review, and /build-fix.

5. Anthropic Official Skills

The official agent skills repository provided directly by Anthropic. Enables specialist tasks such as PDF parsing, Word/Excel/PowerPoint document manipulation, and MCP server creation.

6. gstack

A sprint-process harness by Garry Tan. my-claude installs 26 of its skills plus the gstack root router (27 total). Provides browser-based QA testing (/qa), code review with scope-drift detection (/review), security auditing (/cso), and a full Plan→Review→QA→Ship deployment workflow. Includes a compiled Playwright browser daemon for real-browser testing.

7. superpowers

Jesse Vincent's development-process skill library. my-claude installs 13 of its 14 skills — brainstorming, systematic debugging, TDD, plan writing and execution, and code-review etiquette. (dispatching-parallel-agents is excluded because Boss and Agent Teams already own that path.)

---

How Boss Works

Boss is the meta-orchestrator at the core of my-claude. It never writes code — it discovers, classifies, matches, delegates, and verifies.

User Request
     │
     ▼
┌─────────────────────────────────────────────┐
│  Phase 0 · DISCOVERY                        │
│  Scan agents, skills, MCP, hooks at runtime │
│  → Build live capability registry           │
└──────────────────────┬──────────────────────┘
                       ▼
┌─────────────────────────────────────────────┐
│  Phase 1 · INTENT GATE                      │
│  Classify: trivial | build | refactor |     │
│  mid-sized | architecture | research | ...  │
│  → Counter-propose skill if better fit      │
└──────────────────────┬──────────────────────┘
                       ▼
┌─────────────────────────────────────────────┐
│  Phase 2 · CAPABILITY MATCHING              │
│  P0: gstack skill (if installed)            │
│  P1: Exact skill match                      │
│  P2: Specialist agent (32)                  │
│  P3: Multi-agent orchestration              │
│  P4: General-purpose fallback               │
└──────────────────────┬──────────────────────┘
                       ▼
┌─────────────────────────────────────────────┐
│  Phase 3 · DELEGATION                       │
│  6-section structured prompt to specialist  │
│  TASK / OUTCOME / TOOLS / DO / DON'T / CTX  │
└──────────────────────┬──────────────────────┘
                       ▼
┌─────────────────────────────────────────────┐
│  Phase 4 · VERIFICATION                     │
│  Read changed files independently           │
│  Run tests, lint, build                     │
│  Cross-reference with original intent       │
│  → Retry up to 3× on failure               │
└─────────────────────────────────────────────┘

Priority Routing

Boss cascades every request through a priority chain until the best match is found:

Priority	Match Type	When	Example
P0	gstack skill	Ship / QA / deploy / security workflow	"ship this" → gstack /ship
P1	Skill match	Task maps to a self-contained skill	"merge PDFs" → pdf skill
P2	Specialist agent	Domain-specific agent exists	"security audit" → security-reviewer
P3a	Boss direct	2-4 independent agents	"fix 3 bugs" → parallel spawn
P3b	Sub-orchestrator	Complex multi-step workflow	"refactor + test" → Sisyphus
P3c	Agent Teams	Peer-to-peer communication needed	"implement + review" → Review Chain
P4	Fallback	No specialist matches	"explain this" → general agent

Model Routing

Complexity	Model	Used For
Top-level orchestration	claude-fable-5	Boss
Deep analysis, architecture	claude-opus-5	Sisyphus, Atlas, Hephaestus, Oracle, Metis, Momus, Prometheus
Standard implementation	claude-sonnet-5	Librarian, Multimodal-Looker, OMC specialists
Quick lookup, exploration	claude-haiku-4-5	Lightweight OMC agents, simple advisory

Effort Tiers

Model choice sets which brain runs a task; the effort: frontmatter field sets how hard it thinks. Every self-owned agent declares one.

Effort	Agents
xhigh	Boss, Oracle, Prometheus, Multi-Agent Systems Architect
high	Sisyphus, Hephaestus, Atlas, Metis, Momus
medium	Librarian, Multimodal-Looker, AI Engineer, DevOps Automator

Skills declare effort too — boss-briefing runs at medium, briefing-vault at low. boss-advanced and gstack-sprint deliberately declare none: a skill's effort overrides the session level while it runs, which would quietly downgrade Boss mid-flow.

Precedence is CLAUDE_CODE_EFFORT_LEVEL (env) > frontmatter > session effort level. xhigh is Fable's supported ceiling; max is Opus-class only and silently falls back elsewhere.

3-Phase Sprint Workflow

For end-to-end feature implementation, Boss orchestrates a structured sprint:

Phase 1: DESIGN         Phase 2: EXECUTE        Phase 3: REVIEW
(interactive)            (autonomous)             (interactive)
─────────────────────   ─────────────────────   ─────────────────────
User decides scope      ralph runs execution    Compare vs design doc
Engineering review      Auto code review        Present comparison table
Confirm "design done"   Architect verification  User: approve / improve

Named Workflows

Deterministic multi-agent workflows. install.sh copies them to ~/.claude/workflows/, so they run from any project through the Workflow tool — not just from this repo.

Workflow	What It Does	Invocation
code-review-fanout	Four dimension reviewers (correctness, security, performance, tests) fan out in parallel, then every finding is adversarially verified before it is reported	Workflow({name: "code-review-fanout"}) — args: the review target (branch, commit range, paths). Defaults to the working-tree diff
upstream-audit	One analyst per upstream — pin delta vs origin, allowlist fit, new overlaps, security signals, health — followed by a synthesized action list	Workflow({name: "upstream-audit"}) — for quarterly or pre-sync audits

---

Architecture

┌─────────────────────────────────────────────────────┐
│                    User Request                       │
└───────────────────────┬─────────────────────────────┘
                        ▼
┌─────────────────────────────────────────────────────┐
│  Boss · Meta-Orchestrator (Fable)                     │
│  Discovery → Classification → Matching → Delegation  │
└──┬──────────┬──────────┬──────────┬─────────────────┘
   │          │          │          │
   ▼          ▼          ▼          ▼
┌──────┐ ┌────────┐ ┌────────┐ ┌────────┐
│ P3a  │ │  P3b   │ │  P3c   │ │  P1/P2 │
│Direct│ │Sub-orch│ │ Agent  │ │ Skill/ │
│2-4   │ │Sisyphus│ │ Teams  │ │ Agent  │
│agents│ │Atlas   │ │  P2P   │ │ Direct │
└──────┘ │Hephaes│ └────────┘ └────────┘
         └────────┘
┌─────────────────────────────────────────────────────┐
│  Behavioral Layer                                     │
│  Karpathy Guidelines · Rules (54) · Hooks (8)        │
├─────────────────────────────────────────────────────┤
│  Specialist Agents (32)                               │
│  Boss 1 · OMO 9 · OMC 19 · Vendored 3                │
├─────────────────────────────────────────────────────┤
│  Skills (139)                                         │
│  ECC 79 · gstack 27 · OMC 16 · Superpowers 13       │
│  + Core 4                                             │
├─────────────────────────────────────────────────────┤
│  MCP Layer                                            │
│  Context7 · Exa · grep.app                            │
├─────────────────────────────────────────────────────┤
│  Tooling Layer                                        │
│  LSP (2) · Named Workflows (2)                        │
└─────────────────────────────────────────────────────┘

---

What's Inside

Category	Count	Source
Agents (always loaded)	32	Boss 1 + OMO 9 + OMC 19 + Vendored 3
Skills	139	ECC 79 · gstack 27 · OMC 16 · Superpowers 13 · Core 4
Rules	54 files / 9 sets	ECC 53 (common + 8 language dirs) + Core 1
MCP Servers	3	Context7, Exa, grep.app
Hooks	8 files / 8 events	Delegation guard, telemetry, verification, vault
LSP Servers	2	typescript (typescript-language-server), python (pyright-langserver)
Named Workflows	2	code-review-fanout, upstream-audit
Upstream submodules	4	ecc, omc, gstack, superpowers
CLI Tools	3	omc, omo, ast-grep

Every agent, skill, and rule above is allowlisted in scripts/skill-allowlists.sh and tracked in the install manifest. Anthropic's official document skills (pdf, docx, …) are added separately via claude plugin add anthropics/skills and are deliberately not manifest-tracked.

Core Agent — Boss meta-orchestrator (1)

Agent	Model	Role	Source
Boss	Fable	Dynamic runtime discovery → capability matching → optimal routing. Never writes code.	my-claude

OMO Agents — Sub-orchestrators and specialists (9)

Agent	Model	Role	Source
Sisyphus	Opus	Intent classification → specialist delegation → verification	oh-my-openagent
Hephaestus	Opus	Autonomous explore → plan → execute → verify	oh-my-openagent
Atlas	Opus	Task decomposition + 4-stage QA verification	oh-my-openagent
Oracle	Opus	Strategic technical consulting (read-only)	oh-my-openagent
Metis	Opus	Intent analysis, ambiguity detection	oh-my-openagent
Momus	Opus	Plan feasibility review	oh-my-openagent
Prometheus	Opus	Interview-based detailed planning	oh-my-openagent
Librarian	Sonnet	Open-source documentation search via MCP	oh-my-openagent
Multimodal-Looker	Sonnet	Image/screenshot/diagram analysis	oh-my-openagent

OMC Agents — Specialist workers (19)

Agent	Role	Source
analyst	Pre-analysis before planning	oh-my-claudecode
architect	System design and architecture	oh-my-claudecode
code-reviewer	Focused code review	oh-my-claudecode
code-simplifier	Code simplification and cleanup	oh-my-claudecode
critic	Critical analysis, alternative proposals	oh-my-claudecode
debugger	Focused debugging	oh-my-claudecode
designer	UI/UX design guidance	oh-my-claudecode
document-specialist	Documentation writing	oh-my-claudecode
executor	Task execution	oh-my-claudecode
explore	Codebase exploration	oh-my-claudecode
git-master	Git workflow management	oh-my-claudecode
planner	Rapid planning	oh-my-claudecode
qa-tester	Quality assurance testing	oh-my-claudecode
scientist	Research and experimentation	oh-my-claudecode
security-reviewer	Security review	oh-my-claudecode
test-engineer	Test writing and maintenance	oh-my-claudecode
tracer	Execution tracing and analysis	oh-my-claudecode
verifier	Final verification	oh-my-claudecode
writer	Content and documentation	oh-my-claudecode

Vendored Agents — AI and infrastructure specialists (3)

Snapshotted from agency-agents (MIT) on 2026-07-27, when that submodule was removed. Only the engineering agents with no equivalent elsewhere in the stack were kept; each file carries its upstream attribution header.

Agent	Role	Source
AI Engineer	AI/ML engineering, model integration, data pipelines	agency-agents (vendored)
DevOps Automator	Infrastructure automation, CI/CD, cloud operations	agency-agents (vendored)
Multi-Agent Systems Architect	Agent topology, context management, failure recovery	agency-agents (vendored)

Skills — 139 from 5 sources

Each source is allowlisted in scripts/skill-allowlists.sh — anything not listed there is never installed.

Source	Count	Key Skills
everything-claude-code	79	coding-standards, react-patterns, fastapi-patterns, agent-architecture-audit, e2e-testing
gstack	27	/qa, /review, /ship, /cso, /investigate, /office-hours
oh-my-claudecode	16	autopilot, ralph, team, ultrawork, ralplan, omc-reference
superpowers	13	brainstorming, systematic-debugging, test-driven-development, writing-plans
my-claude Core	4	boss-advanced, boss-briefing, briefing-vault, gstack-sprint

MCP Servers (3) + Hooks (8)

MCP Servers

Server	Purpose	Cost
Context7	Real-time library documentation	Free
Exa	Semantic web search	Free 1k req/month
grep.app	GitHub code search	Free

Behavioral Hooks

Hook	Event	Behavior
Session Setup	SessionStart	Auto-detects missing tools + injects Briefing Vault context
Delegation Guard	PreToolUse	Blocks Boss from directly modifying files
Agent Telemetry	PostToolUse	Logs agent usage to agent-usage.jsonl
Subagent Verifier	SubagentStop	Forces independent verification + logs to Briefing Vault
Completion Check	Stop	Runs profile fallback + guards /boss-briefing execution
Teammate Idle Guide	TeammateIdle	Prompts leader on idle teammates
Task Quality Gate	TaskCompleted	Verifies deliverable quality
Vault Reminder	UserPromptSubmit	Suggests /boss-briefing after 5+ messages

LSP Servers (2)

The plugin declares two language servers in .lsp.json. Claude Code starts them on demand, so agents get diagnostics and code navigation instantly instead of paying for a build round-trip.

Server	Command	Extensions
typescript	typescript-language-server --stdio	.ts .tsx .js .jsx .mjs .cjs
python	pyright-langserver --stdio	.py

install.sh installs both binaries non-fatally — if one is unavailable, only that server is disabled and the rest of the install continues.

---

Briefing Vault

Obsidian-compatible persistent memory. Every project maintains a .briefing/ directory that auto-populates across sessions.

.briefing/
├── INDEX.md                          ← Project context (auto-created once)
├── state.json                        ← Session metadata, counters, lastVaultSync (auto-managed)
├── sessions/
│   ├── YYYY-MM-DD-<topic>.md        ← AI-written session summary (enforced)
│   └── YYYY-MM-DD-auto.md           ← Auto-generated scaffold (git diff, agent stats)
├── decisions/
│   └── YYYY-MM-DD-<decision>.md     ← AI-written decision record (enforced)
├── learnings/
│   ├── YYYY-MM-DD-<pattern>.md      ← AI-written learning note
│   └── YYYY-MM-DD-auto-session.md   ← Auto-generated scaffold (agents, files)
├── references/
│   └── auto-links.md                ← Auto-collected URLs from web searches
├── archives/                         ← PARA: completed/inactive notes (flat)
├── wiki/                             ← LLM-wiki: concept pages
│   └── _schema.md
├── agents/
│   ├── agent-log.jsonl              ← Subagent execution telemetry
│   └── YYYY-MM-DD-summary.md        ← Daily agent usage breakdown
└── persona/
    ├── profile.md                   ← Agent affinity stats (auto-updated)
    ├── suggestions.jsonl            ← Routing suggestions (auto-generated)
    └── rules/                       ← Workflow pattern rules (workflow-*.md)

Sub-Vaults

Path	Description
INDEX.md	Project overview with links to recent decisions and learnings. Auto-created on first session, refreshed by /boss-briefing or Stop hook fallback.
state.json	Session metadata: counters (workCounter, messageCount), lastVaultSync timestamp, sessionStartHead. Auto-managed by hooks.
sessions/	Session summaries. *-auto.md — scaffold with git diff stats and agent counts. <topic>.md — AI-written summary enforced by Stop hook guard.
decisions/	Architecture and design decisions with rationale. AI-written, enforced during active work.
learnings/	Patterns, gotchas, non-obvious solutions. *-auto-session.md — scaffold with file lists. <topic>.md — AI-written.
references/	Web research URLs. auto-links.md — auto-collected from WebSearch/WebFetch calls.
agents/	Agent telemetry. agent-log.jsonl — per-call log with enriched fields {ts, agent_type, phase, seq, task_hint}. YYYY-MM-DD-summary.md — daily usage breakdown.
persona/	User work style profile. profile.md — tool affinity stats. suggestions.jsonl — routing recommendations. Workflow sequence patterns in rules/workflow-*.md. Run /boss-briefing to analyze.
archives/	PARA Archives — completed sessions (30+ days), superseded decisions, inactive learnings
wiki/	LLM-wiki concept pages — distilled knowledge from multiple sessions

Knowledge Management (v2)

BriefingVault v2 integrates three knowledge management methodologies:

Methodology	Applied As
PARA (Tiago Forte)	Directory structure: sessions=Projects, decisions=Areas, references=Resources, archives=Archives
Zettelkasten (Luhmann)	Atomic notes in learnings/, unique IDs (YYYYMMDDHHMMSS), enforced [[wiki-links]]
LLM-wiki (Karpathy)	Concept pages in wiki/ — auto-suggested when keywords appear 3+ times

Claude Code session-end hooks automatically:

• Suggest archiving notes older than 30 days
• Propose wiki pages for frequently mentioned concepts
• Generate unique Zettelkasten IDs for new notes

Session-Specific Diffs

At session start, the current git HEAD is saved to state.json (sessionStartHead field). For non-git projects, a YYYY-MM-DD:cwd identifier is used instead. At session end, diffs are calculated relative to this saved point — showing only changes from the current session, not accumulated uncommitted changes from previous sessions.

Using with Obsidian

1. Open Obsidian → Open folder as vault → select .briefing/
2. Notes appear in graph view, linked by [[wiki-links]]
3. YAML frontmatter (date, type, tags) enables structured search
4. Timeline of decisions and learnings builds automatically over sessions

/boss-briefing

Run /boss-briefing during or at the end of a session to:

• Sync vault: Update profile.md, INDEX.md, and agent summaries
• Detect workflow patterns: Analyze temporal agent call sequences across sessions
• Recover from gaps: Generate recovery summaries if days have passed since the last session
• Propose persona rules: Suggest workflow-based routing preferences (not just frequency)
• Validate session notes: Check that today's session has a proper summary

The Stop hook checks whether /boss-briefing has run today. If not, it blocks session end with a reminder. The existing stop-profile-update.js continues to run as a fallback.

---

Upstream Open-Source Sources

my-claude links 4 MIT-licensed upstream repositories as git submodules, each pinned to an explicit SHA:

#	Source	What It Provides
1	everything-claude-code — affaan-m	79 installed skills + 9 rule sets. Language and stack knowledge: TDD, security, coding standards, framework patterns.
2	oh-my-claudecode — Yeachan Heo	19 specialist agents + 16 installed skills. Orchestration lane: autopilot, ralph, team.
3	gstack — garrytan	27 installed skills for ship, QA, deploy, and security review (Boss P0 lane). Includes Playwright browser daemon.
4	superpowers — Jesse Vincent	13 installed skills covering the dev-process lane: brainstorming, TDD, systematic debugging, plan writing.

Not submodules, but part of the stack:

Source	How It Arrives
oh-my-openagent — code-yeongyu	9 OMO agents (Sisyphus, Atlas, Oracle, etc.), adapted into this repo as standalone .md agents under agents/omo/.
agency-agents — msitarzewski	Submodule removed 2026-07-27. 3 engineering agents vendored into agents/vendored/ with attribution.
anthropic/skills — Anthropic	Installed by install.sh via claude plugin add anthropics/skills (pdf, docx, and friends). Not manifest-tracked.
andrej-karpathy-skills — forrestchang	4 AI coding behavioral guidelines appended to ~/.claude/CLAUDE.md.

---

GitHub Actions

Workflow	Trigger	Purpose
CI	push, PR	Validates JSON configs, agent frontmatter, skill existence, upstream file counts
Smoke	push, PR	4 jobs — hooks (hook execution), shell (install script shape), drift (model drift), routing-refs (dangling agent/skill references)
Update Upstream	every 3 days / manual	git submodule update --remote → refresh SHA pins in SOURCES.json → security scan of the upstream diff → auto-merge only when the scan is clean, otherwise the PR is flagged for human review
Auto Tag	push to main	Reads plugin.json version and creates git tag if new
Pages	push to main	Deploys docs/index.html to GitHub Pages
CLA	PR	Contributor License Agreement check
Lint Workflows	push, PR	Validates GitHub Actions workflow YAML syntax

---

my-claude Originals

Features built specifically for this project, beyond what upstream sources provide:

Feature	Description
Boss Meta-Orchestrator	Dynamic capability discovery → intent classification → 5-priority routing → delegation → verification
3-Phase Sprint	Design (interactive) → Execute (autonomous via ralph) → Review (interactive vs design doc)
Agent Tier Priority	core > omo > omc > vendored deduplication. Most specialized agent wins.
Lane Ownership	Orchestration → OMC, dev process → superpowers, ship/QA/deploy/security → gstack (Boss P0), language and stack knowledge → ECC, AI and domain work → vendored agents
Curated Allowlists	scripts/skill-allowlists.sh is the single source of truth — 139 skills and 9 rule sets survive from thousands upstream, so nothing unlisted ever reaches a session's context
Briefing Vault	Obsidian-compatible .briefing/ directory with sessions, decisions, learnings, references
Agent Telemetry	PostToolUse hook logs agent usage to agent-usage.jsonl
No-op Sync Skip	Upstream sync stages the submodule bumps and SOURCES.json pins, then opens a PR only when that staged diff is non-empty
Agent Dedup Detection	tests/validate-sync.sh compares agent filenames across agents/ and the omc/superpowers submodules and reports collisions

---

Bundled Upstream Versions

Linked via git submodules. Pinned commits are tracked natively by .gitmodules and mirrored as an AI-BOM in upstream/SOURCES.json; install.sh checks out these exact SHAs rather than tracking main.

Source	SHA	Date	Diff
everything-claude-code	4092795	2026-07-27	compare
oh-my-claudecode	590fb98	2026-07-27	compare
gstack	7c9df1c	2026-07-27	compare
superpowers	3dcbd5c	2026-07-27	compare

---

Contributing

Issues and PRs are welcome. When adding a new agent, add a .md file to agents/core/ or agents/omo/ and update SETUP.md.

Credits

Built on the work of: oh-my-claudecode (Yeachan Heo), oh-my-openagent (code-yeongyu), everything-claude-code (affaan-m), gstack (garrytan), superpowers (Jesse Vincent), agency-agents (msitarzewski — 3 vendored agents), anthropic/skills (Anthropic), andrej-karpathy-skills (forrestchang).

License

MIT License. See the LICENSE file for details.