An opinionated Claude Code configuration that trades narrative for signal.
Install • Concepts • Workflow • Commands • Skills • Agents • Credits
Still evolving. Commands, agents, and skills may shift between releases. Breaking changes are flagged in the changelog.
Same question, same model, default Claude Code vs. Atomic Claude:
Default Claude Code:
Sure! I'd be happy to help you understand that. The issue you're experiencing
is likely caused by the middleware not properly validating the token expiry
timestamp. What's happening is that the comparison uses a strict less-than
operator when it should be using less-than-or-equal-to, which means tokens
that expire at exactly the current second are incorrectly rejected.
I'd suggest updating the comparison on line 42 of src/auth/middleware.ts...
Atomic Claude:
Bug in auth middleware. Token expiry check at src/auth/middleware.ts:42
uses `<` not `<=` — tokens expiring at current second get rejected.
Fix: change `token.exp < now` to `token.exp <= now`.
Same accuracy. Fewer tokens. Faster to scan.
Atomic Claude is a coherent system of output styles, discipline skills, subagents, and workflow commands designed to compose. The pieces:
Compressed replies. A tone layer that strips filler from Claude's responses. Three intensity levels switchable mid-session. Opt in via /config then Output style then Atomic. Details in docs/reference/output-style.md.
Project-state awareness. The atomic binary scans your repo and writes context files that Claude loads every session. Claude knows your framework, package manager, and build commands instead of guessing. Details in docs/reference/signals-workflow.md.
User profile. Install creates ~/.claude/.atomic/profile.md, a plain-markdown file with six sections (Identity, Work, Active projects, Interests, People mentioned, Environment). Claude reads it every session and appends new facts as they surface naturally. Facts that hold across all projects go here; project-specific preferences stay in that project's auto memory. Run /atomic-improve to review and accept or discard drift. Run atomic profile refresh to re-detect your dev tooling (runtimes, version managers, containers, shell framework) and rewrite the ## Environment block. The session-start hook fires this automatically with --if-stale 7d so the env stays current. Details in docs/reference/concepts.md.
A spec-to-PR workflow. Verify a hunch with /gather-evidence, plan with /atomic-plan, implement with /subagent-implementation, diagnose failures with /subagent-diagnose, then close the loop with /atomic-improve to turn session friction into durable config changes. Each stage uses fresh-context subagents that write TDD, gate on review, and commit per green checkpoint. Close your laptop, rerun the command next week, pick up where you left off. Or hand the whole thing to /autopilot, which runs plan through ship on its own and asks only how to merge. Details in docs/reference/workflow.md.
Discipline skills that auto-fire. Seven skills trigger on natural language: TDD enforcement, completion verification, debugging, commit messages, code review, prose editing, and documentation routing. No slash command needed. Details in docs/reference/skills.md.
Workflow commands for every git operation. Ten verbs covering every combination of commit, push, squash, PR, and merge-to-base. Plus utilities for CI watching, stale branch cleanup, worktree isolation, reminders, and integrity checks. Details in docs/reference/commands.md.
For a walkthrough of how the pieces fit together, see docs/reference/concepts.md.
Not sure where to begin? Run /atomic-help in any repo. It reads your git state and recommends one next command, or run /atomic-help tour for a four-stage guided walkthrough of the whole system.
Otherwise, pick your depth:
- Compressed replies only. Install, activate the output style via
/config. Done. Everything else is optional. - Project-state awareness. Run
/atomic-setup+/refresh-signalsin your repo. Claude stops hallucinating build commands. - Full plan, implement, review loop. Read the workflow reference.
Two commands. The first downloads the atomic binary (macOS / Linux / WSL2):
curl -fsSL https://raw.githubusercontent.com/damusix/atomic-claude/main/install.sh | bashThe second wires up the artifact bundle into ~/.claude/:
atomic claude installActivate the output style with /config then Output style then Atomic, and you're set.
For prereqs, flags, existing ~/.claude/CLAUDE.md handling, updates, Docker evaluation, and uninstall: docs/guides/install.md.
First session inside a new project. /atomic-setup audits conventions, /refresh-signals teaches Claude the repo's shape:
❯ /atomic-setup
⏺ Audit
┌─────────────────────────────────────┬────────────────────────────────────┐
│ Convention │ Status │
├─────────────────────────────────────┼────────────────────────────────────┤
│ .gitignore has .claude/.scratchpad/ │ ✗ │
│ .gitignore has .worktrees/ │ ✗ │
│ docs/spec/ │ ✗ │
│ docs/design/ │ ✗ │
│ Signals files │ ✗ │
│ CLAUDE.md @-refs │ ✗ │
└─────────────────────────────────────┴────────────────────────────────────┘
Proposed actions
[1] Append .claude/.scratchpad/ and .worktrees/ to .gitignore
[2] Create docs/spec/.gitkeep + docs/design/.gitkeep
[3] Append signals @-refs to CLAUDE.md
Apply which actions? → All (1-3)
✓ Applied. Next: /refresh-signals
❯ /refresh-signals
⏺ atomic signals scan
⎿ Tree · Manifests · Languages → .claude/project/deterministic-signals.md
⏺ atomic-signals-inferrer
⎿ Framework: NestJS · package manager: pnpm
⎿ Build: pnpm build · test: pnpm test · lint: pnpm lint
⎿ → .claude/project/signals.md
signals initialized.
From here: /atomic-plan opens the spec, /worktree-start isolates the branch, /subagent-implementation drives the loop.
| Topic | Link |
|---|---|
| Workflow lifecycle | docs/reference/workflow.md |
| Commands | docs/reference/commands.md |
| Skills | docs/reference/skills.md |
| Agents | docs/reference/agents.md |
| Output style | docs/reference/output-style.md |
| Signals workflow | docs/reference/signals-workflow.md |
| Concepts (how it flows) | docs/reference/concepts.md |
| Conventions | docs/reference/conventions.md |
| Install / update / uninstall | docs/guides/install.md |
| Docker evaluation | docs/guides/evaluations.md |
| Contributing | docs/guides/contributing.md |
| Credits | docs/credits.md |
| Specs | docs/spec/ |
Atomic Claude dogfoods itself. The root artifacts are both the live config and the bundle source. See docs/guides/contributing.md.