By ATTCKDigital Verified
Smith is a Spec-driven development harness for Claude Code with persistent memory, overnight batch builds, model routing, and agency operations tooling.
Browse, search, and manage the Smith Ledger — the accumulated patterns, antipatterns, tool preferences, and edge cases learned from past workflows. Use to review what Smith has learned, search for relevant patterns, prune outdated entries, or inspect the Ledger's evolution over time.
One-time migration of existing flat spec folders into the system-based hierarchy under .specify/systems/.
One-time migration to add `paths:` YAML frontmatter to existing `.specify/systems/<id>/spec.md` files. Scans prose for path-like references (services/<X>/, backend/<X>/, frontend/<X>/, etc.), proposes a per-system list, and writes accepted frontmatter ABOVE the existing body. Use when a project grew system specs as prose and you want the Smith manifest v2 path-resolver tier 1 to bucket files correctly.
Manifest navigator. Returns must-read / should-read / reference file lists for a given task description by reading .smith/index/. Designed to be invoked by context-loader.sh hook AND directly by users (`/smith-navigate "where is auth?"`). Runs as a Haiku 4.5 sub-agent with a 3-second budget.
Start a new feature from scratch or from conversation context. Conversational requirements gathering, planning, questions gate, then fully autonomous build.
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Spec-driven development for Claude Code — requirements, plans, and tasks that ship as working code.
See smith.attck.com for a walkthrough.
Claude Code is a powerful AI coding assistant, but it has no built-in workflow structure. Developers jump straight from a vague idea to generated code with no specification, no plan, and no audit trail. The result is hard to review, harder to maintain, and impossible to trace back to requirements. When something goes wrong — and it will — there is no record of what was intended, what was decided, or why.
Smith fixes this by adding 28 skills that encode a full development workflow into Claude Code. The pipeline flows from spec to plan to tasks to implementation to review to ship. Every step produces a versioned artifact inside a .specify/ directory in your project. Claude reads the output of each step as input to the next, so context accumulates instead of evaporating. You never have to re-explain what you're building.
The outcome: you talk to Claude about what you want to build, Smith handles the structured process, and you get a merged PR with full traceability from idea to code. Hooks log every session automatically and guard against common mistakes — dangerous shell commands, secret exposure, writes to sensitive files. A scheduler can process queued tasks overnight. Everything runs locally on your machine, nothing phones home, and every artifact is a plain text file you can read, diff, and version-control.
skills CLI (skills only)npx skills add ATTCKDigital/smith
This is the fastest path — it copies all 26 Smith skills into ~/.claude/skills/ and nothing else. Use this if you only want the Smith workflow commands.
To update: re-run the same command. npx skills add is idempotent.
To verify: open Claude Code and type /smith — if it autocompletes, you're set.
curl -fsSL https://raw.githubusercontent.com/ATTCKDigital/smith/main/scripts/install.sh | bash
Use this for the full Smith experience. In addition to the skills, it wires up the security / logging hooks and (optionally) the macOS scheduler LaunchAgent. See Installation for details.
Once installed, open any new or existing project in your terminal and run /smith to initialize the vault, bank, and ledger for that project. Once initialized, run /smith-new and describe what you want to build. Smith walks you through requirements, planning, task breakdown, implementation, and PR creation — all without leaving the terminal.
Note: Installation is a one-time global setup into
~/.claude/. Running/smithis a separate per-project step that must be done once in each project before using other commands.
| Category | Commands | Description |
|---|---|---|
| Feature workflow | /smith-new, /smith-explore, /smith-specify, /smith-clarify, /smith-plan, /smith-tasks, /smith-analyze, /smith-implement, /smith-build, /smith-bugfix, /smith-checklist, /smith-finish | End-to-end feature development pipeline |
| Debugging and audit | /smith-debug, /smith-audit | Diagnostic investigation and cross-system audit reporting |
| Knowledge and vault | /smith-vault, /smith-bank, /smith-queue, /smith-todo, /smith-ledger, /smith-reflect | Persistent session logs, idea storage, task queuing, and accumulated learning |
| Reporting | /smith-report, /smith-taskstoissues | Client-facing reports and GitHub issue generation |
| Manifest | /smith-index, /smith-navigate, /smith-migrate-system-paths | Precomputed project index, Haiku navigator, and one-shot path-frontmatter migration for structured context retrieval (see docs/manifest-system.md) |
| Meta | /smith, /smith-update, /smith-constitution, /smith-migrate-specs, /smith-help | Project initialization, version sync, governance, and reference |
npx claudepluginhub attckdigital/smithSkills-first specification-driven development framework with 7 agent skills for planning, implementation, review, and shipping. Natural language activation with intelligent agent orchestration. Includes /plan, /implement, /research commands plus managing-specifications, implementing-features, and reviewing-and-shipping skills.
Spec-driven development using GitHub spec-kit methodology. Constitution-first approach with specify, plan, tasks, and implement phases.
GitHub Spec-Kit integration for Specification-Driven Development - define WHAT and HOW before coding
Autonomous spec-driven development workflow with multi-agent collaboration, specification management, and task orchestration
Specification-driven development workflow: specify → plan → tasks → implement
Context-Driven Development plugin that transforms Claude Code into a project management tool with structured workflow: Context → Spec & Plan → Implement