Comprehensive knowledge of amplihack framework architecture, patterns, and usage
Comprehensive knowledge of amplihack framework architecture, patterns, and usage
npx skills add https://github.com/rysweet/amplihack-rs --skill amplihack-expert이 명령을 Claude Code에 복사하여 붙여넣어 스킬을 설치하세요
Production-ready skill for detecting silent degradation across codebases. Uses multi-wave audit system with 6 specialized category agents, multi-agent validation panel, and convergence detection.
Multi-pass code philosophy audit skill. Performs three distinct passes over target code to verify adherence to the project's PHILOSOPHY.md principles. Produces a structured report with findings and delegates fixes to dev-orchestrator. This is an advisory audit-only skill — it does not modify code directly.
Default task orchestrator for all development and investigation work. Classifies tasks, decomposes into parallel workstreams if appropriate, and routes execution through the recipe runner. Replaces ultrathink-orchestrator.
Builds comprehensive, living code-atlases as multi-layer architecture documents derived from code-first truth. Defaults to both Graphviz DOT and Mermaid. User can override to single format. Language-agnostic (Go, TypeScript, Python, .NET, Rust, Java). Files issues with 'code-atlas-bughunt' label. Treats atlas-building as a multi-agent bug-hunting journey: graph-form reasoning exposes structural bugs, route/DTO mismatches, orphaned env vars, dead code paths, and stale documentation that linear review misses. Three-pass bug hunt with per-journey PASS/FAIL/NEEDS_ATTENTION verdicts. Use when: creating architecture documentation, investigating unfamiliar codebases, hunting structural bugs, setting up CI/CD diagram refresh, or publishing to GitHub Pages/mkdocs.
Generates comprehensive GitHub activity reports across all authenticated accounts. Gathers repos, PRs, features, and themes for configurable time periods (1/5/7/30/90 days). Produces shareable markdown with tables, mermaid charts, and executive summaries. Can create a private repo with GitHub Actions automation and GitHub Pages aggregation site. Use when: "github report", "work report", "activity summary", "what did I work on", "gh-work-report", "show my github activity".
Move the active amplihack CLI session (Copilot/Claude/Amplifier) to a fresh azlin-managed VM, preserving auth, plugins, skills, plan.md, todos, and conversation history. Resumes the session in a detached tmux on the destination host.
| name | amplihack-expert |
| description | Comprehensive knowledge of amplihack framework architecture, patterns, and usage |
| version | 2.0.0 |
| author | amplihack team |
| tags | ["amplihack","framework","architecture","workflows","agents","commands"] |
| triggers | {"keywords":["amplihack","dev-orchestrator","smart-orchestrator","/dev","workstream","recipe runner","workflow","DEFAULT_WORKFLOW","/.claude/","agent system","specialized agent","command system","hook system","continuous work","skill system","extensibility"],"patterns":["How does amplihack.*work","What is.*amplihack","amplihack.*(architecture|structure|design)","How do I.*amplihack","What.*agents.*available","How.*orchestrat","When.*use.*dev|When.*use.*/dev"],"file_paths":["~/.amplihack/",".claude/agents/",".claude/commands/",".claude/workflow/",".claude/skills/"]} |
| token_budget | {"skill_md":800,"reference_md":1200,"examples_md":600,"total":2600} |
| disclosure_strategy | {"quick_answer":"SKILL.md only","architecture_question":"SKILL.md + reference.md","how_to_question":"SKILL.md + examples.md","comprehensive":"All three files"} |
| references | ["reference.md: Comprehensive architecture details","examples.md: Real-world usage scenarios","@~/.amplihack/.claude/context/PHILOSOPHY.md: Core principles","@~/.amplihack/.claude/workflow/DEFAULT_WORKFLOW.md: Main workflow"] |
Engineering system for coding CLIs (Claude, Copilot, Amplifier): 5 mechanisms, 23-step workflow, 30+ agents, 80+ skills. Core entry point: /dev <task> — unified task orchestrator with auto-classification, parallel workstream detection, and goal-seeking execution loop.
| Command | Purpose | Use When |
|---|---|---|
| /dev | Orchestrate | Any non-trivial task |
| /analyze | Review | Check compliance |
| /fix | Fix errors | Common error patterns |
| /amplihack:ddd:* | Doc-driven dev | 10+ file features |
| /multitask | Parallel tasks | Sprint/batch work |
| Agent | Role |
|---|---|
| architect | Design |
| builder | Code |
| reviewer | Quality |
| tester | Tests |
| Name | Steps |
|---|---|
| DEFAULT_WORKFLOW | 23 |
| INVESTIGATION | 6 |
| FIX | 3 |
Quick: SKILL.md | Architecture: reference.md | How-To: examples.md | Deep: all
5 Mechanisms: Workflow (process), Command (entry), Skill (auto), Agent (delegate), Hook (runtime)
Composition: Commands → Workflows → Agents → Skills
Execution: /dev orchestrates — classifies task, detects parallel workstreams, executes via smart-orchestrator recipe, goal-seeking loop (3 rounds max)
Entry point: /dev <task> → dev-orchestrator skill → smart-orchestrator recipe → default-workflow recipe
Entry point: /dev <task> or any non-trivial prompt
Routing:
amplihack:core:analyzer responds directlyamplihack:core:builder (complex)Execution flow:
Session tree: Prevents infinite recursion
AMPLIHACK_MAX_DEPTH=3 (default) — increase to 5 for deep orchestrationAMPLIHACK_MAX_SESSIONS=10 (default) — max concurrent sessions per treeStatus signals:
GOAL_STATUS: ACHIEVED — all criteria metGOAL_STATUS: PARTIAL -- [gaps] — another round will runGOAL_STATUS: NOT_ACHIEVED -- [reason] — final failure statusAMPLIHACK_MAX_DEPTH env varamplihack orch helper: Native Rust JSON extraction (source: crates/amplihack-cli/src/commands/orch.rs)When an upstream API (Copilot, Anthropic) returns a rate-limit error during an
agent step, the recipe runner treats it as a hard failure by default. Non-critical
steps such as step-06c-documentation-refinement set continue_on_error: true
so that documentation polish does not abort the entire recipe. For critical steps,
the agent runtime retries with exponential back-off (configured by the SDK adapter).
If a rate-limit error aborts your workflow, check which step failed — if it is a
polish/review step, adding continue_on_error: true is the recommended fix.