By GregoryHo
Orchestrate AI coding agent workflows with structured planning, parallel task execution, test-driven development, and automated code review across isolated git worktrees, plus session learning and debugging.
Use this agent as the `cross-artifact-alignment` axis of the `arc-auditing-spec` skill. Spawned in parallel with the other two audit axes during Phase 1 fan-out. Read-only by tool grant. Examines semantic alignment between `design.md`, `spec.xml` (+ `details/*.xml`), `dag.yaml`, and the D6 anchor artifacts (`decisions.yml`, `product/vision.md`) of a single arcforge spec family; emits findings addressing misalignment across two or more artifacts (incl. broken delta→decision links and unresolvable `principle_ref`→`P-n`), NOT issues internal to a single artifact.
Use this agent as the `internal-consistency` axis of the `arc-auditing-spec` skill. Spawned in parallel with the other two audit axes during Phase 1 fan-out. Read-only by tool grant. Examines contradictions, dangling references, or self-referential prose **within a single artifact** of the arcforge spec family (e.g., a requirement contradicting its own `<consumes>` entry in the same spec.xml). Does NOT emit cross-artifact findings — those belong to the `cross-artifact-alignment` axis.
Use this agent as the `state-transition-integrity` axis of the `arc-auditing-spec` skill. Spawned in parallel with the other two axes during Phase 1 fan-out. Read-only by tool grant. Examines whether `dag.yaml` recorded state (epic status, worktree pointer, feature status) matches on-disk evidence — `.arcforge-epic` marker files, worktree directory presence, feature-status files. Does NOT examine git history (out of scope; see below).
Use this agent when a completed feature, task, or major project step needs review against the original plan and coding standards — e.g. after finishing a numbered plan step or a significant implementation chunk.
Use this agent to investigate and fix bugs with a 4-phase methodology that isolates root cause before any fix attempt — for failing tests or unexpected behavior where the cause is not obvious.
Use when executing task lists where each task requires isolated execution
Use when the user asks to journal session reflections (/arcforge:arc-journaling), when the PreCompact hook triggers, or at end of a significant work session
Use when optional learning is enabled and observations should become reviewable candidates, inactive drafts, and explicitly activated artifacts.
Use when running autonomous unattended loops — cross-session execution of DAG tasks without human intervention
Use when implementation is complete and all tests pass, and you need to decide how to integrate — works for both epic worktrees (.arcforge-epic present) and regular branches. Step 0 discriminates on the marker.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
arcforge is a minimal, composable skill toolkit for Claude Code, Codex, Gemini CLI, and OpenCode. It gives agents lightweight routing, structured SDD artifacts, and eval-backed quality gates without turning every task into a mandatory workflow.
AI coding agents are powerful but uneven. Left to their defaults, they skip design, ignore review, and lose context across sessions. Heavy always-on process creates a different failure mode: the agent follows workflow ceremony when a direct answer or isolated eval would be better.
arcforge solves this with a small composable toolkit. Skills are available in the session, but they are selected when useful: design when intent is unclear, structured specs when artifacts matter, TDD/debugging/review when implementation risk is present, and verification before completion claims.
The outcome: your agent has disciplined workflows when the task justifies them, while preserving direct execution, harness isolation, and small-task speed when a workflow would be overhead.
ArcForge is split into three layers:
When your coding agent starts a session, arcforge's hooks inject a minimal bootstrap: ArcForge is available, ARCFORGE_ROOT is set, and agents should prefer the smallest useful workflow. Specific skills are read or invoked on demand.
Once a design is approved, ArcForge can build a clear implementation plan and then execute tasks with a two-stage review (spec compliance, then code quality). For larger work, it can create parallel git worktrees so epics can run in isolation.
Skills are tools, not laws. You can enter through arc-using for routing help or call any skill directly when you already know the needed workflow.
Note: Installation differs by platform. Claude Code has a built-in plugin system. Codex and OpenCode require manual setup.
Register the marketplace:
/plugin marketplace add GregoryHo/arcforge
Install the plugin:
/plugin install arcforge@arcforge-dev
Check that commands appear:
/help
# Should see:
# /arcforge:arc-brainstorming - Design exploration
# /arcforge:arc-writing-tasks - Break epics or features into executable tasks
# /arcforge:arc-executing-tasks - Execute tasks with checkpoints
Every skill is directly invocable by name — /arcforge:arc-<name> (e.g. /arcforge:arc-tdd, /arcforge:arc-debugging). Unsure where to start? Invoke /arcforge:arc-using for routing help.
Tell Codex:
Fetch and follow instructions from https://github.com/GregoryHo/arcforge/blob/main/.codex/INSTALL.md
Detailed docs: docs/README.codex.md
Tell Gemini CLI:
Fetch and follow instructions from https://github.com/GregoryHo/arcforge/blob/main/.gemini/INSTALL.md
Detailed docs: docs/README.gemini.md
Tell OpenCode:
Clone https://github.com/GregoryHo/arcforge to ~/.agents/arcforge, then create directory ~/.config/opencode/skills, then symlink ~/.agents/arcforge/skills to ~/.config/opencode/skills/arcforge, then create directory ~/.config/opencode/plugins, then symlink ~/.agents/arcforge/.opencode/plugins/arcforge.js to ~/.config/opencode/plugins/arcforge.js, then restart opencode.
Detailed docs: docs/README.opencode.md
These are the most frequently used commands:
| Command | Purpose | When to Use |
|---|---|---|
/arcforge:arc-using | Routing help + skill index | When unsure which skill or workflow applies |
/arcforge:arc-brainstorming | Design exploration | When starting new work or clarifying requirements |
/arcforge:arc-writing-tasks | Break down into tasks | When you have a clear spec and need executable steps |
/arcforge:arc-executing-tasks | Run task list | When tasks are ready and you want to implement |
/arcforge:arc-journaling | Session journaling | At end of session to capture reflections |
/arcforge:arc-reflecting | Analyze patterns | After 5+ journal entries to summarize learnings |

npx claudepluginhub gregoryho/arcforge --plugin arcforgeHigh-performance multi-line statusline for Claude Code with setup commands and diagnostics
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review