By cjy5507
Structure, ownership, and continuable state for long-running coding agent work.
Forge Analyst — deep codebase analysis via codebase-memory-mcp for architecture mapping, impact analysis, dependency tracing, and quality assessment
Forge Bug Tracker — manages .forge/holes/ directory, tracks issue lifecycle, provides status summaries
Forge CEO — project intake, scope decisions, blocker resolution, delivery sign-off
Forge CTO — architecture design, technical feasibility, interface contracts, code rules, scale-appropriate decisions
Forge Design Team — meta-agent definition for CTO + Designer peer collaboration via Team pattern during design phase
Run codebase analysis via Analyst agent. Triggers: "forge analyze", "analyze codebase", "code analysis", "architecture analysis", "impact analysis", "dependency trace", and first-principles redesign requests that need an assumption audit plus execution plan.
Use when canceling or cleaning up an active Forge project. Removes worktrees, clears state, preserves code already merged to main.
Continue an existing Forge project after a pause, restart, or interruption. Triggers: "forge continue", "pick up where we left off", "resume".
Use when Forge delivers the completed product. Tech writer generates docs, CEO compiles delivery report with spec coverage and known issues.
Use when Forge reaches the design phase or the user asks for technical architecture, UI spec, contracts, or code rules before implementation. Trigger on requests for architecture design, component specification, design tokens, interface contracts, or phased pre-build planning.
Matches all tools
Hooks run on every tool call, not just specific ones
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.
External network access
Connects to servers outside your machine
External network access
Connects to servers outside your machine
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
A deterministic build-and-repair harness OS for long-running coding work.
Forge gives coding agents phase discipline, ownership, continuable state, and guarded verification routing — so work that spans sessions, days, or interruptions doesn't fall apart.
Coding agents are fast but forgetful. On long-running tasks they lose track of what was decided, who owns what, and where things broke. When a session ends, context disappears. When it resumes, you re-explain everything.
Forge fixes this by writing project state to files — not chat history — and structuring work into phases with explicit gates, decision traces, and verification surfaces, so every session picks up exactly where the last one stopped.
forge info — see where you areShows current phase, what's blocking, and what to do next. One glance, not a wall of fields.
Internally this is backed by node scripts/forge-status.mjs, so the compact dashboard
stays consistent with the runtime's canonical next_action.
Forge: my-saas-app (build)
Phase 4/9 — develop
████████████░░░░ 55%
Next action: Finish lane auth-api — JWT middleware done, testing refresh token flow
Active: auth-api, payment-ui. Blocked: db-schema (waiting on contract review)
Lanes: 2/5 done, 1 blocked
Issues: 1 blocker, 0 major, 2 minor
Tag: forge/v1-design
forge continue — restore saved state and keep goingRestores saved workflow state from .forge/ — phase, lane ownership, blockers,
handoff notes — and routes to the most actionable point. No re-prompting,
no re-explaining what you were doing.
When using separate CLIs such as Codex CLI and Claude CLI, resume explicitly with
forge continue from either host against the same repository checkout. The decision
comes from shared .forge/ state; only the host automation depth differs.
Forge: my-saas-app
Phase 4/9 — develop
3 lanes active, 1 blocked (auth waiting on DB schema)
Next action: Finish lane auth-api — JWT middleware done, testing refresh token flow
It picks the right entry point automatically:
forge analyze — map impact before you touch codeRuns Forge's Analyst flow to produce a durable codebase analysis artifact for the active project. This is especially useful before design on existing code, before lane splitting, and before risky fixes.
Typical uses:
forge eval — turn a run into a scorecardCompares baseline vs harness runs and emits both machine-readable JSON and a markdown
report under .forge/eval/. If you do not provide a harness summary, Forge derives one
from the current runtime, status, traceability, and delivery artifacts.
Typical uses:
.forge/events/eval.jsonlforge analytics — inspect the artifact trailSummarizes .forge/events/, .forge/eval/, .forge/evidence/, and
.forge/delivery-report/ without changing runtime state.
forge verification — inspect the latest verification artifactShows the latest durable verification selection/result written by Forge's verification flow.
Forge now supports a few lightweight runtime controls inspired by ECC-style operational guardrails:
FORGE_HOOK_PROFILE=minimal|standard|strictFORGE_DISABLED_HOOKS=hook-a,hook-bFORGE_STOP_BATCH_CHECKS=0 to disable Stop-time batch validationFORGE_PACKAGE_MANAGER=npm|pnpm|yarn|bun to override auto-detectionforge — start a new project or fix an existing oneForge detects whether you're building something new or fixing something broken, and routes accordingly.
Forge is a deterministic harness OS for engineering teams.
That means Forge optimizes first for:
It does not optimize first for giant skill catalogs or toolbox breadth.
npx claudepluginhub cjy5507/forge --plugin forgePlan iron, verify real. Ironclad planning with independent verification chain. Turns any input into a bulletproof plan, executes with TDD, verifies with independent agents.
Turn Claude Code into a plan-execute-validate loop with parallel work, intelligent retry, and memory
Autonomous AI development pipeline that uses GitHub as a structured knowledge graph for Claude Code agents. Adds /work-on, /review-pr, /quality-gate, /orchestrate, and 20+ pipeline commands.
Harness-native ECC plugin for engineering teams - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, MCP conventions, and operator workflows for Claude Code plus adjacent agent harnesses
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
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.