By mwarger
Spec-from-evidence pipeline with autoresearch loops: build subject-named specs with sub-agent fanout, provenance tracking, adaptive clarification, blind scoring, self-replicating bead cycles, and a canonical readiness contract
Create a subject-named specification from any evidence source using a reducer-based Forge workflow. Use this when the user wants a planning-ready spec, a clean-room reverse spec, or an evidence-first feature spec with sub-agent fanout, provenance tracking, adaptive clarification, speculative variants, and a canonical readiness contract.
Render the implementation handoff for a subject spec. Use this when the canonical readiness verdict already exists and you need either a real implementation plan for eligible runs or a withheld handoff plus bounded options for blocked runs.
Reduce typed sub-agent outputs into canon and verify the subject spec before readiness promotion. Use this when you need deterministic merges, provenance review, contradiction handling, verification loops, section drafting, or final review passes across the spec.
Stress-test a subject spec for ambiguity, gaps, contradictions, and untestable claims using dynamic agent teams. Use this when the spec has passed completeness and synthesis-review gates and needs adversarial validation before readiness promotion.
Stamp a self-replicating four-bead autoresearch loop (doer/judge/arbiter/strategist) for autonomous iterative improvement of any artifact. Use after interactive intake when you have a program and want to run an autonomous improvement loop with blind scoring. Triggers on: autoresearch, research loop, autonomous loop, overnight loop, iterative improvement.
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.
Turn a feature request into an implementation-ready spec — stress-tested before any code is written. Then run autonomous improvement loops with blind scoring.
AI-generated specs hallucinate decisions. They fill in blanks with plausible-sounding answers instead of flagging them as open questions. The implementer — human or agent — hits gaps, invents answers on the fly, and ships something that doesn't match what anyone intended.
Nobody stress-tests the spec before handoff. By the time contradictions and missing edge cases surface, they're already bugs.
.ralph-tui/progress.md are consumed by the planning pipeline, so each spec benefits from patterns discovered in prior implementations.A subject spec — a specification document named after its subject (e.g. auth-session-system.md) — with sidecar artifacts:
UBIQUITOUS-LANGUAGE.md) for consistent domain terminologySee a finished example:
specs/feature-flags-system.mdwith full artifacts inspecs/_artifacts/feature-flags-system/.
claude) or Codex (codex)bash, git, python3 (for manual/shell install only)Claude Code (recommended):
/plugin marketplace add mwarger/forge
/plugin install forge@forge-dev
All 10 skills are available immediately. Update with /plugin update forge.
Codex:
Fetch and follow instructions from https://raw.githubusercontent.com/mwarger/forge/main/.codex/INSTALL.md
Start a new session after install and just describe what you want. Forge triggers automatically — no special syntax needed:
Forge a feature to add real-time notifications
Forge a fix for the auth token refresh bug
Forge a migration from REST to GraphQL
Forge a refactor of the payment processing module
Reverse engineer this repo into a spec
Or be more explicit if you prefer:
Create a subject-named spec for this feature request using Forge.
Use the forge-orchestrator skill on this codebase.
Works in both Claude Code (claude) and Codex (codex). The orchestrator skill triggers and routes into the focused sub-skills automatically.
Manual / shell:
./install.sh
This installs the skills into your detected skills dir and a helper command forge-pack.
Useful env vars:
FORGE_INSTALL_MODE=copy ./install.sh
FORGE_SKILLS_DIR="$HOME/.claude/skills" ./install.sh
FORGE_SKILLS_DIR="$HOME/.codex/skills" ./install.sh
Remote install:
curl -fsSL "https://raw.githubusercontent.com/mwarger/forge/main/install.sh?$(date +%s)" | bash
Platform-specific install docs:
.claude/INSTALL.md.codex/INSTALL.mdThe installer auto-detects the target platform:
~/.claude/ exists → installs to ~/.claude/skills/~/.codex/ exists → installs to ~/.codex/skills/FORGE_SKILLS_DIRBy default the installer uses symlinks so local edits in this repo are visible immediately in the installed skills.
npx claudepluginhub mwarger/trace --plugin forgeQRDS-PI cycle: interview-driven SDLC on bd bead state machines
mwargerEvidence-to-artifact pipeline with autonomous improvement loops: spec intake, adaptive clarification, adversarial review, self-replicating autoresearch cycles (doer/judge/arbiter/strategist), and canonical readiness gates
Spec-driven development with task-by-task execution. Research, requirements, design, tasks, autonomous implementation, and epic triage for multi-spec feature decomposition.
tzachbonSkills-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.
jeanlucianoSpecification-Driven Development with Process Discipline for Claude Code
rhussSpecification-driven development workflow: specify → plan → tasks → implement
deepeshBodhGitHub Spec-Kit integration for constitution-based spec-driven development (7-phase workflow)
feiskyerSpec-driven development workflow - transforms ideas into structured specifications (requirements, design, tasks) before implementation.
sanmak