---
title: "valcraft"
description: "Valcraft is a software factory in your repository: it coordinates the work, keeps the requirements and decisions in ordinary files, and checks the code before shipping so you can see what was built and how it was verifie"
canonical: https://agentpluginsdirectory.com/plugins/valcraft
last-updated: 2026-10-21
---

# valcraft
Valcraft is a software factory in your repository: it coordinates the work, keeps the requirements and decisions in ordinary files, and checks the code before shipping so you can see what was built and how it was verified. Skills: tune (repository configuration), cast (project framing), roadmap (business-facing progress), spec (feature and quick contracts), draft (task plans with MSW), forge (reviewed-task implementation and authorized task PRs), review (report-only plan, code, and evidence review), land (authorized finalization and tracker closure), foreman (named-state delivery coordination), temper (local retrospective production), hone (prompt refinement), distill (prompt distillation), and msw (MSW Kernel document judgment).
- Slug: valcraft
- Publisher: valcraft contributors
- Repository: https://github.com/valzav/valcraft
- Manifest: plugins/valcraft/plugin.json
- Version: 0.10.2
- License: MIT
- Category (editorial): agent-tooling
- Skills: 13 (valcraft-cast, valcraft-distill, valcraft-draft, valcraft-foreman, valcraft-forge, valcraft-hone, valcraft-land, valcraft-msw, valcraft-review, valcraft-roadmap, valcraft-spec, valcraft-temper, valcraft-tune)
- MCP servers: 0
- Stars: 22
- Repository created: 2026-08-11
- Repository last pushed: 2026-10-20
- Publisher type: User
- Listing: https://agentpluginsdirectory.com/plugins/valcraft
- Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json

## What valcraft does, in the publisher's words

Agent skills for spec-driven delivery, packaged as one plugin for Claude Code, OpenAI Codex, OpenCode, and Cursor (Teams or Enterprise with marketplace-import authority). At the center is an agentic delivery loop, draft → review → forge → review → land, run over fresh-context worker agents, inside one Claude Code, Codex, or Cursor session or through an orchestrator over several instances. Around it: cast creates the project frame, spec creates every feature or quick contract, and temper learns from what shipped.

Status: alpha.

Valcraft treats spec-driven development as a repository data model, not a session ritual. Product intent, requirements, decisions, tasks, and evidence live in ordinary files with stable IDs. Agents can resume from those artifacts without inheriting another agent's conversation.

From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/valzav/valcraft/HEAD/README.md

## Skills

- valcraft-cast: Create or retrofit a lean spec-driven development project frame: project instructions, README, product brief, architecture and ADR structure, and other justified scaffold artifacts. Use when starting a project or repository, asking to scaffold or set up SDD, retrofitting project structure, or commi…
- valcraft-distill: Distill a prompt artifact, inline prompt text, a markdown prompt file (system prompt, agent instructions, slash command), a skill directory, or a workflow, into its minimal goal-directed essence, a short structured summary of goal, applicability, steps, and load-bearing constraints. Use when expl…
- valcraft-draft: Write or revise one implementation task plan for a feature T-XXX, quick Q-NNN QT-XXX, or existing task plan; apply MSW, commit the reviewable plan, and hand its exact commit to Review. Use for task planning and plan-review remediation, not implementation, feature specification, review, PR creation,…
- valcraft-foreman: Coordinate or take over the spec-driven delivery loop through fresh Spec, Draft, Review, Forge, Land, and Temper workers. Own runtime readiness, takeover inference, task selection, intermediate tracker state, worker lifecycle, backend returns, report validation, approval gates, recovery, and named-…
- valcraft-forge: Implement one feature task, quick task, passed plan, or small fully specified fix; write code and discriminating tests, remediate code findings, and hand Review an exact target. Use for implementation and authorized task publication, not planning, specification, review, merge, or closure.
- valcraft-hone: Refine an existing prompt artifact, inline prompt text, markdown prompt files (system prompts, agent instructions, slash commands), or a complete skill directory, against the official model prompting guides (Anthropic "Prompting Claude Fable 5" + OpenAI GPT-5.6 best practices). Use when explicitl…
- valcraft-land: Finalize reviewed Valcraft work: verify exact final heads and applicable checks, merge an authorized task or spec PR, close the valid tracker targets, reconcile partial completion, or record and close work completed outside git. Use when the operator or Foreman asks to land, merge, finalize, close,…
- valcraft-msw: Apply the MSW Kernel to a markdown document, a plan, spec, skill, prompt, or any other.md file. Derives the document's contract, judges every claim by the kernel's deletion test, deletes the claims that fail, audits every limit for authority, and reports in the kernel's format. Use when explicitl…
- valcraft-review: Independently review an exact plan commit, code target, or completion-evidence record and return reproduced findings without edits. Use plan mode for plans and feature artifacts, code mode for diffs and refs, and evidence mode for Land records. Use valcraft-hone for prompt audits and valcraft-msw f…
- valcraft-roadmap
- valcraft-spec: Create or resume one complete feature contract triplet, including the first MVP feature, or one quick-task file from exactly one local requirements document, selected GitHub PRD issue, or inline quick brief. Use for feature and PRD intake, staged feature completion, Spec review remediation, and aut…
- valcraft-temper: Produce an evidence-based retrospective over completed work, or synthesize prior Temper reports, and deliver one local, gitignored retrospective report with an exact Review handoff. Use for a retrospective, post-mortem, lessons learned, process-improvement analysis, completed feature or quick-task…
- valcraft-tune: Configure the committed `.valcraft/config.yaml` base and user-local `.valcraft/config.local.yaml` overlay. Use for initialization, invalid configuration, or requested tracker, approval, Foreman, backend, branch, or merge-strategy changes. Tune owns only these files; it does not scaffold or run deli…

Descriptions come from the frontmatter of each SKILL.md, punctuation lightly normalized.
