coldpress-os

An AI-native development framework that drives the full lifecycle of AI-assisted software projects — from bootstrap to deployment to evolution.
coldpress-os is Agent Skills-compatible and implements the Spec → Plan → Implement → Review spine, expanded to 11 phases (Shape A — v0.3.0) with sacred-doc governance, multi-agent orchestration, and the forward-carry quartet.
The 11 phases: Bootstrap · Discovery · Tech Stack · Planning · Design · Architecture · Breakdown · Implementation · Deployment · Operate · Evolve. (Phases 5 & 6 added in v0.3.0-alpha — see CHANGELOG.)
One framework. One format. Two runtimes. Pluggable stack packs.
Install
npm install -g @coldpress/core
Requires Node.js >=22 and Claude Code (the CLI or the Agent SDK) on your machine. Python >=3.10 is additionally required for coldpress graph rebuild and the document-ingest skill; install on demand with pip install graphifyy markitdown docling when you reach Phase 2.
Quick Start
coldpress init my-project
cd my-project
claude
coldpress init prompts you for a project name, slug, and user name, then scaffolds:
coldpress.yaml — project configuration (Phase-1 fields only; later phases write back as you progress)
CLAUDE.md — framework routing for Butler (your main Claude Code session)
.claude/agents/ — 11 subagent definitions (analyst, pm, ux-designer, architect, developer, qa, scrum-master, communicator, reviewer, devops, valet)
.claude/skills/ — ~128 thin wrappers pointing at canonical skills (built from 85 atomic + 67 lifecycle SKILL.md sources)
_context/ — produced artefacts (planning, design, implementation, testing, tracking, handoffs, audit, sacred docs)
_input/ — raw inputs, legacy refs, vendor drops, assets
secure/ — credential manifest + pre-commit secret-scan hook
coldpress-os/ — framework files (read-only; upgraded via coldpress update)
AGENTS.md + .cursor/rules/ + .roomodes + .openhands/microagents/ + .clinerules/ — interop outputs so the project feels native in any agent tool
New to coldpress-os? Read the First 10 Minutes or see the TaskPulse example for a full lifecycle demo.
Runtime compatibility
coldpress-os's .claude/ tree loads unchanged under both runtimes:
- Claude Code CLI (interactive dev-time) —
claude in the project directory.
@anthropic-ai/claude-agent-sdk (programmatic / CI pipelines) — import the SDK, point it at the project; all subagents and skills resolve.
See test/agent-sdk-compat.test.ts for the compatibility smoke test.
Commands
coldpress init [project-name] Scaffold a new project
coldpress update Regenerate interop outputs (AGENTS.md, Cursor, Roo, OpenHands, Cline)
coldpress feedback Open GitHub Issues in your browser
coldpress upgrade Print upgrade instructions
coldpress --version Print installed version
Lifecycle phases (Shape A — v0.3.0-alpha)
11 phases, structured as Spec → Plan → Implement → Review expanded with the forward-carry quartet and a silent-divergence guard at the Design → Architecture boundary.
| 1 | Bootstrap | butler | Project init, intake, graph-prime, working-mode detection |
| 2 | Discovery | @analyst | Research, personas, context, idea validation, product brief |
| 3 | Tech Stack | @architect | Stack discovery, evaluation rubric, locking, env provisioning |
| 4 | Planning | @pm | PRD authoring + section-scoped re-validation (PRD-only post-split) |
| 5 | Design (new) | @ux-designer | UX spec, brand guidelines, prototype, narrative, design-deltas |
| 6 | Architecture (new) | @architect | Sacred architecture.md + ADRs incl. REQUIRED ADRs for flagged design-deltas (silent-divergence guard) |
| 7 | Breakdown | @pm + @scrum-master | Epics, stories, parallelization strategy, PERT chart |
| 8 | Implementation | @developer + @qa | Dev-story, code-review, QA automation, wave orchestration |
| 9 | Deployment | @devops | Readiness checks, security scan, dep audit, observability, deploy |
| 10 | Operate | @devops | Sprint status, correct-course, incident-response, ops-deltas |
| 11 | Evolve (final) | @reviewer | Retrospective, product evolution, innovation strategy → next-iteration Phase 1 |
Forward-carry quartet — design-deltas (P5 exit) · architecture-deltas (P7 entry) · implementation-deltas (P11 retrospective) · ops-deltas (P11 retrospective). Each delta resolves via four reconciliation options: accept_into_prd, reject, flag_for_architecture_ADR, park_for_phase_11.
Subagents
11 subagents under Shape A. Two added in v0.3.0-alpha (@reviewer, @devops) take Phase 11 and Phases 9–10 respectively.
analyst | sonnet | 2 | Research, personas, idea validation, synthesis, product brief |
architect | opus | 3, 6 | Tech stack (P3); sacred architecture + ADRs incl. silent-divergence-guard (P6) |
pm | sonnet | 4, 7 | PRD lifecycle (P4); epic oversight + breakdown (P7) |
ux-designer | sonnet | 5 | UX design spec, brand guidelines, prototypes, narrative |
scrum-master | haiku | 7 | Sprint planning (sub-dispatched from @pm at P7) |
developer | sonnet | 8 | Implementation (standard or quick mode) |
qa | sonnet | 8 | Testing (rapid or strategic mode); sub-dispatched from @developer |
devops | sonnet | 9, 10 | Deployment (P9 ship-path) → operate (P10 steady-state); two phase-modes |
reviewer | opus | 11 | Retrospective, product evolution, innovation strategy (final phase) |
communicator | sonnet | cross-cutting | Documentation, narratives, presentations |
valet | sonnet | meta | Framework evolution, meta skills |
Each subagent's definition lives at .claude/agents/<slug>.md in your scaffolded project. Pattern 7 transitions (19 transition records) formalise agent boundaries at phase entry / exit / sub-phase / reconciliation handoffs.
Key concepts
- Subagents are real Claude Code agents with independent context windows, tools, and models. Butler (your main session) dispatches them via the Agent tool. Not prompt-persona costume changes.
- Skills are the atomic unit of work. Each is self-contained with frontmatter + step-files + references.
- Stack packs are pluggable skill sets for specific technology stacks. Six ship in-tree at v0.3.0-alpha:
vibe-coder-fullstack (Convex + Next.js + Clerk), cli-npm-publishable (TypeScript + tsup + Vitest), browser-extension (WXT + Manifest V3), static-single-page, static-multipage-blog (Astro variants), and seo-pack (cross-archetype audit/content/local/schema/technical). Activated via stack_pack: in coldpress.yaml after Phase-3 stack-locking.
- Sacred documents —
_context/sacred/{context,tech-stack,prd,architecture,pert-chart}.md — are protected by governance change workflows in coldpress-os/governance/.
_context/ vs _input/ — produced artefacts vs material fed into the project. Inputs are not written by any skill.
Architecture
@coldpress/core/
├── src/ # CLI + generators (TypeScript)
├── template/ # Scaffolded into consumer projects (11-subagent set)
├── lifecycle/ # 11-phase Shape A skill organisation (67 SKILL.md)
├── skills/ # ~85 atomic reusable skills + 6 stack packs
├── agents/ # Subagent schema + registry
├── schemas/ # ~30 JSON / Zod schemas (sacred docs, handoffs, audit, design)
├── orchestrator/ # Parallelization engine specs
├── governance/ # Sacred-doc change workflows + 4 reconciliation options
├── data/ # Portable knowledge assets (CSV/YAML, method playbook)
├── templates/ # Document / design / infrastructure / prompt-snippet templates
├── plugin/ # Claude Code plugin marketplace tree (~128 spec-compliant SKILL.md)
└── docs/ # Framework documentation
Claude Code plugin marketplace
/plugin marketplace add coldpress-labs/coldpress-os
/plugin install @coldpress/core
Installs the full skill library as an Agent Skills–compliant plugin. Source lives under plugin/skills/, regenerated by npm run build:skills from the rich internal SKILL.md corpus.
Updating
npm update -g @coldpress/core
Inside an existing project:
coldpress update
How it relates to other tools
- Anthropic Agent Skills — compatible + complementary. Coldpress-os's skills conform to the Agent Skills SKILL.md spec (emitted to
plugin/skills/). Installable as a Claude Code plugin via /plugin marketplace add coldpress-labs/coldpress-os. We wrap Anthropic's first-party skills where they overlap with our subagents (docx / pdf / pptx / xlsx under @communicator, webapp-testing under @qa, mcp-builder under @architect, skill-creator under @valet). See docs/agent-skills-compatibility.md.
- BMAD-METHOD — lineage. Coldpress-os is a soft-fork of BMAD v6.2.2 (pinned upstream, no rebase — cherry-pick only). Full attribution in
NOTICE.md.
- GitHub Copilot Workspace — shared spine, richer expansion. Both implement Spec → Plan → Implement → Review. Coldpress-os expands the 4-stage spine into 11 phases (Shape A) with sacred-doc governance and typed inter-phase handoffs. See
docs/spec-plan-implement-review-mapping.md.
- Graphify — vendored upstream. Indexer + retrieval core at
graph/vendor/graphify/. coldpress graph rebuild invokes it; coldpress graph query reads the output.
Documentation
| First 10 Minutes | Hands-on setup walkthrough — zero to a running project |
| Example Walkthrough | Full lifecycle demo with a sample project (TaskPulse) |
| Troubleshooting & FAQ | Common issues and solutions |
| Architecture | Internal technical reference |
| Spec → Plan → Implement → Review mapping | 11-phase Shape A lifecycle grouped under the industry-standard spine |
| Agent Skills compatibility | How coldpress-os fits the Anthropic Agent Skills ecosystem |
| Template Registry | Every template — by category, phase, consuming skill |
| Skill Discovery Index | Every skill grouped by phase + cross-cutting utilities |
| Subagent × Phase Matrix | 11-subagent × 11-phase reference — which subagents do what in which phases |
| Phase → Subfolder Mapping | Canonical _context/* destinations per phase |
coldpress.yaml schema | Per-field phase ownership + write-back contract |
| Interop generator | AGENTS.md, Cursor, Roo, OpenHands, Cline — spec + tool translation |
| SKILL.md generator spec | Agent Skills spec compliance + field mapping |
| Anthropic skill wrapping | Where coldpress-os delegates to Anthropic's first-party skills |
| Secure pattern | secure/manifest.yaml + credential loader pattern |
| Decision trees | How Butler routes your intent to skills |
| Flow map | Visual mapping of phases, skills, and subagents |
| Glossary | Defined terms |
| Stack pack guide | Authoring a new technology stack pack |
| Subagent customization | Modes, overrides, custom agents |
| Step-file spec | Format specification for workflow step files |
Contributing
Feedback flows from projects to coldpress-os via GitHub Issues and PRs. Use coldpress feedback to open the issue form in your browser. See CONTRIBUTING.md for the full guide and CODE_OF_CONDUCT.md for community standards.
Security: please do not open public issues for vulnerabilities. See SECURITY.md for private disclosure.
Acknowledgments
coldpress-os stands on the shoulders of four open-source projects, each of which contributed substantial ideas, code, and craft. All are MIT-licensed, and all are credited in full in NOTICE.md and docs/attribution-audit.md.
- BMAD-METHOD by BMad Code, LLC — the core agent-skill-workflow architecture, document templates, and most utility and review skills. coldpress-os is a direct derivative of BMAD v6.2.2; this framework would not exist in its current form without theirs.
- Creative Intelligence Suite (CIS) by BMad Code, LLC — contributes the brainstorming, design-thinking, problem-solving, innovation-strategy, and storytelling workflows used across Discovery and Planning.
- BMAD-METHOD-WDS (Whiteport Design System) by Mårten Angner / Whiteport Collective — contributes the opinionated UX design workflow (wds-0 through wds-8), design templates, trigger maps, and scenario-driven design methodology powering the
ux-designer subagent.
- Graphify by Safi Shamsi — the indexer + retrieval core behind coldpress-os's knowledge graph. Graphify v4 is vendored verbatim at
graph/vendor/graphify/ and provides tree-sitter-based AST indexing across 20+ languages plus markdown/document ingestion with graph extraction.
"BMad", "BMad Method", "BMad Core", "Whiteport", and "Whiteport Design System" are trademarks of their respective owners. coldpress-os is an independent project and is not affiliated with or endorsed by any of the above.
License
MIT — see LICENSE.
Built by ColdPress Labs. Built with intention. Scaled with purpose.