skills/orchestra-program/SKILL.md
Explain what Orchestra is — a software development lifecycle encoded for agents, its SDLC roots, the pipeline, roles, file structure, status vocabulary, and skill index.
npx skillsauth add mpazaryna/agentic-factory orchestra-programInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Present the orchestra methodology as a complete orientation. This is the program handed to every person who walks into the venue — human or agent, new collaborator or returning contributor.
Orchestra is a software development lifecycle (SDLC) — the same discipline engineers have practiced for decades, encoded so that agents can execute it. The name comes from the roles:
No musician improvises. No step is skipped. The program tells you what will be played, in what order, before the lights go dark.
Every component of Orchestra traces to a proven SDLC discipline. Nothing was invented — it was encoded.
| Orchestra | SDLC Ancestor | The Discipline | |---|---|---| | Score (PRD + Spec + Gherkin) | Requirements → Technical Design → Acceptance Tests | Spec-driven development — written before execution begins | | PRD | Product requirements document | What and why, decided before how | | Spec | Technical design document | How, decided before building begins | | Gherkin | BDD acceptance criteria (Dan North, 2006) | Tests as the contract between intent and implementation | | ADR | Architecture Decision Records (Nygard, 2011) | Decisions as durable artifacts, not conversation | | Implement | Branch / commit / test cycle | Execution against the spec — no improvisation | | Review | Code review + design review | Evidence-based gate before integration | | Merge | Continuous integration gate | The premiere — the permanent record | | Devlog | Engineering journal + commit discipline | Programme notes — context for the audience that follows | | Roadmap | Product roadmap / milestone planning | Vision encoded as checkable states |
The novel layer is the runtime: agents execute against these artifacts as their primary input, rather than humans translating from them.
The score is written before the performance begins.
Agents execute against artifacts, not instructions. A conductor without a score is improvising. An agent without a spec is guessing. The Q&A phases — PRD, then Spec — exist to make the score tight enough that execution requires no interpretation.
Every piece of work moves through the same pipeline in two halves: writing the score, then performing it.
Score — Requirements & Design
| Stage | Skill | What Happens |
|---|---|---|
| Capture | /orchestra-ticket | A work item is captured and scaffolded |
| Define | /orchestra-prd | Q&A with the composer produces the PRD — the what and why |
| Plan | /orchestra-spec | Q&A with the composer produces the spec — the how |
| Validate | /orchestra-gherkin | Acceptance criteria written as testable Gherkin scenarios |
| Approve | (human gate) | Composer signs off — the score is locked |
Performance — Build & Deliver
| Stage | Skill | What Happens |
|---|---|---|
| Implement | /orchestra-implement | A branch is created, spec steps executed, commits made |
| Review | /orchestra-review | Every acceptance criterion checked with evidence — PASS or FAIL |
| Merge | /orchestra-merge | Reviewed branch merged to main, work item closed |
The Q&A before the spec is what makes the rest possible. A tight spec is a spec the implementer can execute without asking questions.
Work items carry a status that tracks exactly where they are in the pipeline. These are stage gates — a recognized SDLC pattern in which each status is a decision point that must be explicitly cleared before the next phase begins. Nothing advances automatically.
draft → approved → in-progress → complete → reviewed → closed
| Status | Meaning |
|---|---|
| draft | PRD or spec written, not yet approved |
| approved | Composer has signed off — ready for next stage |
| in-progress | Implementation underway |
| complete | Implementation done, awaiting review |
| reviewed | Review passed — ready to merge |
| closed | Merged and done |
Everything lives under .orchestra/ in the project root:
.orchestra/
├── roadmap.md — Vision, milestones, active work
├── work/
│ └── {id}-{name}/
│ ├── prd.md — The what and why
│ ├── spec.md — The how
│ └── gherkin-spec.md — Acceptance criteria
├── adr/
│ └── 001-*.md — Architectural decisions, numbered
└── devlog/
└── YYYY-MM-DD-*.md — Developer journals
Work item folders are named {ticket-id}-{short-name}. The ticket ID comes from your issue tracker. If no ticket exists yet, use a descriptive slug.
Score — Requirements & Design:
| Skill | Purpose |
|---|---|
| orchestra-ticket | Capture a work item and scaffold its folder |
| orchestra-prd | Define the what and why for a piece of work |
| orchestra-spec | Generate an execution spec from an approved PRD |
| orchestra-gherkin | Generate Gherkin scenarios from a spec |
| orchestra-plan | Run the full PRD → Spec → Gherkin loop in one session |
| orchestra-roadmap | Read and manage the roadmap |
| orchestra-milestone | Review milestone progress, surface gaps, and reverse-check for orphaned work items not tracked by any milestone |
| orchestra-scaffold | Bootstrap the .orchestra/ structure in a new project |
Performance — Build & Deliver:
| Skill | Purpose |
|---|---|
| orchestra-implement | Execute an approved spec on a branch — includes pre-flight milestone traceability check |
| orchestra-review | Review a completed implementation against its spec |
| orchestra-merge | Merge a reviewed branch to main and close the work item |
| orchestra-eval | Grade skill outputs against acceptance criteria |
Persistent Record — Spans the Lifecycle:
| Skill | Purpose |
|---|---|
| orchestra-adr | Capture an architectural decision |
| orchestra-devlog | Write developer journals |
| orchestra-uml | Generate UML diagrams as Mermaid code |
New to this repo? Run /orchestra-roadmap to see the vision, active milestones, and current work items.
Starting new work? Run /orchestra-ticket to capture it, then follow the pipeline.
Picking up existing work? Find the work item folder under .orchestra/work/, read the spec, check the status, and invoke the skill that matches.
Coming back after a break? Run /orchestra-milestone. It reads roadmap.md, finds the first milestone that isn't Done, checks what's missing from its materials table, and tells you exactly what to do next. One command gets you oriented regardless of where you left off — no need to remember where you were.
When presenting the program on an active project, always follow with the live state. Read:
.orchestra/roadmap.md — current vision and milestones.orchestra/work/*/prd.md — active work items and their statusesPresent this after the program overview. The program tells you what will be played; the live state tells you where tonight's performance stands.
development
Comprehensive Cloudflare platform skill covering Workers, Pages, storage (KV, D1, R2), AI (Workers AI, Vectorize, Agents SDK), feature flags (Flagship), networking (Tunnel, Spectrum), security (WAF, DDoS), and infrastructure-as-code (Terraform, Pulumi). Use for any Cloudflare development task. Biases towards retrieval from Cloudflare docs over pre-trained knowledge.
tools
Send and receive transactional emails with Cloudflare Email Service (Email Sending + Email Routing). Use when building email sending (Workers binding or REST API), email routing, Agents SDK email handling, or integrating email into any app — Workers, Node.js, Python, Go, etc. Also use for email deliverability, SPF/DKIM/DMARC, wrangler email setup, MCP email tools, or when a coding agent needs to send emails. Even for simple requests like "add email to my Worker" — this skill has critical config details.
tools
Build AI agents on Cloudflare Workers using the Agents SDK. Load when creating stateful agents, durable workflows, real-time WebSocket apps, scheduled tasks, MCP servers, chat applications, voice agents, or browser automation. Covers Agent class, state management, callable RPC, Workflows, durable execution, queues, retries, observability, and React hooks. Biases towards retrieval from Cloudflare docs over pre-trained knowledge.
tools
Planning conductor — runs the full PRD → Spec → Gherkin loop in a single interactive session with human approval gates.