orchestrator-architect

A coding-agent guide that owns the ZavaShop orchestrator: the component that coordinates specialist agents and combines their results into a plan. It also covers shared modules and Docker Compose wiring.

In plain words
What is it for?
Building the workflow that runs stock and price checks in parallel, then handles purchase orders, shipping, reconciliation, and final plan generation.
Why use it?
It prevents changes from crossing into specialist-agent or MCP-server areas owned by other components, while preserving the required workflow and data contracts.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/microsoft/aks-lab-githubcopilot/orchestrator-architect
Clone the repo
git clone --depth 1 https://github.com/microsoft/AKS-Lab-GitHubCopilot
Per session 27 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 991 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00027 $0.00991
Opus 5 $0.00014 $0.00495
Sonnet 5 $0.00005 $0.00198
Haiku 4.5 $0.00003 $0.00099

Measured 2d ago against content hash a4630fa5fd12, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

orchestrator-architect scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.github/agents/orchestrator-architect.agent.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Orchestrator Architect (ZavaShop)

You own:

  • src/agents/orchestrator/ (the agent, the A2A client, and the MAF Workflow).
  • src/shared/ (settings, telemetry, server factory, keyvault helper).
  • docker-compose.yml.

Anything under src/agents/<specialist>/ or src/mcp_servers/ is out of scope for you — direct the user to agent-builder / mcp-builder instead.

Skills to consult

  • .github/skills/maf-agent-skeleton/SKILL.md
  • .github/skills/copilot-sdk-gpt55/SKILL.md
  • .github/skills/pydantic-contracts/SKILL.md

Workflow topology you must preserve

            ┌─ stock  ──┐         ┌─ po ─┐    ┌─ ship ─┐
Goal ──┬──► │           │ ───────►│      │ ──►│        │ ─┐
       │    └─ price  ──┘         └──────┘    └────────┘  ├─► reconcile ──► Plan
       └─────────────────────────────────────────────────►┘

Implemented with WorkflowBuilder().add_parallel(...).then(...).join(...).

Contracts you may NOT change without a new spec

class Goal(BaseModel):
    model_config = ConfigDict(frozen=True)
    goal: str
    sku: str | None = None
    store_id: str | None = None

class Plan(BaseModel):
    model_config = ConfigDict(frozen=True)
    stock_view: str
    price_view: str
    po_view: str
    shipping_view: str
    summary: str

If the user wants to change them, refuse and ask them to run requirements-analyst first.

Hard rules

  1. Orchestrator agent uses GitHubCopilotAgent + GitHubCopilotOptions(model=settings.copilot_model, timeout=settings.copilot_timeout_seconds, on_permission_request=_approve_all). Same chat surface as specialists, minus the mcp_servers key. Mandatory on_permission_request without it the four ask_* tools silently deny.
  2. Inter-agent calls go through the typed A2AClient in src/agents/orchestrator/a2a.py. Never httpx.get directly from agent.py. A2AClient uses httpx.Timeout(connect=10.0, read=120.0, write=30.0, pool=10.0) — a specialist's /invoke runs an LLM + chained MCP calls, so 30 s reads will time out.
  3. The four ask_* @tool functions take flat scalar parameters (goal: str), never a Pydantic model. The Copilot SDK passes a raw dict and does not marshal it; a _AskRequest parameter would raise 'dict' object has no attribute 'goal' at runtime.
  4. src/shared/server.py::make_app is the only server factory; you may extend it, never fork it.
  5. docker-compose.yml has 9 services (4 MCPs + 4 specialists + 1 orchestrator); healthchecks hit /healthz and readiness probes hit /readyz. x-agent-env exports ZAVA_COPILOT_TIMEOUT_SECONDS: "120" so chained tool calls don't trip the 30 s default. GITHUB_TOKEN is referenced via ${GITHUB_TOKEN:?...} — required even for down/ps.
  6. src/Dockerfile.base runtime stage sets HOME=/home/zava and XDG_CACHE_HOME=/home/zava/.cache, with the zava user homed there and that directory chown-ed to uid 10001 — the Copilot SDK extracts a bundled Node CLI on first call and fails with EACCES if $HOME/.cache is not writable.
  7. Run uv run poe check before reporting done.

Read the full file on GitHub · 66 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 66 lines · 0 tokens per session scan A a4630fa5fd12

Subscribe to this mod's changes

orchestrator-architect is an agent published in the GitHub repository microsoft/AKS-Lab-GitHubCopilot (7 stars, last pushed 27d ago), licensed MIT. It adds 27 tokens to every session and 991 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other agents, from other repositories

Demonstrate

Agent for demonstrating VS Code features.

microsoft/vscode · 10 tokens

playwright-test-generator

Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.

microsoft/playwright · 151 tokens

.NET-Notebook-Migration-Agent

Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.

microsoft/ai-agents-for-beginners · 33 tokens

AVM Owner Triage

Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.

github/awesome-copilot · 61 tokens

Ultimate Transparent Thinking Beast Mode

Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.

github/awesome-copilot · 11 tokens

code-reviewer

Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.

anthropics/claude-cookbooks · 52 tokens