Cursor rule

1---
2description: Orchestration - routes tasks to the right subagent automatically
3alwaysApply: true
4---
5 
6# Delegation
7 
8After reading memory-bank, check if the current task matches a subagent below. Delegate automatically; do not wait for user to name the agent.
9 
10## When to Delegate
11 
12| Intent / Trigger | Subagent |
13|---|---|
14| Editing `plugin/**`, SimHub SDK, incident detection, iRacing, SimHub dashboards, overlays, Dash Studio | simhub-developer |
15| Writing tech plans, spike plans, technical design docs | simhub-developer |
16| Editing `website/**`, Astro, Tailwind, landing page, docs site, Cloudflare Pages, marketing site | web-developer |
17| "break this down", "scope", "plan feature", brainstorming, user stories, feature ideation, task decomposition | product-owner |
18| Writing PRD specs, feature specs, detailed requirement docs | product-owner |
19| After product-owner creates stories | priority-steward (schedule new tasks) |
20| "what's next?", priorities, sprint, backlog, reprioritize | priority-steward |
21| After completing a task or implementation | priority-steward (update Done, promote Next) |
22| Code review, PR review, diff feedback | code-reviewer |
23| Verify PRD compliance, trace FR-IDs, requirement gaps | prd-compliance |
24| "update memory bank", after major implementation | memory-bank-updater |
25| User gives process/workflow/style guidance | Handle directly: update relevant rules/skills/agents |
26 
27## Document Dependency Orchestration
28 
29When writing multiple dependent documents (PRD specs, tech plans, stories):
30 
31- **Orchestrator handles sequencing.** Read the dependency graph, identify which documents can be written in parallel (shared dependencies resolved), and launch subagents concurrently (max 4).
32- **Tech plans gate PRD specs.** If a PRD spec depends on a tech plan (e.g., spike results), write the tech plan first, then launch the dependent PRDs.
33- **Review at phase boundaries.** Delegate to prd-compliance after completing a phase (not per-document). Fix issues before proceeding.
34 
35## Model Preferences
36 
37- **code-reviewer, simhub-developer, web-developer**: Prefer code-focused model
38- **product-owner, prd-compliance, memory-bank-updater**: Prefer reasoning model
39- **priority-steward**: Default model is fine
40 
41## Rules
42 
43- Memory Bank rule takes priority over this rule when in conflict.
44- All agents must follow `incremental-work.mdc`. Prefer small-to-medium incremental changes; large single-pass artifacts should be escalated to the user first.
45- All agents may search the internet for additional context when project files are insufficient; prefer authoritative sources.
46- All agents may use additional tooling, including command-line workflows and relevant CLIs, when needed to complete tasks.
47- Do not delegate simple one-step tasks; handle them directly.
48- When delegating, pass relevant context from the current conversation in the Task prompt. Subagents read only their domain-specific files (defined in each agent definition), not all memory-bank files.
49 
50## Mandatory Post-Implementation Review Path
51 
52- For implementation in `plugin/**`, use this sequence:
53	1. Delegate implementation to `simhub-developer`.
54	2. `simhub-developer` performs a successful build (msbuild or dotnet build) and self-review, then returns a reviewed, consolidated change set.
55	3. The coding agent executes code changes once (single apply pass) after SimHub self-review is complete.
56	4. **Deploy (mandatory):** The agent that applied plugin code—whether after a simhub-developer handoff or when implementing plugin changes directly—**must** run `plugin/scripts/deploy-plugin.ps1` at the end of that change iteration and report its output. Do not consider plugin implementation complete until deploy has been run and output reported.
57- If plugin changes were made without delegating (e.g. plan execution or small fixes), the agent that made the changes still must run `plugin/scripts/deploy-plugin.ps1` and report output.
58- `code-reviewer` is optional for plugin work unless the user explicitly requests it or confidence is Medium/Low.
59- After implementation, document outcomes in Memory Bank (`activeContext.md`, `progress.md`, and `journal.md` as needed).
60