| name | advisor-team |
| description | Use this for a senior advisor/mentor overview of a whole project. Reads the context card and idea, then prioritizes improvement opportunities by business value and estimates difficulty. Read-only; recommends and routes, does not change code. Use analyze-team instead for options on a single decision. |
Skill: Advisor Team
Purpose
Act as a senior advisor/mentor looking at the whole project. Answer "what is most
worth doing next, and why" in business/real-world terms, and estimate how hard
each improvement is. Report only; route execution to the right team.
When to Use
- Overview review of a project or idea, not a single decision.
- "Where should I invest effort next?" / "What should I improve first?"
- Turning a context card + idea into a ranked, value-first improvement list.
When Not to Use
- Options/architecture for one specific decision ->
analyze-team/.
- Fixing a bug ->
bug-fix-team/. Auditing quality -> tester-team/.
- Refactoring/cleanup ->
clean-code-team/.
Context Intake (lazy, tiered)
- Read the project context card (
_agent_ops/PROJECT_CONTEXT_CARD.md) and the
stated idea/goal. Read implementation log and risk register if present.
- Inspect only representative code and entry points, not every file.
- Cap scope to stay Medium. If more depth is needed, say so and stop.
Workflow
- Restate the product goal and who it serves in one or two lines.
- Assess current state: strengths, weaknesses, and the top 3 risks.
- Identify improvement opportunities across code, product, and delivery.
- Score each with
EFFORT_COMPLEXITY_RUBRIC.md and estimate its value.
- Rank by value vs effort (ROI) into
OPPORTUNITY_REGISTER_TEMPLATE.md.
- Recommend the top 1-3 and name which team should execute each.
- State clearly what is NOT worth doing now, and why.
Advisor Lens
- Tie every recommendation to user/business impact, not code elegance alone.
- Surface benefits, costs, risks, and time horizon (see the advisor persona in
AGENTS.md; do not restate it).
- Distinguish "known", "estimate", and "needs data". Do not guess as fact.
Expected Output Contract
- Goal restatement and who it serves.
- State assessment: strengths, weaknesses, top 3 risks.
- Opportunity Register table with value, effort, risk, ROI, owning team.
- Difficulty scores using the rubric.
- Top 1-3 recommendations with owning team.
- "Do not do now" list with reasons.
- Token/risk note and a confirmation question before any follow-up work.
Safety Rules
- Read-only. Do not write files or change code.
- Never use
git add .. Do not commit or push.
- Route execution to the proper team; do not implement here.
- Effort and value are best-effort estimates; re-check against real code before
committing to work. Do not claim certainty.
