Product Team — Domain Orchestrator & Discovery Loop
This skill is a domain orchestrator for product work. It routes product inquiries to one of 16 specialized sub-skills (RICE, OKRs, UX research, etc.) via a deterministic router, and it can run continuous-discovery loops with machine-checkable gates (cadence tracker and OST linter). It produces digests, discovery logs, linted opportunity solution trees, and goal-scale plans.
✨ What it does
- Routes product inquiries to one of 16 sub-skills using a deterministic signal router.
- Runs a continuous-discovery loop with a cadence tracker (health score) and an OST linter (structural rules).
- Enforces evidence-based decisions: no roadmap without linted OST, no insights from single participants.
- Provides forcing questions per lane to lock key decisions before execution.
- Integrates with the agent-harness for goal-scale runs and close-out gates.
- Generates digests, discovery logs, cadence reports, and linted OST files.
🎯 When to use it
- When you need to prioritize features, set OKRs, or plan a product experiment.
- When you want to run a continuous discovery loop with weekly cadence and validation gates.
- When you have a product goal that spans multiple domains and need to route to the right sub-skill.
- When you need to audit an opportunity solution tree or ensure discovery health.
- When you want to turn a validated spec into a runnable repo with verification.
🚀 How to use
Trigger by describing a product goal or symptom, e.g., 'help me prioritize', 'plan a product experiment', 'run the discovery loop', or 'is our OST sound'. The skill forks context and runs a router script to classify the goal; it may ask one clarifying question if ambiguous. Provide a clear goal statement or a discovery log file. Example prompts:
Help me prioritize our feature backlog using RICE.
Run the discovery loop for our new onboarding flow.
📄 Output: Depending on mode: a sub-skill artifact with a digest, a discovery_log.json with cadence report and linted ost.json, or a .agent-harness/plan.json with state and close handoff.
📦 Add this skill to Claude Code
# 1. Get the skills repo
git clone --depth 1 https://github.com/alirezarezvani/claude-skills /tmp/claude-skills
# 2. Copy this skill into your project (or ~/.claude/skills for all projects)
mkdir -p .claude/skills
cp -r /tmp/claude-skills/product-team/skills/product-skills .claude/skills/product-skillsSkill source: product-team/skills/product-skills/SKILL.md
⚠️ Good to know
Requires the bundled sub-skills and standalone plugins to be installed; discovery data must be in JSON format in the workspace.
❓ FAQ
What happens if the router cannot determine the sub-skill?
It asks one clarifying question naming candidate skills with a recommended answer, or asks you to restate the goal with the deliverable named.
How does the discovery loop decide when to stop?
It stops when the cadence is HEALTHY and an assumption is validated, then graduates to experiment design or PRD; if DORMANT for 4+ weeks, it escalates to the product lead.
Can I cite an opportunity solution tree in a roadmap without linting?
No, the hard rules state no roadmap item cites the OST unless the linter exits 0.
🤖 Overview, features, install steps and FAQ were generated from the project's SKILL.md on Sep 4, 2026. Always check the original source before running commands.