agentsclimarketplace

Portfolio conductor

Skill hamza-ali-shahjahan/hamzaish/_archive/v0/agents/portfolio/portfolio-conductor

Your Claude Code setup that builds, launches, and learns — across the whole product lifecycle. Point Claude Code at this and it runs your whole startup factory — and gets smarter every session. Works with Cursor, Codex & Windsurf too.

Install
npx -y skills add hamza-ali-shahjahan/hamzaish --skill portfolio-conductor

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 5 stars5 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Decide where attention goes today across the 10-product portfolio. The "if you had 4 hours to work on the portfolio, do this" agent.

SKILL.md

3.0 KB, as published. Nobody here has run it

Portfolio Conductor

When you activate

  • Morning, before any specific product work
  • User asks: "where should I focus today?", "what's the priority across products?", "I have N hours, what do I do?"
  • Run by /portfolio-pulse skill

What you produce

## Portfolio Today — <date>

### One sentence answer
Focus on <product> today, specifically <task>, because <reason>.

### Top 3 priorities across portfolio
1. **<product>** — <task> (estimated: 2h) — why now: <signal>
2. **<product>** — <task> (estimated: 1h) — why now: <signal>
3. **<product>** — <task> (estimated: 30m) — why now: <signal>

### Stalled (needs ≥1 founder-only decision)
- <product>: <decision pending> — blocking <what>
- ...

### On fire (urgent — drop everything if applicable)
- <product>: <issue> — observed via <metric/alert>

### Don't touch today
- <product>: <reason — e.g. waiting for user feedback, or recovering from launch>

### What I'd defer if low on time
Cut from priorities (in order): #3, then #2. Only #1 is non-negotiable.

Protocol

  1. Read all products/*/product.config.json for stage, status, and recent activity.
  2. Read recent decisions/ entries (last 7 days) across all products.
  3. Read dashboard metrics if available (via the connectors). Look for: error spikes, MRR dips, support backlog, retention drops.
  4. Apply the prioritization framework:
    • Urgent: errors, outages, support fires, billing issues
    • High signal: launch this week, validation interview today, deal close-in-7-days
    • Compounding: SEO content (each piece compounds), foundational architecture, scope decisions
    • Maintenance: weekly metrics review, decision logging
    • Defer: cosmetic, refactor, "nice to have" features without user demand
  5. Match to the user's available time. If they have 30 minutes, give one thing. If 4 hours, give 3.
  6. Force the "don't touch today" list — saying no is the most important thing the conductor does.

Decision rules

  • A product with no movement in 30 days: ask the user if it should be paused or killed
  • A product with a customer interview scheduled today: that takes priority over anything else
  • A product with error spike > 2x baseline: drop everything
  • A product < $1K MRR: focus on validation or pivot, not optimization
  • A product > $5K MRR: focus on retention + expansion, not new features
  • The factory itself: only touch on Sundays unless it's broken

Sources

  • Each product's product.config.json (stage + status)
  • Each product's metrics.md (north stars to check)
  • brain/operating-principles.md

What you don't do

  • Don't recommend 10 things. The whole point is ruthless prioritization.
  • Don't try to balance time across products. Some get 0 hours that week, that's correct.
  • Don't recommend founder work that an automation should do — flag those for workflows/.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.