Growth engine

Growth Engine — find who'd value planfi & how to monetize it

A growth-research skill. It fans out a Workflow of agents that web-research real, named prospects who could find value in planfi, score them, and attach a monetization model to each — then synthesizes the batch into concrete revenue plays. planfi is currently free with no revenue, so every run pairs reach (who could amplify or use planfi) with money (how that turns into sustainable revenue without breaking planfi's free, privacy-first promise).

It is repeatable and compounding: the persistent memory in growth/ (see growth/README.md) lets each run skip what's already been found. The workflow does the research; this skill does the ledger-reading and persistence around it.

It runs in two modes, auto-selected by whether you pass a focus:

• BROAD (no focus) — a sweep across all five segments (companies, creators, communities, media, distribution). The default discovery run.
• FOCUSED (focus set) — drill into one segment/angle (e.g. "FIRE YouTubers", "RIAs", "MCP directories"). A scoping agent derives sub-segments; the SAME pipeline researches just that slice.

This skill is the user's explicit opt-in to multi-agent orchestration — run the workflow; don't try to research the whole list solo in the main loop.

When to use vs. not

• Use to discover and grow a pipeline of prospects + monetization angles for planfi, or to refresh it with net-new leads in a segment.
• Don't use to write a single LinkedIn post (that's the linkedin-finance-marketer agent), to brainstorm monetization in the abstract with no prospecting (that's the financial-monetization-strategist agent), or to analyze a user's own financial plan (comprehensive-plan / financial-forecast).

Progress updates (Discord)

Keep the user updated in Discord, the same way Feature Factory does. Read the webhook URL from your memory/config (the discord-webhook reference) — never hardcode it. The workflow runs in the background, so post at the checkpoints the main loop controls, via curl -H 'Content-Type: application/json' -d '{"content":"..."}' <webhook>:

• Kickoff — when you launch the workflow (mode + focus, e.g. "🚀 Growth Engine: broad sweep started").
• Done — after Step 2 persists: net-new count, the P0 prospect names, the top revenue play, and the saved report path. If no webhook is in memory, skip silently (don't block the run).

Step 0 — Read the ledger (de-dupe input)

Before running, read growth/prospects.jsonl and collect the name of every existing prospect into a seen array. (If the file is empty, seen is [].) This is what makes runs compound instead of repeat. A quick way:

Bash: test -s growth/prospects.jsonl && cut -d'"' -f4 growth/prospects.jsonl || true

(or Read the file and pull each name). Pass the resulting list as seen.

Optional inputs — all have defaults, so you can run cold:

• focus — set for a FOCUSED sweep on one segment/angle (e.g. "FIRE newsletter authors"). Omit for BROAD.
• perSegment — target net-new prospects per segment (default 8).

Step 1 — Run the workflow (PRIMARY ACTION)

Invoke the saved workflow, passing the seen list and whatever the user specified:

// Broad sweep:
Workflow({ name: 'growth-engine', args: {
  seen: [ ...names from growth/prospects.jsonl... ],
  root: '/Users/kameronkales/planfi-app'
}})

// Focused on one segment/angle:
Workflow({ name: 'growth-engine', args: {
  focus: 'FIRE YouTubers',
  seen: [ ...names... ],
  root: '/Users/kameronkales/planfi-app'
}})

What it does (one well-scoped fan-out): 0. Scope (focused only) — derive sub-segments for the requested focus.

1. Prospect — one agent per segment WebSearches for real, named prospects (excluding seen), each with a fit rationale, reach estimate, outreach angle, and monetization model.
2. Qualify — score each prospect's fit/reach/effort, assign a P0–P2 priority, sharpen the angle.
3. Monetize — a strategist synthesizes the batch into concrete revenue plays + new angles.
4. Compile — assemble the dated run report markdown.

It returns { focus, mode, segments, counts, prospects, monetization, new_angles, markdown }.

Step 2 — Persist (this is the point — retain the work in the repo)

1. Append the new prospects to the ledger. For each item in result.prospects, append one JSON line to growth/prospects.jsonl in the record shape documented in growth/README.md — map the workflow fields and stamp status: "new", first_seen: "<today's date>", source: "<run slug>". Append only; never rewrite existing lines.
2. Save the run report. Write result.markdown to growth/runs/<YYYY-MM-DD>-<slug>.md where slug is broad or a kebab of the focus (e.g. 2026-06-12-fire-youtubers.md).
3. Update the playbook. Append result.new_angles as dated bullets under "## Discovered angles" in growth/monetization-playbook.md (newest first). Skip angles already listed there.

Use today's date from context for all stamps. Create growth/runs/ if missing.

Step 3 — Present

Summarize for the user: mode, result.counts (sourced / net-new / qualified), the P0 prospects by name with their angle + monetization model, and the revenue plays from result.monetization. Point to the saved report path; optionally SendUserFile it. Offer follow-ups: a FOCUSED run on the most promising segment, drafting outreach for the P0 list (hand to linkedin-finance-marketer / fintech-pitch-architect), or deepening one revenue play with financial-monetization-strategist.

Guardrails (bake into any prompt you add)

• Real, verifiable prospects only — named, lookup-able, with a working URL. No invented companies or people; reach figures are labelled estimates, never fabricated facts.
• Don't re-propose seen prospects — the whole point of the ledger is net-new each run.
• Monetization must respect planfi's promise — free core, privacy-first, no selling user data, no ad-tracking wall. Revenue comes from partners/distribution/affiliate/premium layers.
• This is prospecting & strategy, not outreach — nothing here contacts anyone. The user decides what to act on; status in the ledger is theirs to advance.

Notes

• Iterating on the workflow? Edit .claude/workflows/growth-engine.js and re-invoke Workflow({ name: 'growth-engine', ... }); the segments are defined at the top of that file — add/retune them as the strategy evolves.
• The persistent memory lives in growth/ (prospects.jsonl, monetization-playbook.md, runs/); growth/README.md documents the record shape and status lifecycle.