agentleFS
lglucaslglucas/ai-dev-operating-system/.claude/skills/grow-sustainably/SKILL.md

grow-sustainably

Skill12 starsChanged 49 days agoMITView on GitHub

Plan post-first-100-users growth without burning out the founder, the budget, or the user base. Pairs with `first-100-users` (which gets you to 100); this skill takes you from 100 to ~1000 sustainably. Activated when the user says "já tenho 100 usuários, e agora?", "como cresço sem queimar dinheiro?", "o WhatsApp 1-a-1 não escala mais".

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 8 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If any pre-condition is missing, the skill stops and routes to: first-100-users (under 100), usage-monitor (no cost telemetry), or …

    line 26
  2. Distinctive:

    If week-4 retention < 20%, do not scale acquisition.

    line 37
  3. Distinctive:

    If retention is healthy (≥30% at week-4), proceed to Pillar 2.

    line 39
  4. Distinctive:

    Look at the first 100. Where did they actually come from?

    line 43
  5. Distinctive:

    Ask: what made it work? Was it the channel itself, the message, the timing, or the founder's personal voice?

    line 47
  6. Distinctive:

    Only when Pillar 2 plateaus (≥4 weeks of flat or declining yield):

    line 53
  7. Distinctive:

    After 30 days: keep, kill, or extend by another 30 days.

    line 58
  8. Distinctive:

    Run this weekly during scaling:

    line 62
  9. Distinctive:

    If any signal trips, scaling stops until it's fixed.

    line 71
  10. Very specific:

    Always exhaust Tier 1 before Tier 2.

    line 90

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Grow Sustainably

first-100-users is conversation-driven. Beyond 100, conversations alone don't scale. But the answer is NOT to dump money into ads. The answer is to compress what worked, plug retention holes, and add ONE honest distribution channel at a time.

This skill protects against the three classic post-100 failures:

  1. Founder burnout — trying to keep the 1-a-1 cadence past the personal limit.
  2. Vanity acquisition — buying signups that never activate.
  3. Channel diffusion — running 5 channels at 20% effort instead of 1 channel at 100%.

Pre-conditions

This skill assumes:

  • ~100+ active users (active = completed the aha-moment + returned in 7d).
  • Some retention data exists (week-1, week-4 returning users).
  • At least one acquisition pattern that worked manually (a specific community, a specific cold outreach script, a specific content theme).
  • Cost telemetry exists (usage-monitor running) — otherwise scaling will quietly bankrupt the OPEX.

If any pre-condition is missing, the skill stops and routes to: first-100-users (under 100), usage-monitor (no cost telemetry), or daily-standup (no retention data captured).


The four pillars (run in order, never in parallel)

Pillar 1 — Retention before acquisition

A leaky bucket grows slower with more water. Before adding any new acquisition channel:

  • Plot week-1 and week-4 retention.
  • If week-4 retention < 20%, do not scale acquisition. Fix activation/onboarding first.
  • Identify the top 3 reasons inactive users gave (or you suspect) for not returning. Fix the top one first.
  • If retention is healthy (≥30% at week-4), proceed to Pillar 2.

Pillar 2 — Compress what already worked

Look at the first 100. Where did they actually come from?

  • Tag each of the 100 with one acquisition source.
  • Sort by source. Take the top source.
  • Ask: what made it work? Was it the channel itself, the message, the timing, or the founder's personal voice?
  • Codify the answer with /processize — now is the right time, you have ~12+ manual reps.
  • The output is a documented playbook for that single source. Run it deliberately, with light AI assist, until it plateaus.

Pillar 3 — Add ONE new channel deliberately

Only when Pillar 2 plateaus (≥4 weeks of flat or declining yield):

  • Pick exactly one new channel. Not two. Not "a content stack."
  • Define a 30-day experiment with a single primary metric (active users from this channel by day 30).
  • Cap the budget. If money is involved, set a hard ceiling and a circuit breaker.
  • After 30 days: keep, kill, or extend by another 30 days. Never default-renew.

Pillar 4 — Founder bandwidth check

Run this weekly during scaling:

SignalThreshold
Hours per week on the product< 50
Manual user-touchpoints per week< 3h
Days since last full day off< 14
Days since last conversation with a real user< 7

If any signal trips, scaling stops until it's fixed. A burned founder kills more startups than slow growth.


Channels, in rough efficiency order

Order is the median for a B2B / B2C-prosumer SaaS in early stage. Adjust to context.

TierChannelCostCompoundingBest when
1Word-of-mouth (NPS-driven invite)~0YesStrong activation, real value moments
1Niche community presencelowYesPersona is concentrated in <5 communities
2Long-form content (1 channel)lowYesFounder writes/records well; problem is narratable
2Cold outreach (handcrafted, < 20/day)lowNoPersona is on LinkedIn / X with searchable filters
3Programmatic SEOmediumYes (slow)Repeatable structured pages, query volume validated
3Newsletter sponsorship in niche newslettersmediumNoFirst channel hit a ceiling; AOV is high enough
4Paid ads (Google, Meta, X)highNoRetention proven, channels above hit a ceiling, finance allows runway burn
4Influencer placementhighMixedPersona is parasocial; founder won't appear themselves

Always exhaust Tier 1 before Tier 2. Always exhaust Tier 2 before Tier 3. Never skip to Tier 4 without a written justification reviewed via /multi-ai-review.


Hard rules

  • No new paid channel without usage-monitor running. Cost surprise = death.
  • No vanity metrics in the dashboard. Track active users, not signups. Track retention, not impressions.
  • No "agência de growth" before 1k MRR or 1k active users, whichever comes first. Founder owns growth in this phase.
  • Cap automated outreach at 50/day, even after /processize. Past that, deliverability and reputation collapse.
  • Any pricing or packaging change triggers business-plan-impact-review.

Output format

Generate docs/business/growth-plan-<YYYY-MM>.md:

# Growth Plan — <month>

**Active users (start of month):** <n>
**Week-4 retention:** <%>
**Top acquisition source last 30d:** <source> (<n> users)
**Founder bandwidth signals:** <green / yellow / red>

## Pillar 1 — Retention diagnosis

- Week-1 retention: <%>
- Week-4 retention: <%>
- Top 3 churn reasons: ...
- Action this month: ...

## Pillar 2 — Compress what worked

- Source: <source>
- Codified playbook: <link to docs/processes/...>
- Yield trend: ...

## Pillar 3 — New channel experiment (if any)

- Channel: <single channel>
- Hypothesis: ...
- Primary metric: ...
- Budget cap: ...
- 30-day decision date: <YYYY-MM-DD>

## Pillar 4 — Founder bandwidth

- Hours/week on product: ...
- Manual touchpoints/week: ...
- Days since full day off: ...
- Days since last user conversation: ...

## Hard rules check

- [ ] usage-monitor running
- [ ] Dashboard tracks active users + retention (not signups + impressions)
- [ ] No paid channel without justification + `/multi-ai-review`

## Decisions for next month

- ...

Cross-references

  • Principle: ETHOS.md item 14.
  • Predecessor skill: first-100-users.
  • Pairs with: processize, usage-monitor, cost-watchdog, business-plan-impact-review.
  • Inspired by: docs/registry/packs/slavingia-skills.md, docs/registry/packs/gstack.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.