| name | level-up |
| description | Finds the single highest-leverage next move for the user's Cowork AI OS setup. Trigger when the user says "/level-up", "what should I improve next", "what's the next step", "what is missing from my setup", or "where should I focus this week". |
Level-Up — Find the Next Leverage Move
You ask the user 5 quick questions and then recommend the single highest-leverage thing to build next. Different from /audit:
/audit measures the current state
/level-up recommends the next action
You can run /level-up after /audit (most useful) or standalone.
Read these first
about-me/about-me.md, business-brain.md, memory.md, connections.md
- The most recent entry in
about-me/audit-log.md if it exists
The 5 questions
Ask these one at a time. Don't ask the next until they answer.
Q1 — Drudgery
"Walk me through your last week. What did you do 3+ times that felt manual, boring, or copy-paste?"
Q2 — Smart-intern test
"Anything where you thought 'a smart intern could absolutely handle this' — but you ended up doing it yourself because explaining would take longer?"
Q3 — Constraint
"If 100 new clients showed up Monday, what would break first?"
Q4 — Growth lever
"What would give you 100 more clients tomorrow if it ran on autopilot?"
Q5 — Friction
"What's the one thing in your day that makes you sigh?"
If they say "I don't know" to any question, give them a default based on what's in their about-me/ files and ask if it fits.
How to recommend
Look at all 5 answers. Pick the ONE that has the most leverage. Leverage = (impact if solved) × (frequency it happens). Recommend a SINGLE next move, in one of these forms:
Form A — Build a new skill
"Build a skill called ~~name. It would: [specific input → specific output]. The skill would automate [the pain]. I'd hand off to /add-skill to actually build it. Want to do that now?"
Form B — Add a connector (delegates to /browse-connectors)
"The highest-leverage gap is an empty Connection bucket — specifically [Bucket]. Right now you're doing [X] manually because I can't see [tool]. I'm going to hand off to /browse-connectors, which will recommend the specific MCP that fits your existing stack, walk you through enabling it with safe permissions, and confirm it works. Type go when you're ready, or /browse-connectors to do it yourself."
When the user says go, invoke /browse-connectors with the bucket gap as context.
Form C — Refine a context file
"Update business-brain.md to capture [specific gap, e.g. 'your new pricing structure for the Build-a-thon offer']. The drafts I generate will hit closer to your voice. Want to do it now?"
Form D — Schedule a routine
"Add a scheduled task: [name]. It would fire [frequency] and do [specific thing]. Want to set it up?"
Form E — Install a complementary skill / plugin (delegates to /browse-skills)
"The highest-leverage gap is a missing capability — specifically [the thing they're doing manually]. There's likely a community skill or plugin that handles this. I'm going to hand off to /browse-skills, which reads your business-brain and returns 3 ranked recommendations with reasoning. Type go when you're ready, or /browse-skills to do it yourself."
When the user says go, invoke /browse-skills with the gap as context.
Form F — Build a custom skill (delegates to /add-skill)
"Nothing on the marketplace covers this — it's specific to your workflow. I'll hand off to /add-skill, which interviews you and builds a custom slash command via the Skill Creator. Type go when ready."
When the user says go, invoke /add-skill.
Pick by leverage, not by ease
Don't recommend the easiest thing. Recommend the most leveraged thing. The user can say "too big right now" and you'll downshift — but lead with the high-leverage move.
Choosing between the forms
The 6 forms map to the 4 Cs:
- Form C (refine context file) → if the gap is in C1 Context (about-me thin or stale).
- Form B (add connector) → if the gap is in C2 Connections (a bucket is empty and would multiply existing skills).
- Forms E/F (install or build skill) → if the gap is in C3 Capabilities (a workflow needs automation).
- Form A (build new skill) → use Form A when the user's case is highly specific. Use Form E first when there's likely a marketplace skill that fits —
/browse-skills will confirm.
- Form D (schedule a routine) → if the gap is in C4 Cadence (manual rerunning of something that should fire automatically).
When in doubt between Form A (build) and Form E (install), default to Form E — installing a verified community skill is faster and lower-risk than building from scratch. Build (Form A/F) only when nothing on the marketplace fits.
After the recommendation
Wait for the user. If they say:
- yes / go / let's do it — execute the recommendation. If it's "build a new skill," hand off to
/add-skill. If it's "add a connector," walk them through. Etc.
- too big or smaller — recommend the next-highest-leverage move from their answers.
- save for later — append to memory.md as an open loop, exit cleanly.
- already doing this — apologize for missing the context, ask Q1 again with more specificity.
After the recommendation is acted on
Append to memory.md:
### YYYY-MM-DD — Level-up: [recommendation]
- **Trigger:** /level-up
- **Top answer (drove the rec):** "[user's quote]"
- **Recommendation:** [what was recommended]
- **Action taken:** [executed / deferred / refused]
Appendix: The 4 Cs framework (the audit rubric)
Score the user Cowork AI OS on a 0-100 scale. 25 points per C.
C1 - Context (0-25)
How much does Claude know about the user and their business?
- 0-8: No about-me/ files, or files have ~~placeholders unfilled.
- 9-16: About-me/ has the basics but writing-rules.md has fewer than 3 samples.
- 17-22: All 4 about-me/ files complete. Writing samples real. Memory.md has 5+ entries.
- 23-25: Above plus business-brain.md kept current and memory.md gets appended every session.
C2 - Connections (0-25)
How many of the 7 buckets (Revenue, Customer, Calendar, Comms, Tasks, Meetings, Knowledge) are reachable?
- 0-8: 0-2 buckets reachable.
- 9-16: 3-4 buckets reachable.
- 17-22: 5-6 buckets reachable.
- 23-25: All 7 reachable with safe permissions (writes need approval).
C3 - Capabilities (0-25)
How many things can Claude actually do?
- 0-8: Cowork AI OS only, never run a skill beyond /onboard.
- 9-16: Cowork AI OS skills used regularly. No additional plugins.
- 17-22: Cowork AI OS plus 1-2 Anthropic role plugins. 1-2 custom skills built via /add-skill.
- 23-25: Above plus custom skills get refined every few weeks.
C4 - Cadence (0-25)
How much works on a schedule, vs only when asked?
- 0-8: No scheduled tasks.
- 9-16: Morning brief scheduled. Nothing else.
- 17-22: Morning brief plus weekly tune-up plus at least one custom routine.
- 23-25: Multiple routines firing reliably. Computer-awake reliability solved.