commute
Skill1 starsChanged 4 months agoMITView on GitHub
A 10-15 minute phone-sized cramcode session — sync, rapid-fire checks on weak/stale KCs, one thumb-typeable micro-test, sync back. Use when the user says they're on their phone, on the bus/train, or wants a quick session.
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 11
Open (quiet): If memory is git-tracked (synced mode), run python3 scripts/sync.py load first.
- Very specific: line 34
Write evidence: type: "test", "mode": "mobile", verdicts ONLY for KCs genuinely probed, notes quoting their answers.
- Very specific: line 37
If synced mode: python3 scripts/sync.py save.
- Distinctive: line 41
Keep every message short — no walls of text on a phone.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
/commute — the bus session
Everything fits a phone: short messages, no attempt files, answers the user can thumb-type in one or two lines. Target 10–15 minutes total.
Open (quiet). If memory is git-tracked (synced mode), run
python3 scripts/sync.py load first. Then python3 scripts/recommend.py --json
and greet with ONE line: what today's focus is and why.
Round 1 — rapid checks (2–3 KCs, ~5 min). Pick the weakest/stalest
targets from the recommendation. For each, ask ONE concrete prediction
question built from the KC's signals or pitfalls — a micro-scenario, not
trivia:
"Sliding window on [2,1,1,3], constraint 'sum ≤ 4'. Right pointer just added the 3 and the window is [1,1,3]. What happens next, exactly?"
Grade each answer as you go (one sentence of feedback, the fix if wrong).
These become demonstrated / partial / failed verdicts with the user's
answer quoted in the note.
Round 2 — one micro-test (~5 min). Take the recommended test problem and run it phone-style (PROTOCOL.md § "Phone-mode testing"): approach in words, then EITHER 2–3 fill-in-the-blank lines where you scaffold the solution and blank out exactly the KC-bearing lines, OR "you drive, I type" dictation. Then one complexity question. Never ask for full code on a phone.
Close (~2 min).
- Write evidence:
type: "test","mode": "mobile", verdicts ONLY for KCs genuinely probed, notes quoting their answers. Runpython3 scripts/log_episode.pyand show the deltas in one short block. - If synced mode:
python3 scripts/sync.py save. - Sign off with one line queued for the next laptop session: "tonight at a keyboard: full /test on Minimum Window Substring."
Keep every message short — no walls of text on a phone. One question at a time, always.
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
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.