remote-run
Skill1 starsChanged 3 months agoMITView on GitHub
Run any command (e.g. the CircuitGenome test suite) on the beefy TUM box against a synced mirror of the local working tree, kept alive in tmux, then report the exit code and tail. Explicit-invoke only.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- NAIVlab/run-skillstheir own version, first seen Nov 8, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- NAIVlab/run-skills★ 5same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository19 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ask-mattSkill
- code-reviewSkill
- diagnosing-bugsSkill
- domain-modelingSkill
- explain-codeSkill
- explain-code-with-figSkill
- grillingSkill
- grill-meSkill
- grill-with-docsSkill
- implementSkill
- meeting-prepareSkill
- researchSkill
- setup-matt-pocock-skillsSkill
- smalltalkSkill
- tddSkill
- to-issuesSkill
- to-prdSkill
- youtube-summarySkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 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.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 19
All orchestration lives in scripts/remote-run.sh — prefer it over hand-writing ssh/rsync/tmux commands.
- Distinctive: line 46
Do not proceed until preflight is clean.
- Very specific: line 62
Default tmux session is cg-remote.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
remote-run
Offload a heavy command to the remote TUM machine because the local machine isn't powerful enough (SPICE/ngspice runs, the full pytest suite, long jobs). The command is arbitrary; the machine and project are fixed:
| local | /Users/phuocpham/workspace/projects/CircuitGenome |
| remote | pham@10.157.146.6:/home/pham/code/analog-ml/CircuitGenome |
| ssh | key auth, -o ServerAliveInterval=240 |
All orchestration lives in scripts/remote-run.sh —
prefer it over hand-writing ssh/rsync/tmux commands.
The flow
Each run is: sync → launch detached in tmux → poll until done → report.
The remote run survives SSH drops because it lives in a tmux session; the log is
written outside the checkout (~/.remote-run/<session>.log) so rsync --delete
never wipes it mid-run.
Let RR=".claude/skills/remote-run/scripts/remote-run.sh" (path relative to the
skills repo; adjust to wherever you invoke from).
1. Preflight (first run of a session, or on any failure)
bash "$RR" preflightConfirms the box is reachable non-interactively, that tmux/uv/rsync
exist, and the checkout is present. If it prints Permission denied, key auth
isn't set up — tell the user to run once:
ssh-copy-id -o ServerAliveInterval=240 pham@10.157.146.6Do not proceed until preflight is clean.
2. Launch
Pick the command based on the task. Everything after -- is run verbatim on the
remote, inside the checkout:
bash "$RR" run -- uv run pytest # whole suite (default heavy job)
bash "$RR" run -- uv run pytest tests/test_spice_sim.py -x # chase one failure
bash "$RR" run -- uv run python examples/foo.py # any non-test jobrunrsyncs the local tree first (--delete, excluding.git/ .venv/ __pycache__/ .pytest_cache/), then starts the command detached and returns immediately.- Default tmux session is
cg-remote. If it's already busy,runrefuses to clobber it and prints that run's status — wait,tail, orkillit. To run a second thing concurrently, pass--session cg-remote-<label>.
3. Poll until done
bash "$RR" poll # -> "running" | "done EXIT=N" | "idle"Loop with a short sleep between polls (e.g. 15–30s for a full suite). Keep going
until it prints done EXIT=N. A dropped connection costs nothing — just poll
again.
4. Report
bash "$RR" tail -n 40 # last lines of output (pytest's own summary lands here)Report to the user: exit code (0 = success) plus the tail. On a non-zero
exit, tail more (-n 100) or pull the whole log back for the failing
tracebacks:
bash "$RR" fetch-log ./cg-remote.logThe full log always stays on the remote; only fetch it when needed.
Housekeeping
bash "$RR" kill # kill the default session
bash "$RR" kill --session cg-remote-trainNotes
- All
sshusesBatchMode=yes— it never hangs on a password prompt; it fails fast instead, which is what preflight detects. - The command is shipped base64-encoded, so quotes/globs/pipes in it survive intact.
uv runauto-syncs the remote env fromuv.lockbefore running, so no separate install step is needed.
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.

