opus-et-statuslisted

Use when the user asks what is running or failed on the cluster, whether a SLURM job finished, which phase or human gate the pipeline is at, to read a failed job's log, to browse QC images, to compare tilt series by CTF fit, dose or alignment, to see template-matching score distributions, or to change a tuning knob in species.conf without editing it by hand. Polls the cluster over SSH from the local machine.
alncat/opus-et-agent · ★ 0 · Data & Documents · score 72

Install: claude install-skill alncat/opus-et-agent

# OPUS-ET Status Server A dependency-free (stdlib-only) web dashboard that polls the cluster over SSH and shows SLURM job state, pipeline phase progress, gate status, failure log tails, and a narrow typed config editor. ## Launch ```bash python3 opus-et-status/scripts/status_server.py \ --host <CLUSTER_HOST> \ --work-dir <RUN_DIR> \ --port 8080 ``` Then open <http://127.0.0.1:8080>. - `--host` is an SSH host or alias that already works non-interactively (`ssh <CLUSTER_HOST> true` must succeed without a password prompt; the client uses `BatchMode=yes`). - `--work-dir` is the run directory on the cluster. - WARP folders come from the run's `pipeline.conf` (`FRAMESERIES_DIR`, `PROCESSING_DIR`, `TILTSTACK_DIR`, and `M_DIR` for the Refinement panel; read, never executed), else the `warp_*` / `m` defaults. Point `--pipeline-conf` at the conf the jobs actually source (default `opus-et-warp/pipeline.conf`), or name a tree directly with `--tiltseries-dir` / `--frameseries-dir` / `--tiltstack-dir` / `--m-dir` -- e.g. after a rebuild into `warp_tiltseries_v2`, a hard-coded `warp_tiltseries/` shows a stale or empty tree. The Frames notes say which folders were read. - `--species-conf` defaults to `species.conf`; pass it only to start on a different one. Naming a conf that is not there is not fatal -- the Config tab says which file it could not read and every other panel still fills. - One run per instance. To watch a second run, start another instance