fgcz-context

Provides canonical institutional context for FGCZ infrastructure, tools, and conventions.

Updated May 28, 2026
One-click install
npx skills add https://github.com/cpanse/skills --skill fgcz-context-cpanse
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: fgcz-context
Source: https://github.com/cpanse/skills/tree/main/meta-skills/skills/fgcz-context
Command: npx skills add https://github.com/cpanse/skills --skill fgcz-context-cpanse

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? FGCZ-specific knowledge — paths, hostnames, tool relationships, and conventions — is tribal knowledge that agents and new staff must rediscover every session. This skill grounds all other FGCZ skills in a shared institutional substrate so downstream workflows do not re-derive context. ## Core Features & Use Cases - Institutional substrate: Documents what FGCZ is, its five organizational groups, user personas, and the canonical data flow from B-Fabric metadata through sequencing, demultiplexing, SUSHI analysis, and gStore delivery. - Tech stack reference: Maps the full operational stack — B-Fabric LIMS, SUSHI, ezRun, Draugr, gStore/gtools, SLURM, GitLab, ShinyProxy — including how the components compose and who owns them. - Convention defaults: Codifies agent rules such as gStore read-only semantics (write only via g-req copynow), read-only gi_* conda environments, GPU --constraint=L40S preference, and ezRun report styling (300 DPI, polychrome, qs2). - Use Case: At the start of a session involving a SUSHI app or a pXXXXX project path, load this skill so the agent correctly interprets /srv/gstore/projects/pNNNNN/, knows not to rm in gStore, and routes deliverables through gtools. ## Quick Start Load the fgcz-context skill at the start of any FGCZ-related session before invoking other marketplace skills.

Frequently Asked Questions about fgcz-context

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
What is the FGCZ tech stack for bioinformatics workflows?

FGCZ uses B-Fabric as LIMS, SUSHI as the workflow framework with ezRun as the R analysis backend, Draugr for demultiplexing, gStore for long-term storage, SLURM on Debian 12 for compute, self-hosted GitLab, and ShinyProxy for app delivery.

How do I write files to gStore at FGCZ?

gStore is POSIX read-only by design to protect long-term deliverables. The only sanctioned write path is the gtools CLI, using `g-req copynow <src> <dst>` from a working filesystem like /srv/GT/analysis/. Never use rm, mv, or direct cp into /srv/gstore/.

What is the difference between B-Fabric project and order IDs?

A project ID (pNNNNN) and an order ID (oNNNNN) are distinct B-Fabric concepts used for per-project accounting. Project IDs map to gStore paths like /srv/gstore/projects/pNNNNN/.

Can I install packages into FGCZ gi_* conda environments?

No. Production conda environments with the gi_ prefix are read-only. Use the _dev variants for installable environments, or use Pixi for project-local rapidsai/conda-forge dependencies.

When should I not use the fgcz-context skill?

Do not use it as a primary how-to source. It describes the institutional world, while operational skills like fgcz-infrastructure and bfabric-lims describe how to act in it. Load it once per session as grounding context.

How do I submit GPU jobs on the FGCZ cluster?

Prefer `--constraint=L40S` (or Blackwell) in your SLURM submission rather than `--nodelist=fgcz-r-NNN`. Targeting a specific node risks queuing forever when that node is busy.