Agent Skills by KwongFuk
Showing 54 vetted skills indexed across 1 GitHub repositories.
conversation-history-recorder
Record structured conversation history for workspace projects as Markdown files.
palmetto-cpu-request-advisor
Estimate and right-size Palmetto CPU requests for job submissions.
codex-custom-provider-pack
Configure Codex for custom provider endpoints and maintain model registries.
slurm-self-healing-watchdog
Monitor Slurm sbatch jobs via sacct and squeue, apply minimal repairs, and resubmit failures.
palmetto-apptainer-libcuda-fix
Patch sbatch launchers to bind host CUDA libraries into Apptainer containers.
path-storage-policy
Enforce path storage policies by routing project files between /home and /scratch.
paper-write
Generate LaTeX paper drafts from PAPER_PLAN.md and NARRATIVE_REPORT.md.
repro-record-workflow
Automate per-subtask logging into CODEX_RECORD.md for reproducible research workflows.
paper-poster
Extract paper content and generate a print-ready LaTeX conference poster.
palmetto-conda-scratch-guard
Enforce Palmetto storage policy with preflight checks and scratch-based environments.
proof-writer
Assess theorem feasibility and draft a structured PROOF_PACKAGE.md with dependency map.
formula-derivation
Construct step-by-step research formula derivations with labeled steps and remarks.
training-check
Detect WandB metric anomalies during active model training sessions.
result-to-claim
Judge experimental results against intended claims and route verdicts to next actions.
mermaid-diagram
Generate Mermaid diagrams from plain text requirements and validate syntax.
palmetto-slurm-workflow
Validate Palmetto Slurm job submissions with cluster guardrails and execution modes.
exact-training-resume-guard
Preserve full training state and resume from the latest complete checkpoint.
research-refine-pipeline
Coordinate research refinement and experiment planning into finalized proposal and roadmap files.
auto-review-loop-llm
Automate iterative research review loops with OpenAI-compatible LLMs.
experiment-plan
Plan claim-driven experiments with phases, run orders, and evaluation protocols.
idea-creator
Generate and rank publishable research ideas from broad directions.
scratch-trash-guard
Require a mode choice and confirmation code before deleting files or directories.
slurm-auto-retry-watchdog
Submit Slurm sbatch scripts and retry retriable failures up to a fixed limit.
grant-proposal
Draft grant proposals in KAKENHI, NSF, NSFC, ERC, DFG, SNSF, ARC, NWO, and GENERIC formats.