Agent Skills by SaucerHi
Showing 53 vetted skills indexed across 1 GitHub repositories.
general-research
Answers research questions using project context from config, methodology, logs, and papers.
idea-refine
Convert rough research ideas into structured proposals with problem statements and validation plans.
theory-counterexample
Identify minimal counterexamples that violate assumptions in mathematical claims.
evolve-suggest
Aggregate G2 feedback into per-skill value updates and propose SKILL.md edits.
proof-formalize
Convert validated proofs into LaTeX writeups with theorem environments.
progress-capture
Extract progress updates from conversations and append timestamped YAML entries to daily logs.
checklist-update
Mark completed checklist items done and append new tasks with timestamps.
memory-retrieve
Score MEMORY.md entries by tag overlap, keyword match, recency, and importance.
paper-illustrate
Generate publication-ready diagrams as TikZ or SVG sources.
evolve-apply
Archive SKILL.md versions and apply approved edits with audit logging.
code-debug
Reproduce, isolate, identify root causes, and verify software fixes.
checklist-verify
Verify completed checklist items by validating artifacts, scripts, data, and cross-references.
plan-milestone
Compute milestone progress, days to deadline, and risk from config.yaml and logs.
design-converge
Weigh design options into a criteria matrix with scores and a recommendation.
proof-critique
Classify proof steps and identify logical gaps in plain text or LaTeX proofs.
memory-forget
Remove or archive low-value memories and update MEMORY.md.
session-open
Load config, memory, and digest files to display a session status banner.
session-close
Summarize and archive completed conversations into structured digest logs.
project-integrate
Integrate an unpacked SER distribution into an existing research project.
memory-consolidate
Cluster unconsolidated episodes into topics and procedures by tag overlap.
experiment-plan
Translate refined ideas into structured experiments mapped to claims and saved as plan.md.
experiment-monitor
Monitor active experiments by reading logs and verifying process health.
checklist-status
Recompute progress counts across checklist branches and leaf items.
decision-analyze
Compare decision options by pros, cons, effort, risk, and reversibility.