Back to authors
poorrican

poorrican

357 Skills published on GitHub.

simpo-training

Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance (+6.4 points on AlpacaEval 2.0). No reference model needed, more efficient than DPO. Use for preference alignment when want simpler, faster training than DPO/PPO.

[Post-TrainingSimPOPreferenceOptimizationAlignmentDPOAlternativeReference-FreeLLMAlignmentEfficientTraining]
UncategorizedView skill →

slime-rl-training

Provides guidance for LLM post-training with RL using slime, a Megatron+SGLang framework. Use when training GLM models, implementing custom data generation workflows, or needing tight Megatron-LM integration for RL scaling.

[ReinforcementLearningMegatron-LMSGLangGRPOPost-TrainingGLM]
UncategorizedView skill →

unsloth

Unsloth: 2-5x faster LoRA/QLoRA fine-tuning, less VRAM.

[Fine-TuningUnslothFastTrainingLoRAQLoRAMemory-EfficientOptimizationLlamaMistralGemmaQwen]
UncategorizedView skill →

chroma

Open-source embedding database for AI applications. Store embeddings and metadata, perform vector and full-text search, filter by metadata. Simple 4-function API. Scales from notebooks to production clusters. Use for semantic search, RAG applications, or document retrieval. Best for local development and open-source projects.

[RAGChromaVectorDatabaseEmbeddingsSemanticSearchOpenSourceSelf-HostedDocumentRetrievalMetadataFiltering]
UncategorizedView skill →

faiss

Facebook's library for efficient similarity search and clustering of dense vectors. Supports billions of vectors, GPU acceleration, and various index types (Flat, IVF, HNSW). Use for fast k-NN search, large-scale vector retrieval, or when you need pure similarity search without metadata. Best for high-performance applications.

[RAGFAISSSimilaritySearchVectorSearchFacebookAIGPUAccelerationBillion-ScaleK-NNHNSWHighPerformanceLargeScale]
UncategorizedView skill →

pinecone

Managed vector database for production AI applications. Fully managed, auto-scaling, with hybrid search (dense + sparse), metadata filtering, and namespaces. Low latency (<100ms p95). Use for production RAG, recommendation systems, or semantic search at scale. Best for serverless, managed infrastructure.

[RAGPineconeVectorDatabaseManagedServiceServerlessHybridSearchProductionAuto-ScalingLowLatencyRecommendations]
UncategorizedView skill →

qdrant-vector-search

High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance.

[RAGVectorSearchQdrantSemanticSearchEmbeddingsSimilaritySearchHNSWProductionDistributed]
UncategorizedView skill →

obsidian

Read, search, create, and edit notes in the Obsidian vault.

UncategorizedView skill →

airtable

Airtable REST API via curl. Records CRUD, filters, upserts.

[AirtableProductivityDatabaseAPI]
UncategorizedView skill →

computer-use

|

[computer-usedesktopautomationguicross-platform]
desktopView skill →

data-preserving-doc-rewrite

Use when rewriting a large machine-assembled document (compendium, report, data dump) into better prose/tables WITHOUT changing any numbers, IDs, dates, or empirical claims — especially when parallelizing across sections with subagents and needing to prove zero data loss.

UncategorizedView skill →

documentation-curation

Use when documents disagree or must change without loss: adjudicating versions against implementation or deployment evidence, reconciling canonical contracts across co-referencing documents, preserving information in rewrites and expansions, consolidating records into append-only logs, or producing source-verified documentation across many files.

UncategorizedView skill →

google-workspace

Gmail, Calendar, Drive, Docs, Sheets via gws CLI or Python.

[GoogleGmailCalendarDriveSheetsDocsContactsEmailOAuth]
UncategorizedView skill →

linear

Linear: manage issues, projects, teams via GraphQL + curl.

[LinearProjectManagementIssuesGraphQLAPIProductivity]
UncategorizedView skill →

maps

Geocode, POIs, routes, timezones via OpenStreetMap/OSRM.

[mapsgeocodingplacesroutingdistancedirectionsnearbylocationopenstreetmapnominatimoverpassosrm]
productivityView skill →

nano-pdf

Edit PDF text/typos/titles via nano-pdf CLI (NL prompts).

[PDFDocumentsEditingNLPProductivity]
UncategorizedView skill →

notion

Notion API + ntn CLI: pages, databases, markdown, Workers.

[NotionProductivityNotesDatabaseAPICLIWorkers]
UncategorizedView skill →

ocr-and-documents

Extract text from PDFs/scans (pymupdf, marker-pdf).

[PDFDocumentsResearchArxivText-ExtractionOCR]
UncategorizedView skill →

petdex

Install and select animated petdex mascots for Hermes.

[petdexmascotdisplayclituidesktop]
productivityView skill →

powerpoint

Create, read, edit .pptx decks, slides, notes, templates.

UncategorizedView skill →

teams-meeting-pipeline

Operate the Teams meeting summary pipeline via Hermes CLI — summarize meetings, inspect pipeline status, replay jobs, manage Microsoft Graph subscriptions.

[TeamsMicrosoftGraphMeetingsProductivityOperations]
UncategorizedView skill →

using-proton-pass-cli

Interact with Proton Pass via the `pass-cli` command-line tool. Use this skill when the user wants to manage passwords, vaults, items, SSH keys, secret injection, or any credential management through the Proton Pass CLI. Triggers on mentions of pass-cli, proton pass, secret references, pass:// URIs, vault management, or SSH agent integration with Proton Pass.

UncategorizedView skill →

godmode

Jailbreak LLMs: Parseltongue, GODMODE, ULTRAPLINIAN.

[jailbreakred-teamingG0DM0D3ParseltongueGODMODEuncensoringsafety-bypassprompt-engineeringL1B3RT4S]
UncategorizedView skill →

analysis-judgement

Use when an empirical conclusion is forming, a result confirms expectations, a design needs pressure-testing, or an experiment could resolve uncertainty. Covers falsifiable challenge questions, denominators and aggregation grain, data and release boundaries, result audits, residual attribution, and choosing useful spikes. A source of judgement prompts and analytical traps, not a fixed procedure.

UncategorizedView skill →

arxiv

Search arXiv papers by keyword, author, category, or ID.

[ResearchArxivPapersAcademicScienceAPI]
UncategorizedView skill →

blogwatcher

Monitor blogs and RSS/Atom feeds via blogwatcher-cli tool.

[RSSBlogsFeed-ReaderMonitoring]
UncategorizedView skill →

domain-intel

Passive domain reconnaissance using Python stdlib. Subdomain discovery, SSL certificate inspection, WHOIS lookups, DNS records, domain availability checks, and bulk multi-domain analysis. No API keys required.

UncategorizedView skill →

duckduckgo-search

Free web search via DuckDuckGo — text, news, images, videos. No API key needed. Use the Python DDGS library or CLI to search, then web_extract for full content.

[searchduckduckgoweb-searchfreefallback]
UncategorizedView skill →

experiment-log-interview

Conducts a structured Socratic interview to produce or update a single experiment log entry — the durable record of what was run, what it showed, and what it means. Use this skill whenever the user wants to log an experiment, write up results, record a backtest, capture a finding, pre-register a run, document a study, or update an existing entry with new results or a revised interpretation. Trigger on phrases like "log this experiment," "write up the results of...", "I ran X, help me document it," "pre-register this," "update the entry for...", or when the user shares results and asks for help interpreting and recording them. The skill enforces the four-way separation between what happened, what it means, what it implies, and what comes next; challenges the user's interpretations with evidence requests and alternative explanations; and writes incrementally to keep context clean and the entry always grounded.

[researchexperiment-loginterviewdocumentation]
UncategorizedView skill →

experiment-log-structure

Use when an agent needs to produce, update, validate, or normalize a standardized experiment-log entry without running an interview. Defines the canonical structure, pre-registration rules, evidence/interpretation split, calibration tags, and append-only revision model for durable experiment records.

[researchexperiment-logstructurestandardizationagents]
UncategorizedView skill →

llm-wiki

Karpathy's LLM Wiki: build/query interlinked markdown KB.

[wikiknowledge-baseresearchnotesmarkdownrag-alternative]
researchView skill →

ml-paper-writing

Write publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. Use when drafting papers from research repos, structuring arguments, verifying citations, or preparing camera-ready submissions. Includes LaTeX templates, reviewer guidelines, and citation verification workflows.

[AcademicWritingNeurIPSICMLICLRACLAAAICOLMLaTeXPaperWritingCitationsResearch]
UncategorizedView skill →

polymarket

Query Polymarket: markets, prices, orderbooks, history.

UncategorizedView skill →

research-paper-writing

Write ML papers for NeurIPS/ICML/ICLR: design→submit.

[ResearchPaperWritingExperimentsMLAINeurIPSICMLICLRACLAAAICOLMLaTeXCitationsStatisticalAnalysis]
researchView skill →

research-proposal-interview

Conducts a structured Socratic interview to produce a comprehensive markdown research proposal that handles cascading uncertainty (fixed end-question, branching experiments). Use this skill whenever the user wants to write a research proposal, research plan, study design, experiment plan, thesis proposal, RFC, or "spec out" a research direction — even if they don't explicitly say "interview me." Trigger when the user says things like "help me plan this research", "I want to design experiments for X", "draft a proposal for...", "think through a research direction", or shares a half-formed research idea and asks for help structuring it. The skill interviews the user, challenges their priors with evidence requests and falsifiers, optionally uses sub-agents to explore prior art, and builds the proposal markdown incrementally so context stays clean and the document is always grounded.

[researchproposalinterviewplanning]
UncategorizedView skill →

research-proposal-structure

Use when an agent needs to produce, update, validate, or normalize a standardized research proposal artifact without running an interview. Defines the canonical structure, confidence-tag semantics, decision logic, and completion checks for proposal.md-style research plans.

[researchproposalstructurestandardizationagents]
UncategorizedView skill →

openhue

Control Philips Hue lights, scenes, rooms via OpenHue CLI.

[Smart-HomeHueLightsIoTAutomation]
UncategorizedView skill →

xurl

X/Twitter via xurl CLI: post, search, DM, media, v2 API.

[twitterxsocial-mediaxurlofficial-api]
UncategorizedView skill →

yuanbao

Yuanbao (元宝) groups: @mention users, query info/members.

[yuanbaomentionatgroupmembers元宝派艾特]
UncategorizedView skill →

agent-orchestration

Use when planning or coordinating multi-agent work: contract-first fan-outs, parallel correctness audits and diff reviews, implementation plans and handoffs, dependent issue chains, transcript-based recovery after crashes or file loss, and mining past sessions for unresolved threads.

UncategorizedView skill →

hermes-agent-skill-authoring

Author in-repo SKILL.md: frontmatter, validator, structure, and writing-quality principles.

[skillsauthoringhermes-agentconventionsskill-md]
UncategorizedView skill →

hermes-tui-development

Inspect and extend Hermes Agent TUI capabilities, launch flags, hotkeys, composer input, and terminal behavior.

[hermes-agenttuitypescriptinkclikeybindings]
UncategorizedView skill →

subagent-driven-development

Execute plans via delegate_task subagents (2-stage review).

[delegationsubagentimplementationworkflowparallel]
UncategorizedView skill →

debugging-hermes-tui-commands

Debug Hermes TUI slash commands: Python, gateway, Ink UI.

[debugginghermes-agenttuislash-commandstypescriptpython]
UncategorizedView skill →

debugging

Use for root-causing bugs, test failures, and unexpected behavior; tracing values and failures through call chains and component boundaries; choosing Python debuggers such as pdb and debugpy; and diagnosing Linux authentication, device, memory-pressure, Btrfs, and block-device problems.

UncategorizedView skill →

linux-system-debugging

Class-level Linux debugging workflow for authentication/PAM lockouts, desktop hardware/device stacks, and RAM/swap/CPU/process resource pressure.

[linuxdebuggingpamfaillockdesktophardwarebluetoothmemoryswapperformance]
UncategorizedView skill →

node-inspect-debugger

Debug Node.js via --inspect + Chrome DevTools Protocol CLI.

[debuggingnodejsnode-inspectcdpbreakpointsui-tui]
UncategorizedView skill →

performance-and-throughput

Use when a pipeline, model, training loop, or data path is slow or memory-hungry; for cost attribution, performance-claim review, RSS growth, CPU/GPU benchmarking, loader starvation, CPU-bound training, container-only CUDA faults, streaming-cache thrash, and memory or numerical blowups.

UncategorizedView skill →

python-debugpy

Debug Python: pdb REPL + debugpy remote (DAP).

[debuggingpythonpdbdebugpybreakpointsdappost-mortem]
UncategorizedView skill →

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

UncategorizedView skill →

Page 3 of 8 · 357 results