humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training.
Drewgent configuration for opencode — agent profiles, skills, scripts, and tooling
This repository also configures its own agents. See what opencode-drewgent tells them →
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Parameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods. Use when fine-tuning large models (7B-70B) with limited GPU memory, when you need to train <1% of parameters with minimal accuracy loss, or for multi-adapter serving. HuggingFace's official library integrated with transformers ecosystem.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Expert guidance for Fully Sharded Data Parallel training with PyTorch FSDP - parameter sharding, mixed precision, CPU offloading, FSDP2.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Expert guidance for fast fine-tuning with Unsloth - 2-5x faster training, 50-80% less memory, LoRA/QLoRA optimization.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Automate read-only inspection and supervised destructive operations on the user's Synology DS920+ NAS via expect + SSH. Covers connection config, sudo+password handling, read-only diagnostics, and the read-only-first policy for external system debugging.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
NeuronFS-style brain governance for {{AGENTNAME}}. Manages filesystem-based AI constraints via folder structure and special tokens (vorq, 禁).
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Default NeuronFS governance rules — demonstrates vorq harness and 禁 micro-opcode patterns. Load to establish baseline AI behavior constraints.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Maps NeuronFS 7-layer brain architecture to {{AGENTNAME}} prompt layers. Explains how P0-brainstem rules override all others.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Skill "obsidian-cli" from humanerd-drew/opencode-drewgent, covering obsidian cli, command reference, syntax, file targeting and vault targeting.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
name: obsidian-markdown title: Obsidian Flavored Markdown Skill description: Create and edit valid Obsidian Flavored Markdown with wikilinks, embeds, callouts, properties, vault graph health type: document space: concept tags: [concept, obsidian, vault, graph] created: 2026-05-24 updated: 2026-06-11 links.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
A guide for connecting a Korean payment gateway using the PortOne V2 SDK, including card and easy-payment methods and mobile redirects.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Airtable REST API via curl. Records CRUD, filters, upserts.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Huly Cloud (huly.app) — All-in-One Project Management. Covers PM, Chat, Docs, HR, CRM, Storage. Agent's primary work hub replacing Linear + kanban.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Manage Linear issues, projects, and teams via the GraphQL API. Create, update, search, and organize issues. Uses API key auth (no OAuth needed). All operations via curl — no dependencies.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
An analysis of Manyfast, a product-planning service, and a way to represent its requirements, features, and detailed specifications in an agent system.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Geocode, POIs, routes, timezones via OpenStreetMap/OSRM.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Operate the Teams meeting summary pipeline via Hermes CLI — summarize meetings, inspect pipeline status, replay jobs, manage Microsoft Graph subscriptions.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex needs its repo
A guide for changing Quartz, a static site generator, so pages marked as drafts are left out of the public build.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Jailbreak API-served LLMs using G0DM0D3 techniques — Parseltongue input obfuscation (33 techniques), GODMODE CLASSIC system prompt templates, ULTRAPLINIAN multi-model racing, encoding escalation, and {{AGENTNAME}}-native prefill/system prompt integration. Use when a user wants to bypass safety filters, jailbreak a…
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Skill "reference-gap-assessment" from humanerd-drew/opencode-drewgent, covering reference-gap-assessment, trigger conditions, core principle, user preference embedding and workflow.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
Never write a secret to disk in plaintext. Ever.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
An automated pipeline that collects articles from SEO and marketing RSS feeds, summarizes and tags them, scores their relevance, and produces trend reports.
humanerd-drew/opencode-drewgent
Skill Claude CodeCodex
An SEO and GEO audit guide with commands for checking a website. GEO means optimising content so AI search tools can discover, understand, and cite it.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: