Agent Stuff

Agent skills and plugins I use with Claude, Codex, and other agents. Some skills are specific to my own setup (like my Obsidian vault).

Agent skills format

A skill is a folder of instructions and any files an agent needs to follow them. Skills here use the Agent Skills format:

skill-name/
├── SKILL.md              # Metadata and instructions
├── references/           # Optional: detailed docs or examples
└── scripts/              # Optional: helper scripts

The metadata at the top of SKILL.md gives the skill its name and description. The description tells the agent when to load it; the instructions explain how to do the work.

Install plugins

Plugins here use the Agent Plugins format to install a set of related skills at once. You can still install any of those skills on its own or download its ZIP. If you install a plugin, don't also install its skills separately, or you'll end up with duplicates.

Each skill keeps the layout shown above inside the plugin's skills/ folder:

plugin-name/
├── plugin.json              # Portable plugin metadata
├── .claude-plugin/
│   └── plugin.json          # Claude metadata
├── .codex-plugin/
│   └── plugin.json          # Codex compatibility metadata
└── skills/
    ├── first-skill/
    │   └── SKILL.md
    └── second-skill/
        └── SKILL.md

Browse the available bundles under Catalog → Plugins.

Claude

On a paid plan, open Customize → Plugins → Add → Add marketplace and enter nweii/agent-stuff. The plugins will then appear under Discover; add the ones you want. They're saved to your account and also load in Claude Code when you sign in. See Claude's plugin guide.

ChatGPT and Codex

As of Oct 2026, personal ChatGPT accounts need manual ZIP uploads to install this repository's plugins at the account level. GitHub marketplace installation is only available for local Codex use and does not sync to web and mobile at this time.

Personal ChatGPT account

1. Download the plugin ZIP from zips/plugins/.
2. In ChatGPT, open Plugins → Personal → Upload plugin archive and select the ZIP.

The uploaded plugin is available to your account, including on web and mobile. This install won't update automatically; new versions require uploading a fresh ZIP under "Upload new version."

Local desktop and Codex access

In the ChatGPT desktop app, open Plugins → Add → Add a marketplace and enter nweii/agent-stuff. Install the plugins you want for local use.

In the Codex CLI, add the marketplace, then install a plugin:

codex plugin marketplace add nweii/agent-stuff
codex plugin add obsidian-tools@agent-stuff

ChatGPT workspace admins: import from GitHub

Workspace admins have a separate GitHub import and sync route. This is distinct from the personal-account ZIP uploads and local marketplace installs above. To keep a workspace's copies in sync with this repo:

1. Open Admin → Plugins → Add → Import marketplace.
2. Set Source to https://github.com/nweii/agent-stuff. Leave Path and Branch, tag, or commit empty.
3. Import, authorize GitHub access, and make the plugins you want available to workspace members.
4. Members install them from Plugins.

ChatGPT checks for updates daily. See OpenAI's marketplace import guide for permissions and sync controls.

Use individual skills without installing

With Bun and terminal access available, tell your agent to run:

bunx skills use nweii/agent-stuff

This lists the skills here (excluding internal skills). Your agent can choose one and read it with --skill SKILL_NAME, without installing it.

Install individual skills

Copy whatever looks useful.

With Bun or Node.js installed, use the skills CLI to choose skills and the coding agents that should load them:

bunx skills add nweii/agent-stuff

# or
npx skills add nweii/agent-stuff

Install a specific skill by name:

bunx skills add nweii/agent-stuff --skill obsidian-clis

By default, skills install in the project where you run the command. Add -g to make a skill available across projects.

Update installed skills:

bunx skills update

# Global installs
bunx skills update -g

To refresh one skill, run skills add again with its name. Use skills CLI 1.5.24 or later. Older versions can lose track of a skill when I reorganize this repo. Internal skills have one update quirk, explained under Internal skills.

ZIP downloads and manual copies

Every general-purpose skill has a ZIP in zips/. For Claude, follow its skill setup guide to enable code execution and upload the ZIP. Uploaded copies need manual updates.

You can also copy a skill folder into the skills directory your agent reads. Copy the whole folder so the agent can find any references or scripts its instructions use. Some skills use other apps or services. Installing the skill doesn't install those.

Internal skills

Skills under the Internal heading in the catalog have metadata.internal: true. They're personal workflows tied to my own setup (vault paths, tools, naming conventions) and are excluded from the ZIP downloads. "Internal" just means they're written for my setup; the source is still public.

Install it as-is (when the skill's assumptions already match your setup)

Some don't hardcode anything of mine. They assume a convention (a related property, a daily/weekly note hierarchy) or a tool (Granola MCP, gh) and otherwise work anywhere. If nothing in the SKILL.md names a folder or template you don't have, just install it:

bunx skills add nweii/agent-stuff --skill SKILL_NAME

Replace SKILL_NAME with the name listed in the catalog. Two CLI behaviors matter here:

• bunx skills add nweii/agent-stuff skips internal skills. Naming one with --skill opts it in.
• After I move a skill to another folder in this repo, bunx skills update can falsely report internal skills as deleted, even ones I haven't moved. Use INSTALL_INTERNAL_SKILLS=1 bunx skills update (add -g for global installs), or decline removal so it doesn't delete a skill that's still here. If the command can't ask for confirmation, it leaves those skills alone.

For a one-off run, follow Use individual skills without installing. Naming an internal skill explicitly works with this route too.

Tailor a copy (when it carries folder conventions you don't share)

For the ones carrying my folder structure and template names throughout, have your agent read mine and write you your own version instead of installing and then editing:

If I haven't named a skill, run `INSTALL_INTERNAL_SKILLS=1 bunx skills use nweii/agent-stuff` to list the choices, including internal skills. Load the chosen skill with `bunx skills use nweii/agent-stuff --skill SKILL_NAME`, replacing SKILL_NAME with its name. Read the returned instructions and relevant supporting files from the downloaded directory as source material. If the CLI is unavailable, fetch the skill and its supporting files from GitHub.

This is an internal skill written for someone else's setup, so some folder paths, tools, and naming conventions won't match mine. List what needs substituting and ask me about anything you can't infer from my setup.

Write the complete tailored skill folder into my skills directory under a name of my choosing. Include a `metadata.credit` line noting what it was adapted from, and preserve supporting files so references and scripts still work. Leave everything that isn't setup-specific alone. Don't install the original.

Giving your copy its own name keeps a later add or update from overwriting it.

Skill folders

Most skill folders are in skills/. Skills grouped in a plugin are in plugins/<id>/skills/.

Catalog

Auto-generated by scripts/update-catalog.py

Plugins

• Obsidian Tools — Operate your Obsidian vault, write note descriptions, and create Templater and Web Clipper templates using your own conventions.

Skills

• archive-conversation — Create analytical archival summaries of AI conversations, capturing intellectual journeys, key insights, and technical logs. Use when archiving, saving, or documenting a chat session.
• audit-transcription — Audit an existing transcript for likely mistranscriptions. Propose vocabulary repairs or interpretations for review before editing, and flag unrecoverable passages.
• bb-agent-coordination — Use alongside bb-cli when coordinating work across bb threads: delegation, ownership, environment boundaries, steering, reports, review, reconciliation, or handoff. bb-cli owns commands; this skill owns collaboration.
• better-work — Use when improving a prompt before execution or revisiting work the user finds too shallow.
• build-lab — Set up a durable /lab section in a website: a sidebar shell over design experiments, a tested registry, and the host agents-file conventions that make future lab authoring work.
• commit — Create well-formatted commits with conventional commit messages. Use when user asks to commit, wants to commit changes, or needs help with commit messages.
• drafting-writing — Help the user develop their own writing through questions, fragment mining, shaping drafts, or tightening phrases. Use for writing partner sessions that preserve the user’s authorship.
• enrich-from-transcript — Add substantive detail from a transcript to an existing meeting or interview note, or scaffold a note when only the transcript exists. Preserve evidence and exact phrases.
• extract-flow-scenario — Use when extracting an actual workflow, user journey, or operational scenario from a conversation into a sequence of actors, state changes, and pain points.
• file-naming — Propose meaningful filenames from file contents and context. Use for renaming documents or creative files, or standardizing a folder’s filenames.
• herdr-agent-coordination — Use alongside the managed herdr skill whenever recognized agents communicate through Herdr: peer tasks, reports, questions, corrections, return paths, origin checks, or stalled delivery. This skill owns the peer protocol; herdr owns runtime control.
• obsidian-app-css — Use when writing, debugging, auditing, or reorganizing CSS snippets or app themes for Obsidian desktop/mobile, including DOM selectors, CSS variables, platform classes, plugin/theme compatibility, specificity, or cascade conflicts. Excludes hosted Publish, publish.css, and publish.js.
• obsidian-clis — Use when operating Obsidian from the terminal: notes, search, properties, links, tasks, Bases, or plugin development. Uses the app CLI, with notesmd-cli as a headless fallback. Included in Obsidian Tools.
• obsidian-publish-customize — Use when theming or scripting an Obsidian Publish site through publish.css or publish.js, including client-side workarounds. Excludes ordinary Publish usage, desktop themes, and plugins.
• obsidian-templater — Help with templates/snippets for the Obsidian Templater plugin. Use to help generate Obsidian templates from natural language, understand and debug existing tp.* snippets, and adapt vault notes and workflows to Templater when users mention Templater, tp.*, templater syntax. Included in Obsidian Tools.
• obsidian-web-clipper — Use when creating, importing, or debugging Obsidian Web Clipper templates: JSON, variables, filters, selectors, interpreter prompts, and URL or schema triggers. Does not handle general scraping. Included in Obsidian Tools.
• quantify-impact — Draw out defensible metrics from descriptions of work through questions and explicit estimates. Use when a resume bullet, case study, or pitch needs concrete scale or outcome evidence.
• raycast-extensions — Build Raycast extensions — commands, List/Form/Grid views, hooks like useFetch, AI integration, manifest. Use when the user says 'raycast extension', 'add a raycast command', or is working in a project with @raycast/api and a Raycast manifest.
• react-useeffect — React useEffect best practices. Use when writing Effects, derived values, or data fetching. Teaches when NOT to use Effects and better alternatives like useMemo or key props.
• repo-docs — Use when writing, revising, or auditing READMEs, repo docs, RFCs, PR descriptions, or commit messages, or preparing docs for public release. Agent-facing files such as AGENTS.md belong to writing-for-agents.
• semantic-compression — Compress wordy language or loosely expressed thinking into a smaller, sharper form without flattening its meaning. Use when revising an explanation, phrase, sentence, paragraph, concept, label, or coined term whose nuance, implication, or texture must survive the edit.
• set-note-description — Use when writing a note's description property for recall or retrieval. Summarizes a note's own content; parent-period synthesis belongs to periodic-rollup or the environment's specific rollup workflow. Included in Obsidian Tools.
• software-test-quality — Audit software tests for assertion, fixture, mock, or flakiness problems; choose the test level for a risk; or add meaningful regression coverage for recent changes.
• spec-shaping — Shape product ideas into actionable specs and sprint plans. Use when interviewing about a product plan, breaking specs into sprints, or turning vague ideas into well-scoped work.
• suggest-lucide-icons — Pick Lucide icons for a concept, UI placement, or vault note. Searches the full Lucide catalog for real, verified icon names. Use when the user says 'what icon for X', 'suggest a Lucide icon', 'pick an icon', or needs an icon for note frontmatter, a button, or a section header.
• teach-me — Use when the user asks to learn a concept or understand completed work through stepwise tutoring, checks for understanding, or a saved walkthrough or learning record.
• things-app — Use Things 3 to read, capture, schedule, or update tasks and projects, including Today/Inbox views and searches. Supports local or remote access.
• tirith-config — Operate tirith, the terminal guard against homograph URLs, ANSI injection, and pipe-to-shell. Use when editing ~/.config/tirith/policy.yaml, debugging a blocked command or paste, choosing trust / allowlist / download-and-run / TIRITH=0, or quieting a noisy rule.
• transcribe-audio — Transcribe audio or video to text with a locally-run speech-to-text model. Use when the user wants a transcript or captions from a recording, wants to run speech-to-text locally instead of a cloud service, or wants to set up or switch their local transcription model.
• update-changelog — Add or revise a project's Keep-a-Changelog-format changelog, including version bumps and change-type groupings. Use when the user says 'update the changelog', 'log this change', or 'bump the changelog' for a project or plugin.
• use-tailwind-v4 — Tailwind CSS v4 syntax and v3→v4 migration reference. Use when the user mentions 'Tailwind v4', 'upgrade Tailwind', @theme, @utility, or @import 'tailwindcss', or is working in a project whose config indicates v4.
• validating-startup-ideas — Find and validate startup ideas by mining user complaints, crafting premises, and navigating the idea maze. Use when discovering product opportunities, validating ideas, shaping solutions, researching user pain points, or exploring what to build.
• visual-keywords — Generate search keywords for images, screenshots, visual references, or typeface specimens. Use for fuzzy retrieval or searchable Eagle metadata; output keywords rather than prose descriptions.

Internal

Personal workflows tied to my own setup (vault paths, tools, naming conventions). Installable, but expect to adapt SKILL.md to your own folders and conventions before they're useful.

• archive-to-brain — Save an archival summary of an AI conversation to Nathan's Obsidian vault, using the Thinking note template and vault folder conventions to capture intellectual journeys, key insights, and technical logs. Use when archiving a chat session to the vault.
• clip-skills — Save supplied bunx skills output as a repository or individual-skill note in Brain, Nathan’s personal Obsidian vault. Use when capturing a CLI skills listing.
• code-to-pantry — Use when looking up reusable components in Nathan's pantry, bringing one into a project, or saving a code technique to the pantry for reuse.
• create-topic-note — Create a topic note grouping related notes under a common theme, with automatic backlinking to source notes. Triggers: 'group these under a topic', 'create topic note for [[A]], [[B]], [[C]]'.
• nweii-skills — Use when authoring, auditing, migrating, installing, or publishing skills and plugins in Nathan’s library; loading repository skills on demand; or reflecting on a session to improve them.
• obsidian-granola — Import Granola meetings into Obsidian, or process an imported meeting through a saved transcription audit and reviewed enrichment. Resume after a transcript is pasted or corrections are confirmed.
• obsidian-log-commits — Fetches today's GitHub commits across personal and org repos, filters noise, and appends a summary as bullets into the ## Log section of today's daily note. Use when asked to log what was worked on today, check today's commits, or update the daily note with dev activity.
• people-notes — Use for person-centered work in Nathan's Brain vault: prepare context from a profile and linked records, create a canonical person note, or reconcile an existing one.
• periodic-rollup — Use when synthesizing child periodic-note descriptions into a parent week, quarter, or year, or compiling project/topic history. Prefer a vault-specific rollup workflow when the environment provides one.
• save-term — Use when saving an encountered term or a concept coined in conversation as a Brain Term note, with its source quote or a brief gloss.

License

MIT. See LICENSE.