By tal
Refine plans and specs through in-depth interviews using strategic questioning
Execute action items from a plan file in the `plans/` directory.
Create an implementation plan through deep, in-depth interviewing
Fill out the next unfilled phase with implementation details from spec and codebase
Break a spec into high-level, independently testable implementation phases
Refine a plan or spec through in-depth strategic questioning
Uses power tools
Uses Bash, Write, or Edit tools
Has parse errors
Some configuration could not be fully parsed
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Personal plugin marketplace for both Claude Code and Codex.
Add the marketplace from inside Claude Code:
/plugin marketplace add tal/plugin-marketplace
Then install a plugin:
/plugin install <plugin-name>@tal-marketplace
For example:
/plugin install sort@tal-marketplace
/plugin install plan-refiner@tal-marketplace
/plugin install smart-notifications@tal-marketplace
Or from the shell with the claude CLI:
claude plugin marketplace add tal/plugin-marketplace
claude plugin install <plugin-name>@tal-marketplace
Each plugin has its own README with the full feature/command/skill inventory — follow the links for details.
Alternatively, register the marketplace in ~/.claude/settings.json:
{
"extraKnownMarketplaces": {
"tal-marketplace": {
"source": {
"source": "github",
"repo": "tal/plugin-marketplace"
}
}
},
"enabledPlugins": {
"plugin-name@tal-marketplace": true
}
}
Add the marketplace:
codex plugin marketplace add tal/plugin-marketplace
Then install a plugin:
codex plugin install <plugin-name>@tal-marketplace
Each plugin lives under plugins/<plugin-name>/ and can expose metadata for both clients side by side:
plugins/my-plugin/
.claude-plugin/
plugin.json
.codex-plugin/
plugin.json
commands/
skills/
agents/
hooks/
Marketplace catalogs live at:
.claude-plugin/marketplace.json.agents/plugins/marketplace.jsonplugins/.plugins/my-plugin/.claude-plugin/plugin.json if the plugin should be installable from Claude Code.plugins/my-plugin/.codex-plugin/plugin.json if the plugin should be installable from Codex.Claude entry:
{
"name": "my-plugin",
"source": "./plugins/my-plugin",
"description": "What the plugin does",
"version": "0.1.0"
}
Codex entry:
{
"name": "my-plugin",
"source": {
"source": "local",
"path": "./plugins/my-plugin"
},
"policy": {
"installation": "AVAILABLE",
"authentication": "ON_INSTALL"
},
"category": "Productivity"
}
npx claudepluginhub tal/plugin-marketplace --plugin plan-refinerSkills for configuring and writing Karabiner-Elements rules on macOS
OSX desktop notifications for Claude Code events
Generate and apply custom macOS folder icons. Triggers on any 'give these folders icons' / 'iconify my downloads' / 'make my project folders look nicer' request — the skill inspects a directory's subfolders, infers the right style by reading any existing custom icons on peers and the parent (emoji-on-folder, SF Symbol, AI illustration), gathers context from each folder's contents and siblings, and proposes a per-folder icon. Generation falls back gracefully — uses gpt-image-1 / imagen if an API key is present in the environment or a `.env` file, otherwise just emoji or SF Symbol composited onto Apple's GenericFolderIcon. Always shows an HTML preview before touching the filesystem and applies icons via NSWorkspace.setIcon. macOS only — bails immediately on non-Darwin systems.
Resurrected /simplify: reviews changed code for reuse, quality, and efficiency via three parallel agents, then fixes what they find.
Sort and process files in any folder. `/sort` dispatches on file type — videos, images, archives, DMGs, app bundles, documents — and routes each to the right pipeline. Output lands under `<target>/AI Library/` where `<target>` defaults to the folder Claude Code was launched in (so the skill works on ~/Downloads, ~/Desktop, a project folder, or anywhere else). Videos delegate to `/sort-videos` for transcription, OCR, talk/lecture detection with extended summaries, and tagged MP3 export. Images are type-detected (screenshots/memes/photos) and topic-tagged via a vision agent. Documents are text-extracted and classified by an agent into existing folders, with a sensitivity flag for personal info. Archives and DMGs are peeked and routed to app or document or image pipelines. Installers whose app is already in /Applications/ are auto-deleted; nothing else is deleted without confirmation. User overrides and per-extension rules via sort.md / sort.local.md at project and user scope; interactive rule creation via /sort:add-rule.
Interview command for fleshing out big feature plans and specifications
Adaptive interview-driven spec generation with quality review. Automatically adjusts depth based on plan complexity.
Conducts a multi-round interview to refine a rough SPEC.md into a complete, implementation-ready specification with tasks and post-implementation review
'MUST BE USED PROACTIVELY when user mentions: planning, PRD, product requirements document, project plan, roadmap, specification, requirements analysis, feature breakdown, technical spec, project estimation, milestone planning, or task decomposition. Use IMMEDIATELY when user says "create a PRD", "plan this feature", "document requirements", "break down this project", "estimate this work", "create a roadmap", "write specifications", or references planning/documentation needs. Expert Technical Project Manager that creates comprehensive PRDs with user stories, acceptance criteria, technical architecture, task breakdowns, and separate task assignment files for sub-agent delegation.'
Interviewing to understand requirements, and then implementing them using a Manus-style approach.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.