Productivity plugins

Save conversation checkpoints as markdown files and resume them later with full context — shared between Claude Code and Codex

1 skill

Daily Turkish HTML radar report of trending and talked-about projects, with a persistent watchlist (Claude Code)

1 skill

Turns any rough request into a structured, domain-aware expert prompt tuned to the detected Claude or OpenAI/Codex model, with a 12-gate quality review

1 skill

AI Spend Tracker brings your organisation’s OpenAI and Anthropic API usage and costs into one view for finance teams. Connect your provider accounts to understand AI spending without switching between dashboards. Built with love by Moss, the spend management platform.

1 skill

Read and edit Word, PowerPoint, Excel and Hancom files (docx, pptx, xlsx, hwpx) as text, keeping everything the text does not show

1 skill

Lets your personal AI agent protect your inbox, keep important mail visible, and research unfamiliar offers before they reach you.

1 skill

Reverse-engineer several apps, compare and merge their features, check compatibility, and build and test one unified application.

1 skill

Focused proofreading that never rewrites you. Hand it a draft and get back an interactive two-pane page: your text on the left, every suggestion on the right, each one clickable. Accept changes one at a time, or choose between alternative phrasings from a dropdown. Two passes — spelling and grammar, then sentence construction. It corrects form, never substance: it won't change what a sentence claims, strip a hedge, or make casual prose corporate.

1 skill

Build, review, and improve investor pitch decks (seed / Series A, deep tech). Based on YC, Sequoia, a16z and DocSend research.

1 skill

Clean and score Spanish text (0-100) so it doesn't read like AI: invisible characters, Unicode bold, filler, "it's not X, it's Y", em dashes, hashtag walls. Local heuristics, no API. Also an MCP server for 14 other agents.

1 skill

Receive pages scanned with the Mokutori app for iPhone and iPad straight in this agent: pair with a QR code, then read each batch of corrected page images and OCR text.

1 skill

PDF-guided ADT audit and repair for Codex, with optional teacher/reviewer notes and shareable review reports.

1 skill

A working method and tools for driving QGIS with Claude Code: approved plan, numbered PyQGIS scripts, result checks, factual work report.

1 skill

Skill, starter tree and tools for a versioned, secret-free, cross-referenced notes tree: capture changes, audit, bootstrap, merge handoffs, restore drills.

1 skill

A live dashboard of your coding agents, drawn as a pixel-art office. Spot the agent waiting on you, see which projects got the hours, and replay your day, week or month as a timelapse, across Claude Code, Codex, Gemini, Cursor and every account. Type /theoffice.

1 skill

Turn a software goal into parallel work, atomic commits, agent prompts, and clear integration handoffs.

1 skill

Turn a UI decision into real shadcn-composed variants, screenshotted and chosen by clicking instead of describing a preference in words.

1 skill

version-manager CLI usage guide and .appversion.yml authoring skills for AI agents and humans.

Interaction Swift library API guide for AI agents and developers.

Watch source files for AI!/AI?/AI. marker comments and surface them in the session (plugin-monitor port of the pi-watcher extension).

Coordinate Claude Code, Codex, and generic CLI agents across sibling Git worktrees.

Visual workspace for Claude Code — Excalidraw diagrams with inline PNG previews, tldraw canvases, and Markdown notes with wikilinks.

Asana CLI and agent skills, with an opt-in MCP server

Play the latest Rabbi Eli Stefansky / MDY Daf Yomi shiur while Claude Code works.

Generate product or company names, then screen them for brand, GitHub, and domain collisions.

Opt-in bundle: one install pulls in brain plus every plugin in Anthropic's official marketplace (except the output styles that contradict terse mode). Heavy by design — each enabled plugin adds context every turn; service connectors still need your own login, LSP plugins their language binaries.

Routes Agent calls to the pool with spare quota, appends the verdict to `bullswarm run` results, and draws the runs strip and the workflow pane. Needs the bullswarm CLI on PATH and CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1 (function hooks are early access).

Guardrails for a memories/topics/<topic>/units/ content workflow: a unit's frontmatter taxonomy (channels, type, tags) selects which writing rules apply, script rules catch what LLMs are bad at (character limits, banned phrases), a judge weighs the rest, and a unit cannot reach status: published without a cited user approval and a published_url.

Guardrails for a memories/tasks/ workflow: a task's body must be the human's ask (cited, and nothing else), its evidence must resolve, its in_review claim is judged, and a turn may not end with open work. Four use-case guardrails in the nature format, shipped as a self-testing plugin. Requires the sloprail plugin (it installs the engine hooks).

The missing smart dispatcher and dashboard for Codex tasks