Productivity plugins

Learn from your workflow patterns and suggest skills, hooks, and agents to automate repetitive work

2 skills · 1 command · 2 agents

Evidence-first startup ideation skills for problem inventory, problem discovery, assumption grilling, validation tests, interview synthesis, and narrow v1 scoping.

5 skills

Checkpoint and resume work across Claude Code, Cursor and Codex. Records what was ruled out so the next agent does not repeat it. Local-first: your memory is markdown on your disk.

2 skills · 1 command · 1 MCP server · hooks

Dataverse ALM CLI — plugin registration, web resources, and solution deploy via Git-tracked, unmanaged solutions. Includes migration guidance off spkl, Daxif, PACX, and ALM Accelerator.

5 skills

Multi-model orchestration: routing table (scout recon, mechanical, reasoning, cross-vendor peer tiers), six-field delegation contracts with dispatch templates, DONE/BLOCKED status vocabulary, budget modes (economic/thorough), escalation ladder, blind-parallel high-stakes path with reconcile rules, and a verification stage. Installs the /orchestrate command and three model-pinned agents (deep-reasoner→Opus, fast-worker→Sonnet, scout→Haiku). Codex peer requires the Codex CLI (degrades gracefully without it).

1 skill · 1 command · 3 agents

Puts the project's committed mnema record into the session's opening context without the agent asking, and connects the mnema MCP server so it can record its own work.

3 skills · 1 MCP server · hooks

Coordinated staged implementation with planner, coordinator, bounded orchestrator, implementer, and validator roles.

5 skills

autoresearch: a skill-only autonomous experiment loop — propose → measure → keep winners → repeat, against any numeric objective, with noise-floor gating so nothing is banked on noise. /ar:start, /ar:resume, /ar:status, /ar:report, /ar:stop. State stays in the target repo's ./.ar/. No hooks, no MCP.

5 skills

Fantasy sports analysis for your ESPN, Yahoo, and Sleeper leagues.

4 skills · 1 MCP server

Learn harnesses by building a world of collaborating bot characters and tools.

5 skills

Writing and review skills for engineering prose drafted by agents: commit messages, PR descriptions, issues, review comments, code comments, design and intent docs, READMEs, and explanations. Written for the reader who saw none of the session that produced the text.

5 skills

Croatian plant knowledge and connected garden workflows.

3 skills · 1 MCP server

Give Codex deterministic repository impact analysis, authored intent, and fail-closed migration planning.

2 skills · 1 MCP server · hooks

Session capture installed once per host: every Claude Code event under <root>/<session_id>/, every file a tool call writes copied to written/, every publication (Artifact, GitHub, commit, push, npm) listed with its prompt, and a binding line per session start, end and rename with its tmux pane, command line, name history, team and chronicle episode, and every /btw side question with the story of its fork. Carries the session-continuity skill (T1).

2 skills · 1 command · hooks

Evaluates a folder's outstanding work (typically team-status's stage-map) and gets it moving on whatever track it actually needs. `triage` sorts outstanding items into direct-fixable housekeeping, real work that needs team-intake first, and things only a human can do — then dispatches the first two. `dispatch` decides whether build-ready items should build in parallel worktree efforts, sequentially, or as a single normal session, then dispatches and babysits the runs through to merge. Generic — works on any project using the delivery-team pipeline conventions.

1 skill · 3 agents

Lessons that coding agents record, an independent gate agent trusts, and later sessions reuse.

2 skills · 1 agent · hooks

Turn daily work into language practice. Auto-detects and corrects target language in your workflow.

3 skills · hooks

Sync Claude Code sessions to Jieli threads and attach Jieli thread trailers to Claude-created commits.

3 skills · hooks

Always-on anti-slop rules for Claude Code: no filler, no invented facts, no sycophancy. Adds profession modules, a slop linter, and MCP tools.

1 skill · 1 command · 1 MCP server · hooks

The default edition. Full build plan with tracker, per-step prompts, verify lines, rollback, landmines, and a not-building list. The agent never commits: it hands you one commit + push block per step.

4 skills

Tool-neutral edition. Same full format with generic model tiers and a commit convention you choose per plan (human commits, agent commits, or agent commits and pushes).

4 skills

Lightweight edition for 3-8 step features and fixes. One compact plan file: scope, tracker, opening prompt, and per-step prompts with verify lines. Small and cheap to load.

4 skills

Sync Claude Code sessions to Jieli threads and attach Jieli thread trailers to Claude-created commits.

3 skills · hooks

Audit docs, preview gated README edits, review trust policies and draft unreleased notes with local tools.

4 skills

STE writing and visual explanations without host-tool restrictions. Native checked tools require a configured runtime; see the host support matrix.

4 skills

Local-only agent skills maintained for both runtimes — ingest, ai-usage-card, sitrep.

4 skills

Industry Rockstar Brain — talk to your Rockstar / IR agents and get every reply as live UI: inline widgets in chat, a per-user command center, and one-click live artifacts. Bundles the IR MCP server.

1 skill · 2 commands · 1 MCP server

Organize browsers, downloads, desktop, and notes. Backs up first, shows you counts, waits for approval, and hands you the undo command. Adding support for a new app is one markdown file.

4 skills

Draft, clean up, and refine GitHub issues — optimized for AI agent consumption

3 skills

A curated set of Claude skills built by Karbon for accounting firms. Free to the profession.

3 skills