amitkot
UserClaude Code plugin marketplace with Git fundamentals and GitButler CLI skills.
Categories
Indexed Skills (20)
gitbutler-cli
MANDATORY when on 'gitbutler/workspace' branch. Use `but` commands instead of git for commits, pushes, staging. Key commands: `but status -v`, `but stage <file> <branch>`, `but commit -m 'msg' <branch>`, `but push <branch>`. Invoke this skill BEFORE any git commit/push when gitbutler/workspace is the current branch.
comment-non-obvious-steps
Use when writing or editing code that contains repeated-looking calls on different structures, bitmask/bit-fiddling, ordering-sensitive sequences, or multi-step state mutations across data structures — adds a short inline comment per step explaining WHY each step is distinct, so a reader doesn't stop and ask "wait, why are there three of these?". Does not apply to trivial code with self-explanatory identifiers.
one-test-one-thing
Use when writing or reviewing a unit/integration test — enforces a clear scenario-shaped name (no `_and_`), Arrange/Act/Assert structure with visible separation, and one logical behaviour per test. A failing test name should describe exactly what regressed without opening the file. Multiple scenarios become multiple tests, not one test with branches or `_and_` in the name.
conversation-search-setup
Use when the user asks to "set up conversation search", "install conversation search", "install cc-conversation-search", "/conversation-search:setup", or when the `conversation-search` run skill reports that the CLI is missing. Idempotently installs (or upgrades) the upstream `cc-conversation-search` PyPI tool via `uv tool` and initialises the local SQLite FTS5 index at `~/.conversation-search/index.db`. Safe to re-run.
conversation-search
Find and resume Claude Code conversations by searching topics or filtering by date. Returns session IDs and project paths for easy resumption via 'claude --resume'. Use when user asks "find that conversation about X", "what did we discuss", "what did we work on yesterday", "summarize today's work", "show this week's conversations", "recent projects we accomplished", or wants to locate past work by topic, date, or time period (yesterday, today, last week, specific dates).
git-fundamentals
Essential Git version control knowledge including repositories, commits, branches, remotes, and common workflows. Use when helping with basic Git operations, teaching Git concepts, or as foundation for advanced Git tools. Activate for questions about Git basics, standard workflows, or version control fundamentals.
marketplace-author
This skill should be used when the user asks to "create a skill in this repo", "add a skill to claude-code-tools", "add a skill to the marketplace", "update an existing skill", "ship this skill", "bump skill version", or otherwise authors/edits skills in the amitkot/claude-code-tools marketplace. Handles directory layout, version bumps across SKILL.md / plugin.json / marketplace.json, CHANGELOG entries, and the create → commit → push → PR → /plugin → /reload-plugins workflow.
offline-questions
Use when the user wants to park a question for later — fires a background subagent that researches with full session context and appends a structured answer to ~/Desktop/Offline_Questions_and_Answers.md. Trigger phrases include "park this question", "research this offline", "I care about this but not now", or invocation via /offline-question.
ralph
Interactive planning and build loop for Claude Code. Generates specs, implementation plans, and iterative build infrastructure (loop.sh) through clarifying questions. Use when starting a multi-step feature, bugfix, or refactoring task.
deliverable-on-disk
Use when about to mark a task as complete, done, finished, or shipped — verifies the deliverable actually exists as a file on disk, in a commit, or as a posted artefact. Blocks claiming completion based on chat output alone.
ephemeral-state-in-tmp
Use when about to write a state, cursor, marker, scratch, or progress file for a cron loop, monitor task, polling job, recurring schedule, or any session-scoped/ephemeral data — directs the file to /tmp/<name>.json instead of ~/.claude/ or other config dirs.
extend-not-duplicate
Use when about to write a new script, command, or file that is near-identical to an existing one with small variations — adds a CLI flag, parameter, or config option to the existing script instead of creating a parallel copy.
plans-and-investigations-location
Use when about to write an implementation plan, investigation note, design sketch, post-mortem, or any markdown capturing personal process inside a project repo — and especially when about to save under `docs/plans/`, `docs/investigations/`, or `docs/design/`. Routes personal/in-flight work to `.local/<kind>/` (gitignored) and reserves `docs/<kind>/` for shared, team-reviewed artefacts.
deepwiki
ALWAYS use when the user mentions an external repository, library, or protocol by name. Queries any public GitHub/GitLab/BitBucket repository to answer questions about its code, APIs, behavior, or implementation without cloning it.
pr-update-description
Update a PR with a clear explanation of the changes. Use when the user asks to update, write, or improve a PR description, or mentions "pr description", "update PR", or "describe PR".
execute-beads-plan
This skill should be used when the user asks to "execute implementation plan", "run beads plan", "start working on plan", "execute tasks from plan", or wants to run an implementation plan that was created with plans-to-beads. Orchestrates parallel task agents with verification and change handling.
plans-to-beads
This skill should be used when the user asks to "convert plan to beads", "create beads from plan", "break down plan into tasks", "create implementation plan", or has a plan file to track with beads. Transforms a plan into a tracked implementation plan with beads, dependency graph, and verification criteria.
cmux-browser
Use this when testing the dashboard with a web client in interactive sessions. Best for testing frontend tools, verifying UI behavior, checking page loads, validating URL routing, or interacting with the browser. Preferred over playwright-testing for interactive work — the browser opens as a visible split pane and results return inline.
playwright-testing
ALWAYS use this when testing a website with a web client. This skill should be used when testing frontend tools, verifying UI behavior, checking page loads, validating URL routing, or interacting with the browser in any way.
slack-notify
Use when the user asks to send Slack notifications, create task threads, post progress updates, read thread replies, or wants session activity reported to Slack.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.