Loading...
Loading...
Found 5,124 Skills
Set up and run an autonomous experiment loop for any optimization target. Use when asked to start autoresearch or run experiments.
Use when bootstrapping skill-loop in a repository, creating a GitHub issue loop starter, or writing and updating skill-loop.yml and starter skills.
Leverage OpenAI Codex/GPT models for autonomous code implementation. Triggers: "codex", "use gpt", "gpt-5", "let openai", "full-auto", "用codex", "让gpt实现". Use this skill whenever the user wants to delegate coding tasks to OpenAI models, run code reviews via codex, or execute tasks in a sandboxed environment.
This skill should be used when the user asks to "edit a video to music", "create a beat-synced edit", "make a montage", "sync cuts to beats", "cut a video to the beat", "make a music video edit", "edit clips to a song", "build FFmpeg filters for video editing", or mentions combining video clips with audio tracks using timed cuts. Provides knowledge of the EditPlan format, FFmpeg filter_complex construction, and beat-sync editing workflows.
Use when a coding task should be driven end-to-end from issue intake through implementation, review, deployment, and acceptance verification with minimal human re-intervention.
Create subtasks in Jira from a previously generated task plan. Reads the plan from docs/<TICKET_KEY>-tasks.md and creates one Jira subtask per task under the parent ticket. Use when the user says "create subtasks", "push tasks to Jira", "sync plan to Jira", "create Jira tickets", "make subtasks for PROJECT-1234", or anything about turning a plan into Jira issues. Also triggered by the orchestrating-jira-workflow skill as Phase 4. Requires the task plan to already exist (run planning-jira-tasks first if it does not). Use this skill even if the user just says "push to Jira" or "create the tickets" after a planning phase — those are subtask creation requests.
Execute a single task from a Jira task plan using a structured pipeline of specialist subagents: planning, testing, refactoring, implementation, documentation, code-quality review, architecture review, security audit, and requirements verification. The user must specify which task number to execute. Use when the user says "execute task 3", "work on task 2", "implement task 1", "start task 5 for PROJECT-1234", or "run task N". Also triggered by the orchestrating-jira-workflow skill as Phase 5 of the end-to-end pipeline (called once per task). Requires that the task plan exists at docs/<TICKET_KEY>-tasks.md. Executes ONLY the specified task — never continues to the next one without explicit user approval.
Automatically create PRs for registering insights, patterns, and workflows obtained from the current project as skills to the TBSten/skills repository. It performs the full process consistently: collecting insights from the project's CLAUDE.md, .claude/rules/, .claude/skills/, and codebase, packaging them into reusable skills, and creating the corresponding PR. Use when requested: "Register insights to the skill repository", "contribute skill", "Share this insight", "Register as a skill", "Compile insights into a PR", "Turn this pattern into a skill". gh CLI and git must be installed.
Use Claude Code to simplify the current pull request by safely reducing unnecessary scope, complexity, and noise while preserving the intended outcome.
Run Gemini CLI review against the current branch, fix only the review comments that are still valid for the current codebase, and leave invalid comments unchanged.
Complete GitHub pull requests by iterating on CI and review feedback until the PR is ready.
Run Codex, Claude Code, and Gemini CLI reviews against the current branch concurrently, deduplicate the findings, and fix only the review comments that are still valid for the current codebase.