发现 Skills

Canonical browser-facing QA workflow for report-only, acceptance-gate, and fix-loop checks using Browser, Chrome, Playwright, web-evidence, browser-session, and workflow evidence primitives.

Deliver a dispatch-ready plan by sprint: split plan tasks into PR lanes, dispatch subagent work, enforce acceptance gates, and advance through final integration without main-agent implementation.

Logs system prompts for automatic versioning and component-level optimization. Upgrades an existing Greenflash integration with prompt tracking and analytics.

Find prompt and model quality issues using real conversation data, with specific optimization recommendations. Can implement prompt fixes and model switches directly in your codebase.

Analyzes and protects dependencies in Next.js + TypeScript projects. Invoked by prompt-classifier on DEPS. Checks compatibility, conflicts, and security. Chains verification and commits changes on approval.

Abandon the currently claimed issue, releasing the local claim so another agent can pick it up.

Tools and workflows for managing the Supabase backend in the FastAPI Todo App.

Audits and optimizes the Design System of a Claude Design project (claude.ai/design). Verifies completeness (fonts, tokens, brand refs, a11y), identifies gaps, applies fixes when possible, and delivers a report. Intended to be invoked before any large-scale design generation campaign to ensure outputs inherit from a clean system. Relies on `design:design-system`, `design:design-critique`, `design:accessibility-review`, `marketing:brand-review`.

Detects and resolves pending Git conflicts from the inter‑PC auto-sync system. Reads the marker `.conflicts-pending.yaml`, analyzes the two backup versions (ours/theirs) in plain human language per file, identifies the nature of each divergence (different additions, edits in the same section, deletion vs keep), proposes clear A/B/C options with a motivated recommendation (caveman OFF), executes the chosen resolution after validation by Florent, then cleans up the marker. Triggers: "/conflicts", "resolve conflicts", "git conflicts", "there's a conflict", "fix conflict". Automatically invoked as soon as the reminder "⚠️ N Git conflicts pending" injected by the UserPromptSubmit hook appears.

Launches the Wisper app (speak-app-dev) in the background. PROACTIVE — automatically launch after any code change in speak-app-dev/ so the user can test. Do not wait for them to request it.

AI-assisted pair programming with multiple modes (Driver, Navigator, Switch), real-time verification, quality monitoring, and comprehensive testing. Supports TDD, debugging, refactoring, and learning sessions. Features automatic role switching, continuous code review, security scanning, and performance optimization with truth-score verification.

One-prompt API integration pipeline. Takes any API (docs URL, name, or description) and produces a working CLI wrapper + skill + tests.

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

Assess and plan repository-level refactors: architecture, structure, ownership, dependency direction, target shape, migration order, execution, review, verification, and .cflow resume. Use after the diagnostic frame is confirmed; use cf-mr-wolf first when framing is unclear or unconfirmed, and cf-architecture when architecture context is missing or stale.

Create a WBS (Work Breakdown Structure) based on the contents of the PJC. Follow the skill's references (concept, five steps, structure definition, checklist), and keep the success criteria and scope consistent with the PJC. Activated by commands such as "create WBS" and "reflect WBS into spreadsheet".

Creates Bug, Improvement, or Task tickets in the current sprint on ClickUp using the correct template for the active project. Use whenever the user wants to log a bug, report an issue, propose an improvement, or create a task ticket. Also triggers when the user shares a screenshot or video of a problem and wants it tracked. Never attach files — photos or videos shared by the user are for context only.

API Platform 4.2 reference for building REST and GraphQL APIs with Symfony. Use this when creating API resources, configuring operations, working with DTOs, state providers/processors, serialization, filters, validation, security, pagination, or any API Platform-related code. Triggers on: API Platform, ApiResource, ApiProperty, ApiFilter, State Provider, State Processor, DTO, normalization groups, denormalization, OpenAPI, Hydra, GraphQL, REST API, collection operations, item operations, subresources, ApiTestCase.

Capture frames or clips from RTSP/ONVIF cameras.

Operate the boring-budget Go CLI in deterministic, agent-safe mode using stable JSON envelopes, exit/error codes, and non-destructive data semantics. Use when tasks involve setup, category/label/entry/cap workflows, reports and balances, or data import/export/backup/restore for boring-budget.

Databricks engineering standards for safe, efficient interaction with Azure Databricks workspaces via the CLI. Use when running SQL queries, exploring Unity Catalog, auditing permissions, monitoring jobs, or managing warehouses. Covers authentication, safety guardrails, and operational patterns.

Explains your product's technical architecture in plain English — so you understand what's being built, can make informed decisions, and can describe it to others. Use when an agent has set up your tech stack and you're not sure how the pieces connect, or when you need to explain your system to a potential co-founder, investor, or hire. Produces an architecture-overview.md in plain language with a simple diagram.

Use when exporting LSPRAG reference analysis (getReferenceInfo) as a standalone module or wiring it to custom LSP/MCP clients.

GitHub issue CRUD -- create, read, update, close with label/milestone/assignee suggestions and Jira linking

Manage Synnovator platform data: 7 content types (event, post, resource, rule, user, group, interaction) and 9 relationship types via a file-based YAML+Markdown engine. Data stored in PROJECT_DIR/.synnovator/. Use when performing CRUD operations on platform content, managing activities/competitions, teams, posts, user interactions (likes/comments/ratings), or testing user journeys. Trigger: any request involving Synnovator platform data, user journey simulation, or content management.