bdk

Cross-cutting architectural analysis - layer boundaries, DI, design patterns, data flow, directory structure, import direction

Layer-group code reviewer - deep review of assigned source files and their tests, produces structured findings

Find unused functions, methods, variables, and unreachable code blocks using reference checking

Find duplicated code and extractable patterns - searches changed symbols for literal duplicates, structural patterns, and intra-function duplication

Fast codebase exploration with code-review-graph MCP priority - searches code, symbols, patterns, dependencies

Delegate here to analyze stderr output, error logs, stack traces, and debug command failures. Fast triage of what went wrong.

Detect and run project-appropriate static analysis tools (lint, format, type check) across Python, JS, Go, Rust and other stacks

Dry-run implementation plans step by step - trace concrete data through proposed code, invent edge cases from problem description, cross-check against impact analysis. Used by verify-plan skill.

Run test suite and report results. Pass test targets (files, dirs) in prompt or omit for full suite.

Internet research for debugging, finding solutions, and gathering technical information. Searches GitHub issues, Stack Overflow, Reddit, forums, and documentation.

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

Generate conventional commit message based on git changes

Run code review with dynamic agent scaling (3-13 agents based on change size)

Create Architecture Decision Records (ADRs) based on MADR template. Use when the user asks to "create an ADR", "document a decision", "write an ADR", or provides decision context that needs to be formalized.

Create comprehensive implementation plan with exploration and trade-off analysis

Debug issues through structured investigation, failing test creation, and targeted fixes. Use when the user provides an error message, traceback, steps-to-reproduce, or describes unexpected behavior.

Execute an implementation plan, then run tests and static analysis, fixing issues

Generate comprehensive architecture documentation for complex code modules with Graphviz diagrams and examples. Use when user asks to "explain this code", "document the architecture", or wants to understand how a module/system works.

Systematically debug issues using graph-powered code navigation — trace call chains, find execution paths, correlate with recent changes.

Navigate and understand codebase structure using the knowledge graph — stats, architecture overview, community modules, call relationships.

Plan and execute safe refactoring using dependency analysis — find dead code, preview renames, verify no critical paths broken.

Risk-aware code review using knowledge graph — detect changes, trace impacted flows, check test coverage, recommend merge decision.

Compile Graphviz diagrams (.dot files) to SVG in-place. Use when working with .dot files or markdown with Graphviz code blocks, or when user says "compile diagrams", "generate SVG from dot files", "update documentation diagrams".

Propose better object-oriented architecture for complex, procedural code

Restore saved work progress from .bdk/save-progress/ to resume a previous session. Loads referenced plans, creates TaskList items for remaining work.

Save current work progress to .bdk/save-progress/ for resuming in a future session. Use when context is large, work is in progress, and you want to stop and continue later.

Initialize .bdk/settings.json for this project. Run once per project when BDK blocks session start with missing settings.

Rigid TDD process for writing and verifying tests before implementation. Use when implementing any feature or bugfix. Receives test case bullet points from the plan and enforces red-green cycle.

Use when existing architecture documentation needs refreshing after code changes, when running periodic doc maintenance, or when user asks to "update docs", "refresh documentation", or "sync docs with code".

Verify implementation plans before execution using a 4-agent pipeline. Use when you have a written plan (from /bdk:create-plan or manual) and want to check if it will actually work before writing code.

7 hooks across 2 events