Bkit plugin
A Claude Code plugin links planning, implementation and task checks, retains progress and controls how independently agents work.
- 605 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- L4 runs development stages and tools without ordinary pauses, using Claude Code permissions.
Install
- Works in
Manual install
claude plugin install bkitThe first plugin-install command in the README quick start. Requires a compatible Claude Code.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- L4 runs development stages and tools without ordinary pauses, using Claude Code permissions.
This is third-party code. Review the repository files before installing.
What it does
Bkit organizes development around a plan and specification. An agent clarifies the task, designs a solution, implements it, compares results with requirements and prepares a report. Larger releases can be split into dependent sprints whose state survives sessions. The plugin adds PDCA commands, specialist agents and stage gates. Users choose how much work proceeds before the next pause. Optional team execution distributes tasks among several Claude Code agents.
Who it is for. For developers and small-team leads needing a persistent Claude Code plan with stage checks.
Good fit when
- Split a release into dependent tasks.
- Compare implementation with an agreed specification.
- Resume work after a session ends.
- Use specialist agents for separate tasks.
Not a fit when
- You want a plugin that leaves the development workflow untouched.
- Agents cannot edit code or invoke tools.
- A match score would replace tests and human review.
Example request
Plan a small feature with Bkit and pause for approval before editing code.Limitations
Installation needs Claude Code 2.1.143 or later; some agents and modes require newer versions, so check compatibility before updating. High autonomy lets stages proceed without ordinary pauses; run it under an account whose permissions you have restricted. Multiple agents share model quota and need budget oversight. Specification-match scores are agent assessments, so separately inspect the final patch and the checks actually performed.
How to disable. Disable or remove Bkit through the Claude Code plugin manager. Retain or delete project plans, reports and edits separately.
FAQ
How do I start with one feature?
The quick start uses `/pdca pm my-feature`.
How do I control autonomy?
Use `/control level N`, shared by PDCA and sprints.
How do I plan a release?
`/sprint master-plan` accepts a feature list.
Are agent teams required immediately?
No. Parallel team execution is enabled separately.
What happens when a gate fails?
It can pause progression or trigger a repair loop depending on the condition.
Discussion
No questions or reviews yet.
Related
PluginsA universal workflow plugin for Claude Code: 7 agents, 26 commands, 14 skills and 14 hooks for any project
Three Claude Code plugins: the savvy-flow orchestrator with tiered subagents and two mods for a progress panel and a cache timer
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
A Claude Code and Codex plugin with skills for code research, tests, refactoring and independent review of finished changes
A ready-made Claude Code setup: agents, slash commands, skills, safety hooks, rules and MCP connections in one install
Claude Code plugins with development recipes: agree on the outcome, design, implement with checks and run an independent review
README in short
Bkit retains plans, designs, analysis and completion reports, preserving the reasoning behind changes across interruptions.