PLUGIN brewcode
On this page
Brewcode Overview
Brewcode is a Claude Code authoring toolkit: skills (recipes) define what to brew, agents (brewers) do the work, hooks (processes) handle session start and recovery after compaction.
Install
Install prompt- if Claude refuses, paste this into the chat
Execute these commands in this Claude Code session, one by one, show full output for each, do not skip any, do not summarize:
1. claude plugin marketplace add https://github.com/kochetkov-ma/claude-brewcode
2. claude plugin install brewcode@claude-brewcode
After all commands succeed, run `/reload-plugins`. If `/reload-plugins` is unavailable, tell me to type `exit` and run `claude` again. Run the commands now.What you get
| Part | What it does | Page |
|---|---|---|
setup-status | Dashboard of 11 setups and 23 stamp carriers across brewcode, brewtools, brewdoc; upgrade brings the stale ones current | Skill |
semble-setup | Semantic code-search MCP: install, audit, reindex | Skill |
convention-setup | Coding, testing and architecture conventions with reversible loading | Skill |
teams-setup | Dynamic project agent teams (5-20) with tracing and tuning | Skill |
superreview-setup | Generates a project-tailored deep-review skill | Skill |
rules, skills, agents, e2e | Recurring tools: rules, skill and agent management, E2E tests | All skills |
| 7 agents | skill-, agent-, hook-, mod-creator, bash-expert, internal bc-rules-organizer and setup-upgrader | Agents |
| 2 hooks | session-start and compact-recall (plan, intent, task graph after compaction) | Hooks |
brewcode mods: Prompt placeholders · Plan-mode effort · Reminders · all mods
How it fits
Brewcode layers on top of Claude Code with no external runtime. Skills spawn agents, agents write rules and project files, and the two SessionStart hooks keep sessions anchored. Generic implementation and review roles are generated per project into .claude/agents/ by /brewcode:teams-setup. The brewcode mods add prompt placeholders and plan-mode effort.
Quick start
- See what is already here/brewcode:setup-status
One table of every setup skill: current, stale, partial, missing, plus the exact command per row.
- Convention foundation/brewcode:convention-setup install
Extract coding, testing and architecture patterns into
.claude/convention/before teams and review. - Team install/brewcode:teams-setup install
Create project agents that use those conventions.
- Deep review/brewcode:superreview-setup install
Generate a review skill wired to your agents, gates and conventions.
One setup per session
Interactive -setup runs (install, interviews) fan out subagents and ask questions, so run them one at a time, ideally in a fresh session. setup-status upgrade refreshes the stale setups itself, one agent after another.
Technical details
Naming: -setup vs bare #
A -setup skill installs a mechanism (agents, hooks, a rule, a generated skill) that you then use. Run it once, upgrade it later. Bare-named skills are recurring tools. Every setup answers status, install, upgrade, enable, disable, uninstall, purge. A bare call means status when installed and install when absent; semble-setup and the four brewtools guard setups (manager, agent-deadline, agent-return, agent-router) always default to status, convention-setup to install. Full rule: Skills.
Artifact metadata #
Every artifact a -setup skill installs carries the same stamp across brewcode, brewtools and brewdoc. The version always comes from .claude-plugin/plugin.json, never hardcoded and never the literal unknown: a writer that cannot resolve it refuses to write.
| Field | Values | Where |
|---|---|---|
doc_type | llm | user | skip, unquoted | .md frontmatter only, never JSON |
version | quoted "X.Y.Z", plugin version at install | every carrier |
content_version | quoted "X.Y.Z", last release that changed the content | every carrier |
generated_by | quoted "<plugin>:<skill>" | every carrier |
last_updated | quoted "YYYY-MM-DD" | every carrier except a byte-copied .mjs/.sh/.md (the date would churn the drift check) |
| Carrier | Applies to |
|---|---|
| JSON top-level keys | JSON artifacts |
.md frontmatter | generated .md skills and docs |
// brewcode-meta: one-liner, line 2 | byte-copied .mjs/.sh |
header table (| Version |, | Generated by |, | Last update |) | team.md |
<!-- brewcode-meta: ... -->, line 1 | byte-copied .md |
/brewcode:setup-status reads the stamps back and flags any artifact whose content_version is behind the plugin. One skill’s full contract: superreview-setup.
Plugin structure #
brewcode/
├── .claude-plugin/plugin.json # manifest
├── hooks/ # hooks.json, session-start.mjs, compact-recall.mjs, lib/utils.mjs
├── agents/ # 7 agents (5 public + 2 internal: bc-rules-organizer, setup-upgrader)
├── skills/ # 9 skills
└── templates/ # rule templatesFull tree: file-tree.md.
Agent contracts and tests #
Creator agents and their on-demand references follow Claude Code authoring contracts checked through 2.1.309. They distinguish ordinary agents, conversation forks and skill forks, and return unresolved decisions to the main caller. Local regression suites cover creator contracts, hooks, team lifecycle, convention lifecycle and semantic search; each prints current pass/fail totals, and none replaces a live evaluation.
Related
Installation
Full setup, update and troubleshooting.
Brewcode mods
Prompt placeholders and plan-mode effort.
View on GitHub
Source code, README and configuration.
Update: /brewtools:plugin-update, /reload-plugins, then /brewcode:setup-status upgrade - FAQ