skills/kochetkov-ma__claude-brewcode
1
0
Fork 0
mirror of https://github.com/kochetkov-ma/claude-brewcode.git synced 2026-09-14 20:16:41 +08:00
Code Issues Packages Projects Releases Wiki Activity

v4.0.1: purge v3.19.0 task-system doc debt from site; delete workflow page; reconcile brewcode skill/agent/hook counts; document grepai self-install

Browse Source
This commit is contained in:
kochetkov-ma
2026-06-30 21:23:10 +01:00
parent 3055aaf33b
commit b06d8179d4
24 changed files with 145 additions and 517 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.claude-plugin/marketplace.json
+5 -5
View File
@@ -6,13 +6,13 @@
},
"metadata": {
"description": "Claude Code plugin suite: brewcode for infinite task execution, brewdoc for documentation tools, brewtools for text utilities, brewui for UI/visual/creative tools",
"version": "4.0.0"
"version": "4.0.1"
},
"plugins": [
{
"name": "brewcode",
"description": "Brewcode - full-featured development platform for Claude Code: infinite focus tasks, prompt optimization, skill/agent creation, quorum reviews, rules management",
"version": "4.0.0",
"version": "4.0.1",
"category": "productivity",
"keywords": [
"brewcode",
@@ -46,7 +46,7 @@
{
"name": "brewdoc",
"description": "Brewdoc - Claude Code documentation tools: auto-sync for skills/agents/rules, my-claude installation docs, memory optimization, md-to-pdf conversion",
"version": "4.0.0",
"version": "4.0.1",
"category": "productivity",
"keywords": [
"brewdoc",
@@ -74,7 +74,7 @@
{
"name": "brewtools",
"description": "Brewtools - universal utilities for Claude Code: text optimization, humanization, secrets scanning",
"version": "4.0.0",
"version": "4.0.1",
"category": "productivity",
"keywords": [
"brewtools",
@@ -102,7 +102,7 @@
{
"name": "brewui",
"description": "Placeholder for future UI/visual/creative tools (currently empty, installable)",
"version": "4.0.0",
"version": "4.0.1",
"category": "productivity",
"keywords": [
"ui",
RELEASE-NOTES.md
+18
View File
@@ -2,6 +2,24 @@
---
## v4.0.1 (2026-06-30)
> Docs-only follow-up to v4.0.0 — purges leftover v3.19.0 task-system documentation debt from the site so docs match the real plugin.
> Docs: [brewcode overview](https://doc-claude.brewcode.app/brewcode/overview/) | [agents](https://doc-claude.brewcode.app/brewcode/agents/) | [grepai](https://doc-claude.brewcode.app/brewcode/skills/grepai/) | [superreview](https://doc-claude.brewcode.app/brewcode/skills/superreview/) | [getting-started](https://doc-claude.brewcode.app/getting-started/) | [quickstart](https://doc-claude.brewcode.app/quickstart/) | [installation](https://doc-claude.brewcode.app/installation/)
### docs
#### Removed
- **workflow page:** deleted `brewcode/workflow.mdx` (described the removed KNOWLEDGE/plan/start/handoff pipeline) + its `navigation.ts` entry
- **dead skill/agent refs:** purged all links to removed skills (`/brewcode:setup`, `/brewcode:plan`, `/brewcode:start`, `/brewcode:standards-review`, `/brewcode:teardown`) and removed agents (`bc-coordinator`, `bc-knowledge-manager`) across getting-started, quickstart, installation, overview, agents, and skill/agent pages
- **task-system framing:** removed dead KNOWLEDGE auto-accumulation / post-task / coordinator-loop / "infinite task" descriptions
#### Changed
- **brewcode counts reconciled:** 9 skills (spec, grepai, convention, rules, superreview, skills, agents, teams, e2e), 10 agents, 2 hooks; suite total 4 hooks
- **grepai:** documented self-install behavior on its MDX page
- **superreview:** now documented as the generator of the project-tailored deep-review skill (review + standards merged)
---
## v4.0.0 (2026-06-30)
> **MAJOR / breaking** — injection-hook architecture removed across all plugins; plugin-root now resolved natively via `${CLAUDE_PLUGIN_ROOT}`.
brewcode/.claude-plugin/plugin.json
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "brewcode",
"version": "4.0.0",
"version": "4.0.1",
"description": "Brewcode - full-featured development platform for Claude Code: infinite focus tasks, prompt optimization, skill/agent creation, quorum reviews, rules management",
"author": {
"name": "Maksim Kochetkov",
brewcode/README.md
+1 -1
View File
@@ -4,7 +4,7 @@
| Field | Value |
|-------|-------|
| Version | 4.0.0 |
| Version | 4.0.1 |
| Skills | 13 |
| Agents | 12 |
| Hooks | 2 |
brewcode/package.json
+2 -2
View File
@@ -1,6 +1,6 @@
{
"name": "claude-plugin-brewcode",
"version": "4.0.0",
"version": "4.0.1",
"description": "Infinite task execution with automatic handoff for Claude Code",
"keywords": [
"claude-code",
@@ -36,6 +36,6 @@
},
"claude-plugin": {
"name": "brewcode",
"version": "4.0.0"
"version": "4.0.1"
}
}
brewdoc/.claude-plugin/plugin.json
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "brewdoc",
"version": "4.0.0",
"version": "4.0.1",
"description": "Brewdoc - Claude Code documentation tools: auto-sync for skills/agents/rules, my-claude installation docs, memory optimization",
"author": {
"name": "Maksim Kochetkov",
brewtools/.claude-plugin/plugin.json
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "brewtools",
"version": "4.0.0",
"version": "4.0.1",
"description": "Brewtools - universal utilities for Claude Code: text optimization, humanization, secrets scanning",
"author": {
"name": "Maksim Kochetkov",
brewui/.claude-plugin/plugin.json
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "brewui",
"version": "4.0.0",
"version": "4.0.1",
"description": "Brewui -- placeholder for future UI/visual/creative tools",
"author": {
"name": "Maksim Kochetkov",
web/docs/src/content/docs/brewcode/agents.mdx
+6 -10
View File
@@ -1,6 +1,6 @@
---
title: "Agents"
description: "All 12 Brewcode agents: models, triggers, tools"
description: "All 10 Brewcode agents: models, triggers, tools"
order: 12
---
@@ -24,10 +24,8 @@ from developer to task coordination.
| 6 | agent-creator | opus | Read, Write, Edit | Agent creation and improvement |
| 7 | hook-creator | opus | Read, Write, Edit, Bash | Hook creation and debugging |
| 8 | bash-expert | opus | Read, Write, Edit, Bash | Bash/sh script creation |
| 9 | bc-coordinator | haiku | Read, Write, Edit, Bash | Internal: spawned by /brewcode:start + post-task hook |
| 10 | bc-knowledge-manager | haiku | Read, Write | Internal: spawned by /brewcode:start |
| 11 | bc-grepai-configurator | opus | Read, Write, Edit, Bash | Internal: spawned by /brewcode:grepai |
| 12 | bc-rules-organizer | sonnet | Read, Write, Edit, Bash | Internal: spawned by /brewcode:rules |
| 9 | bc-grepai-configurator | opus | Read, Write, Edit, Bash | Internal: spawned by /brewcode:grepai |
| 10 | bc-rules-organizer | sonnet | Read, Write, Edit, Bash | Internal: spawned by /brewcode:rules |
<CardGrid>
<Card title="developer" icon="code" href="/brewcode/agents/developer/">
@@ -57,7 +55,7 @@ from developer to task coordination.
</CardGrid>
:::note[Internal agents]
`bc-coordinator`, `bc-knowledge-manager`, `bc-grepai-configurator`, and `bc-rules-organizer` are internal. They are spawned automatically by brewcode skills and hooks, are not user-invokable, and have no dedicated page.
`bc-grepai-configurator` and `bc-rules-organizer` are internal. They are spawned automatically by brewcode skills and hooks, are not user-invokable, and have no dedicated page.
:::
## Models and cost
@@ -66,12 +64,10 @@ from developer to task coordination.
|-------|--------|-------------|
| **opus** | [developer](/brewcode/agents/developer/), [reviewer](/brewcode/agents/reviewer/), [architect](/brewcode/agents/architect/), [skill-creator](/brewcode/agents/skill-creator/), [agent-creator](/brewcode/agents/agent-creator/), [hook-creator](/brewcode/agents/hook-creator/), [bash-expert](/brewcode/agents/bash-expert/), bc-grepai-configurator | Complex tasks, deep analysis, code generation |
| **sonnet** | [tester](/brewcode/agents/tester/), bc-rules-organizer | Balanced tasks, review, optimization |
| **haiku** | bc-coordinator, bc-knowledge-manager | Lightweight operations: statuses, knowledge extraction, compaction |
<Callout type="tip" title="Automatic selection">
Agents are selected automatically based on the Phase Registry in PLAN.md.
The [/brewcode:start](/brewcode/skills/start/) skill reads agent assignments from the phase table
and launches the corresponding agents via Task API.
Agents are launched automatically by brewcode skills via the Task API.
Each skill spawns the relevant agent for its domain — no manual agent selection required.
</Callout>
<Callout type="note" title="Moved Agents">
web/docs/src/content/docs/brewcode/agents/agent-creator.mdx
+1 -1
View File
@@ -123,7 +123,7 @@ Description patterns by ambiguity level:
| Some — 1-2 other agents | Single-line + detailed Triggers | `"Creates bash scripts. Triggers: create script, bash script, shell script"` |
| High — creator/meta agents | Multi-line + 2-3 `<example>` blocks | See agent-creator.md source |
Subagent spawn constraint: CC 2.1.172 allows nested subagent spawning up to 5 levels deep. However, brewcode workflow requires spawning only from the main conversation — the 2-step report protocol (`post-task.mjs`) binds the task lock to one session and delivers report/coordinator instructions back to the spawning conversation. Nested spawns bypass session binding, `KNOWLEDGE` injection, and the coordinator loop. This is a **brewcode workflow policy**, not a Claude Code platform limitation.
Subagent spawn note: CC allows nested subagent spawning, but prefer spawning from the main conversation so the orchestrator can integrate each agent's result directly. Brewcode agents are stateless and resolve plugin paths via `${CLAUDE_PLUGIN_ROOT}`.
Known active bugs: [#29423](https://github.com/anthropics/claude-code/issues/29423) (subagents skip CLAUDE.md), [#27736](https://github.com/anthropics/claude-code/issues/27736) (`skills:` in plugin agents), [#31392](https://github.com/anthropics/claude-code/issues/31392) (global agents not discovered in v2.1.70+).
</details>
web/docs/src/content/docs/brewcode/agents/reviewer.mdx
+1 -1
View File
@@ -161,7 +161,7 @@ DECISION: Approve / Changes required / Reject
Full agent definition, disallowed tools, and review protocol.
</Card>
<Card title="Brewcode overview" icon="document" href="/brewcode/">
All brewcode agents and skills — infinite task execution, quorum reviews, handoff.
All brewcode agents and skills -- quorum reviews and multi-agent workflows.
</Card>
</CardGrid>
web/docs/src/content/docs/brewcode/agents/skill-creator.mdx
+2 -2
View File
@@ -118,7 +118,7 @@ skill-name/
|---|---|
| **Progressive Disclosure** | Always — L1 name+desc in context, L2 SKILL.md on trigger, L3 references on demand |
| **Reference Splitting** | 2+ modes, >50 lines/mode, >300 lines total — detect mode → load only matching `references/{mode}.md` |
| **Agents-as-References** | Coordinator with multi-step workflow — pass file path to subagent, subagent reads itself. 0 tokens in coordinator context |
| **Agents-as-References** | Orchestrator with multi-step workflow -- pass file path to subagent, subagent reads itself. 0 tokens in orchestrator context |
| **Context Fork** | Standalone task, no history needed, &lt;4 phases — `context: fork` + isolated subagent |
| **Executable Bash** | `**EXECUTE** using Bash tool:` keyword + `&& echo "✅" \|\| echo "❌"` + `> STOP if ❌` |
| **Pushy Description** | LLM-invocable skills — action verb + `Triggers: "exact user phrases"`. Raises rate 20% → 50-72% |
@@ -197,7 +197,7 @@ Root-level `SKILL.md`: a plugin with a `SKILL.md` at the plugin root (no `skills
### Subagent spawn constraint
CC 2.1.172 allows nested subagent spawning up to 5 levels deep. However, brewcode workflow requires spawning only from the main conversation — the 2-step report protocol (`post-task.mjs`) binds the task lock to one session and delivers report/coordinator instructions back to the spawning conversation. Nested spawns bypass session binding, `KNOWLEDGE` injection, and the coordinator loop. This is a **brewcode workflow policy**, not a Claude Code platform limitation.
CC allows nested subagent spawning, but prefer spawning from the main conversation so the orchestrator can integrate each agent's result directly. Brewcode agents are stateless and resolve plugin paths via `${CLAUDE_PLUGIN_ROOT}`.
</details>
web/docs/src/content/docs/brewcode/agents/tester.mdx
+1 -1
View File
@@ -144,7 +144,7 @@ Tester fixes test code and configuration. Any failure that requires changing pro
Agent definition, role boundaries, and output format specification.
</Card>
<Card title="Brewcode overview" icon="rocket" href="/brewcode/">
All brewcode agents and skills — infinite task execution, quorum reviews, session handoff.
All brewcode agents and skills -- quorum reviews and multi-agent workflows.
</Card>
</CardGrid>
web/docs/src/content/docs/brewcode/overview.mdx
+20 -26
View File
@@ -14,24 +14,24 @@ Install the entire suite:
<InstallPrompt plugin="all" mode="install" />
Brewcode is a plugin for Claude Code that turns single sessions into an **infinite task pipeline**
with automatic context handoff between sessions. The name is a nod to brewing: skills (recipes) define
what to brew, agents (brewers) do the work, and hooks (processes) manage the lifecycle.
Brewcode is a plugin for Claude Code for spec authoring, semantic code search, and a toolkit of
skills and agents. The name is a nod to brewing: skills (recipes) define what to brew, agents
(brewers) do the work, and hooks (processes) manage prompt-time injection.
## Key Features
<CardGrid>
<Card title="Infinite context" icon="infinity">
Automatic knowledge transfer between sessions via KNOWLEDGE.jsonl and the handoff mechanism.
Tasks continue after compaction without losing progress.
<Card title="Native plugin paths" icon="infinity">
Agents resolve paths via native `${CLAUDE_PLUGIN_ROOT}`, substituted in each agent .md at Task spawn.
No hook injection, no shared runtime state.
</Card>
<Card title="9 skills (recipes)" icon="terminal">
Full cycle: from project analysis (<a href="/brewcode/skills/setup/">/brewcode:setup</a>) to task execution (<a href="/brewcode/skills/start/">/brewcode:start</a>)
and code review (<a href="/brewcode/skills/standards-review/">/brewcode:standards-review</a>).
From specification (<a href="/brewcode/skills/spec/">/brewcode:spec</a>) to semantic search (<a href="/brewcode/skills/grepai/">/brewcode:grepai</a>)
and multi-agent review (<a href="/brewcode/skills/superreview/">/brewcode:superreview</a>).
</Card>
<Card title="12+ agents (brewers)" icon="users">
<Card title="10 agents (brewers)" icon="users">
Specialized agents: [developer](/brewcode/agents/developer/), [tester](/brewcode/agents/tester/), [reviewer](/brewcode/agents/reviewer/), [architect](/brewcode/agents/architect/), [skill-creator](/brewcode/agents/skill-creator/),
[agent-creator](/brewcode/agents/agent-creator/), and coordinators.
[agent-creator](/brewcode/agents/agent-creator/), [hook-creator](/brewcode/agents/hook-creator/), and bash-expert.
</Card>
<Card title="2 hooks (processes)" icon="settings">
Prompt-time skill-activation reminder and session-start version-check.
@@ -62,8 +62,8 @@ all context is managed through built-in Claude Code events.
│ │
│ Skills (recipes) Agents (brewers) │
│ ┌──────────────┐ ┌──────────────────┐ │
│ │ setup, spec │ │ developer, tester │ │
│ │ plan, start │ ──────> │ reviewer, architect│ │
│ │ spec, grepai │ │ developer, tester │ │
│ │ superreview │ ──────> │ reviewer, architect│ │
│ │ convention,. │ │ + dynamic teams │ │
│ └──────────────┘ └──────────────────┘ │
│ │ │ │
@@ -100,20 +100,14 @@ brewcode/
│ ├── agent-creator.md # Agent creation (opus)
│ ├── hook-creator.md # Hook creation (opus)
│ ├── bash-expert.md # Bash scripts (opus)
│ ├── bc-coordinator.md # Task coordinator (haiku)
│ ├── bc-knowledge-manager.md # Knowledge manager (haiku)
│ ├── bc-grepai-configurator.md # grepai configurator (opus)
│ └── bc-rules-organizer.md # Rules organizer (sonnet)
├── skills/ # 9 skills
│ ├── setup/ # Project analysis, templates
│ ├── spec/ # Specification creation
│ ├── plan/ # Plan creation
│ ├── start/ # Task execution
│ ├── grepai/ # Semantic search (self-install hooks)
│ ├── convention/ # Convention analysis
│ ├── rules/ # Rule extraction
│ ├── grepai/ # Semantic search
│ ├── teardown/ # Cleanup
│ ├── standards-review/ # Standards review
│ ├── superreview/ # Multi-agent quorum review
│ ├── skills/ # Skill management
│ ├── agents/ # Agent management
│ ├── teams/ # Dynamic agent teams
@@ -126,19 +120,19 @@ brewcode/
<Tabs>
<TabItem label="Skills">
9 skills (recipes) cover the full task lifecycle.
9 skills (recipes) cover spec, search, review, and meta-tooling.
See [Skills](/brewcode/skills/) for details.
| Group | Skills | Purpose |
|-------|--------|---------|
| Core cycle | setup, spec, plan, start | Task creation and execution |
| Quality | standards-review, convention, rules | Code analysis and review |
| Spec | spec | Specification authoring via research + interaction |
| Quality | superreview, convention, rules | Multi-agent review, conventions, rule extraction |
| Dynamic | teams | Agent team creation and management |
| Testing | e2e | E2E testing orchestration with BDD scenarios |
| Utilities | grepai, teardown, skills, agents | Tools and configuration |
| Meta | grepai, skills, agents | Semantic search + skill/agent management |
</TabItem>
<TabItem label="Agents">
12+ agents (brewers) with different models and specializations.
10 agents (brewers) with different models and specializations.
See the [Agents](/brewcode/agents/) section for details.
| Role | Agents | Model |
@@ -147,7 +141,7 @@ brewcode/
| Review | reviewer | opus |
| Creation | skill-creator, agent-creator, hook-creator | opus |
| Optimization | bash-expert | opus |
| Coordination (internal) | bc-coordinator, bc-knowledge-manager, bc-grepai-configurator, bc-rules-organizer | haiku/sonnet/opus |
| Internal | bc-grepai-configurator, bc-rules-organizer | opus/sonnet |
+ dynamic project agents created by [`/brewcode:teams`](/brewcode/skills/teams/)
</TabItem>
web/docs/src/content/docs/brewcode/skills/convention.mdx
+1 -1
View File
@@ -161,7 +161,7 @@ import { Badge, Callout, Card, CardGrid, Steps, UpdateNotice } from '../../../..
<CardGrid>
<Card title="Rules skill" icon="document" href="/brewcode/skills/rules/">
Turn KNOWLEDGE.jsonl and analysis outputs into structured `.claude/rules/*.md` files.
Turn analysis outputs into structured `.claude/rules/*.md` files.
</Card>
<Card title="GitHub source" icon="link" href="https://github.com/kochetkov-ma/claude-brewcode/tree/main/brewcode/skills/convention">
Source code, scripts, layer definitions, and convention doc templates.
web/docs/src/content/docs/brewcode/skills/grepai.mdx
+7 -3
View File
@@ -10,7 +10,11 @@ import { Card, CardGrid, Badge, Callout, Steps, UpdateNotice } from '../../../..
<Badge variant="primary" text="sonnet" /> <Badge variant="secondary" text="session" />
<Callout type="caution" title="Prerequisites">
Run [/brewcode:setup](/brewcode/skills/setup/) first — it installs Homebrew, Ollama, and grepai CLI automatically in Phase 0.
Run `/brewcode:grepai setup` — it checks/installs Homebrew, Ollama, and the grepai CLI, builds the index, and self-installs the grepai hooks into your project.
</Callout>
<Callout type="note" title="Self-install hooks">
`setup` mode detects whether grepai hooks are present in your project. If not, it copies `grepai-session.mjs` (SessionStart auto-start/health) and `grepai-reminder.mjs` (PreToolUse:Bash reminder) from the plugin's `skills/grepai/assets/` into `.claude/grepai/hooks/`, then merges the SessionStart and PreToolUse:Bash entries into `.claude/settings.json` using jq (python3 fallback, no-clobber, idempotent). A summary of created files is printed when done.
</Callout>
<Callout type="tip" title="Environment">
@@ -133,8 +137,8 @@ The skill always prints a status table after any mode completes:
</details>
<CardGrid>
<Card title="setup skill" icon="document" href="/brewcode/skills/setup/">
Run setup first — it installs all prerequisites including Ollama and grepai CLI.
<Card title="Hooks" icon="document" href="/brewcode/hooks/">
Auto-start and reminder hooks that setup mode installs into your project.
</Card>
<Card title="Source on GitHub" icon="link" href="https://github.com/kochetkov-ma/claude-brewcode/tree/main/brewcode/skills/grepai">
Skill scripts, SKILL.md, and configuration templates.
web/docs/src/content/docs/brewcode/skills/review.mdx
+8 -8
View File
@@ -1,6 +1,6 @@
---
title: "Review"
description: "Dynamic per-project quorum code review skill — generated by /brewcode:setup, adapted to your tech stack, agents, and conventions."
description: "Dynamic per-project quorum code review skill — generated by /brewcode:superreview, adapted to your tech stack, agents, and conventions."
order: 18
---
import { Badge, Callout, Card, CardGrid, Steps, UpdateNotice } from '../../../../components/mdx';
@@ -8,11 +8,11 @@ import { Badge, Callout, Card, CardGrid, Steps, UpdateNotice } from '../../../..
# review — quorum code review
<Callout type="caution">
**`/brewcode:review` is not a static plugin skill.** It is generated into each target project by [`/brewcode:setup`](/brewcode/skills/setup/). The generated skill knows your tech stack, your project agents, your coding conventions, and your etalon classes — every project gets its own version.
**`/brewcode:review` is not a static plugin skill.** It is generated into each target project by [`/brewcode:superreview`](/brewcode/skills/superreview/). The generated skill knows your tech stack, your project agents, your coding conventions, and your etalon classes — every project gets its own version.
</Callout>
<Callout type="tip">
**Not the same as `/brewcode:standards-review`.** Standards-review is a static plugin skill for rule/style/convention compliance checks. Review is a deliberative multi-agent pipeline with quorum consensus — it checks if the code is actually good, not just if it follows the rules.
**`/brewcode:superreview` merges quorum review and standards checking into one generated deep-review skill.** There is no separate static skill for rule/style compliance — superreview produces a single project-local skill that covers both convention enforcement and deliberative quorum consensus. It checks if the code is actually good, not just if it follows the rules.
</Callout>
## Quick reference
@@ -115,7 +115,7 @@ import { Badge, Callout, Card, CardGrid, Steps, UpdateNotice } from '../../../..
**Merge rules** within a confirmed cluster: longest description wins, highest severity, first non-null suggestion, averaged confidence, min/max line range, full list of contributing agents.
**How the skill is generated** — run [`/brewcode:setup`](/brewcode/skills/setup/). Phase 3.5 runs `bash scripts/setup.sh review`, copies `brewcode/templates/skills/review/SKILL.md.template` into `.claude/skills/brewcode-review/SKILL.md`, and fills 8 placeholders from parallel analysis agents:
**How the skill is generated** — run [`/brewcode:superreview`](/brewcode/skills/superreview/). It dispatches parallel analysis agents that inspect your repo topology, stack, agents, and coding conventions, then emits a project-tailored `SKILL.md` with all 8 placeholders filled:
| Placeholder | Filled with |
|-------------|-------------|
@@ -142,11 +142,11 @@ The skill is committable — check it into version control so your whole team sh
</details>
<CardGrid>
<Card title="/brewcode:standards-review" icon="check" href="/brewcode/skills/standards-review/">
Static plugin skill for rule and convention compliance. Complementary — run both for full coverage.
<Card title="/brewcode:superreview" icon="star" href="/brewcode/skills/superreview/">
Generated per project by /brewcode:superreview — merges review + standards into one deep-review skill.
</Card>
<Card title="GitHub template source" icon="link" href="https://github.com/kochetkov-ma/claude-brewcode/tree/main/brewcode/templates/skills/review">
The 460-line template instantiated per project by <code>/brewcode:setup</code>.
<Card title="GitHub skill source" icon="link" href="https://github.com/kochetkov-ma/claude-brewcode/tree/main/brewcode/skills">
Source for the brewcode skill suite.
</Card>
<Card title="brewcode overview" icon="rocket" href="/brewcode/">
All brewcode skills and agents in one place.
web/docs/src/content/docs/brewcode/skills/spec.mdx
+6 -8
View File
@@ -8,7 +8,7 @@ import { Badge, Callout, Card, CardGrid, Steps, UpdateNotice } from '../../../..
# spec — research and write spec
<Callout type="caution">
**Skipping spec means rewriting later.** Without a specification, parallel agents in `/brewcode:plan` and `/brewcode:start` have no shared ground truth — they duplicate work, contradict each other, and miss edge cases. The fix costs more than the spec would have.
**Skipping spec means rewriting later.** Without a specification, parallel agents and reviewers have no shared ground truth — they duplicate work, contradict each other, and miss edge cases. The fix costs more than the spec would have.
</Callout>
<Callout type="tip">
@@ -55,8 +55,6 @@ Files Created:
SPEC: .claude/tasks/20260417_143052_auth_feature_task/SPEC.md
Task Dir: .claude/tasks/20260417_143052_auth_feature_task/
Next Step:
/brewcode:plan .claude/tasks/20260417_143052_auth_feature_task/
```
## Flow
@@ -65,7 +63,7 @@ Next Step:
<li>
<div>
<strong>Check adapted templates</strong>
<p>Verifies <code>.claude/tasks/templates/SPEC.md.template</code> exists. Missing template means <code>/brewcode:setup</code> has not run — the skill stops here rather than writing a spec into the wrong structure.</p>
<p>Verifies <code>.claude/tasks/templates/SPEC.md.template</code> exists. Missing template means the project structure is not initialized — the skill stops here rather than writing a spec into the wrong structure.</p>
</div>
</li>
<li>
@@ -155,11 +153,11 @@ MAX 3 iterations — then surface remaining remarks via AskUserQuestion
</details>
<CardGrid>
<Card title="Plan" icon="document" href="/brewcode/skills/plan/">
Next step — turn the SPEC into a phased execution plan with task breakdown.
<Card title="Semantic Search" icon="document" href="/brewcode/skills/grepai/">
Search the codebase semantically to ground the spec in real code patterns and dependencies.
</Card>
<Card title="Start" icon="play" href="/brewcode/skills/start/">
Execute the plan with infinite handoff and automatic session continuity.
<Card title="All Brewcode Skills" icon="star" href="/brewcode/skills/">
Browse every skill available in the brewcode plugin.
</Card>
<Card title="GitHub source" icon="link" href="https://github.com/kochetkov-ma/claude-brewcode/tree/main/brewcode/skills/spec">
SKILL.md, references, and SPEC-creation.md consolidation rules.
web/docs/src/content/docs/brewcode/skills/teams.mdx
+1 -1
View File
@@ -144,7 +144,7 @@ Every generated agent contains a self-selection protocol. When a task arrives, t
**Integration with core skills**
Team agents are preferred by `/brewcode:spec`, `/brewcode:plan`, and `/brewcode:start` when matching domains are found. Priority: team agent → project agent → plugin agent.
Team agents are preferred by `/brewcode:spec` when matching domains are found. Priority: team agent → project agent → plugin agent.
</details>
web/docs/src/content/docs/brewcode/workflow.mdx
-381
View File
@@ -1,381 +0,0 @@
---
title: "Workflow"
description: "Detailed walkthrough of the full Brewcode cycle: spec, plan, start, handoff, knowledge"
order: 14
---
import { Card, CardGrid, Callout, Tabs, TabItem, Steps, Badge } from '../../../components/mdx';
# Workflow
This section is a deep dive into Brewcode mechanics. Unlike the [Quick Start](/quickstart/),
here every stage is covered: from specification creation to automatic context handoff
between sessions and artifact structure.
## High-level overview
```
/brewcode:spec /brewcode:plan /brewcode:start
| | |
Research Phase generation Execution loop
(5-10 agents) + quorum review + handoff
| | |
v v v
SPEC.md PLAN.md artifacts/
phases/ KNOWLEDGE.jsonl
KNOWLEDGE.jsonl FINAL.md
```
## Phase 1: Specification (/brewcode:spec)
The [spec](/brewcode/skills/spec/) skill creates a structured task specification through parallel codebase research
and user dialog.
<Steps>
<li class="step step-primary">
<div>
<strong>Template check</strong>
<p>The skill checks for <code>SPEC.md.template</code> in <code>.claude/tasks/templates/</code>.
If the template is missing, you need to run <code>/brewcode:setup</code> first.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Input parsing</strong>
<p>Determines the input type: text description, path to a requirements file, or reference
from <code>.claude/TASK.md</code>. Supports the <code>-n</code> (--noask) flag for autonomous mode.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Clarifying questions</strong>
<p>3-5 questions via AskUserQuestion across categories: Scope (what's in/out),
Constraints (libraries, compatibility), Edge cases (concurrency, null inputs).
Skipped in <code>-n</code> mode.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Split check</strong>
<p>If requirements cover more than 3 independent areas or are estimated at 12+ phases --
a split is suggested. If accepted, only the first part gets a SPEC.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Parallel research</strong>
<p>5-10 agents are launched <strong>in a single message</strong> (in parallel).
Each agent analyzes its area: Controllers, Services, DB/Repos, Tests, Config, Docs.</p>
<pre><code class="language-text">Task(subagent_type="developer", prompt="Analyze services...")<br/>Task(subagent_type="tester", prompt="Analyze test patterns...")<br/>Task(subagent_type="reviewer", prompt="Analyze quality...")<br/>Task(subagent_type="Explore", prompt="Find library docs...")</code></pre>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Consolidation into SPEC.md</strong>
<p>Agent results are merged with deduplication. A task directory
<code>.claude/tasks/&#123;TS&#125;_&#123;NAME&#125;_task/</code> and SPEC.md file are created from the template.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Specification review</strong>
<p>Reviewer checks: completeness, consistency, feasibility, risks.
Fix loop: continues while critical/major issues exist (max 3 iterations).</p>
</div>
</li>
</Steps>
## Phase 2: Planning (/brewcode:plan)
The [plan](/brewcode/skills/plan/) skill transforms a specification into a detailed plan with phases, dependencies, and completion criteria.
### Input sources
| Input | Action |
|-------|--------|
| Path to `{TS}_{NAME}_task/` | Reads SPEC.md from the directory |
| Path to `SPEC.md` | Derives task dir from parent folder |
| `.claude/plans/LATEST.md` | Plan Mode: parses the plan, creates task dir, skips SPEC |
| Empty | Checks `.claude/TASK.md` for the latest task |
### Phase generation
Each task is split into 5-12 phases. For every execution phase, a verification phase is created,
and a Final Review is added at the end.
```
phases/
├── 1-create-entity.md # Execution: developer
├── 1V-verify-entity.md # Verification: reviewer
├── 2-implement-service.md # Execution: developer
├── 2V-verify-service.md # Verification: tester
├── 3-add-api-endpoint.md # Execution: developer
├── 3V-verify-endpoint.md # Verification: reviewer
└── FR-final-review.md # Final Review: reviewer+tester+architect
```
### Phase Registry in PLAN.md
PLAN.md uses a "slim" v3 format -- only the Phase Registry table and metadata.
Phase details are in separate `phases/*.md` files.
### Plan quorum review
3 agents review the plan in parallel. The 2/3 rule: an issue is accepted only if confirmed
by 2+ agents.
| Agent | Review focus |
|-------|-------------|
| Plan | Phase Registry, completeness, phase quality |
| [architect](/brewcode/agents/architect/) | Architectural decisions, dependencies, technologies |
| [reviewer](/brewcode/agents/reviewer/) | Completion criteria, granularity, risks |
After the quorum -- requirements tracing: every SPEC item must have a phase in the plan.
## Phase 3: Execution (/brewcode:start)
The [start](/brewcode/skills/start/) skill is the heart of Brewcode. It executes tasks with infinite context via Task API,
the 2-step protocol, and automatic handoff during compaction.
### Initialization
```
/brewcode:start -> Load PLAN.md -> Parse Phase Registry
|
v
TaskCreate for each phase -> TaskUpdate dependencies
|
v
Execution Loop: TaskList() -> pending+unblocked -> spawn agents
```
1. **Resolve path** -- from arguments or `.claude/TASK.md`
2. **Initialize** -- bc-coordinator creates the lock, sets status to `in progress`
3. **Load context** -- reads PLAN.md and KNOWLEDGE.jsonl (NOT phases/ -- those are for agents)
4. **Create tasks** -- TaskCreate for each Phase Registry row
5. **Set dependencies** -- TaskUpdate with addBlockedBy
### Execution loop
```
LOOP while pending tasks remain:
1. TaskList() -> find pending + unblocked
2. Same group -> launch in PARALLEL (one message)
3. For each:
a. TaskUpdate(status="in_progress")
b. Task(agent) -> execution
c. WRITE report -> artifacts/{P}-{N}{T}/{AGENT}_output.md
d. CALL bc-coordinator (knowledge extraction)
e. TaskUpdate(status="completed")
4. On verification failure -> fix loop (max 3 iterations)
5. Deadlock check
```
### 2-step protocol
After every agent, two steps must be performed:
<Steps>
<li class="step step-primary">
<div>
<strong>WRITE report</strong>
<p>The manager writes the agent report to <code>artifacts/&#123;P&#125;-&#123;N&#125;&#123;T&#125;/&#123;AGENT&#125;_output.md</code></p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>CALL bc-coordinator</strong>
<p>The coordinator reads the report from disk, extracts 3-10 knowledge entries into KNOWLEDGE.jsonl,
and updates the Phase Status table.</p>
</div>
</li>
</Steps>
### Failure escalation
| After | Action |
|-------|--------|
| 1st failure | R&D task: root cause investigation |
| 2nd failure | Split the phase into sub-phases |
| 3rd failure | Model switch, reassignment, AskUserQuestion |
When escalation is exhausted -- cascading failure: dependent tasks are marked failed,
independent ones continue execution.
## Handoff mechanism (context transfer)
Handoff is Brewcode's key mechanism that provides "infinite context". When Claude Code
approaches the context window limit, automatic compaction kicks in.
### What happens during compaction
<Steps>
<li class="step step-primary">
<div>
<strong>PreCompact hook fires</strong>
<p>The <code>pre-compact.mjs</code> hook detects the approaching limit.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>KNOWLEDGE.jsonl compaction</strong>
<p>Deduplication, stale entry removal, trimming to maxEntries.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Handoff entry written</strong>
<p>A special entry in KNOWLEDGE.jsonl captures the current state: which phase was executing,
what was done, what remains.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Status update</strong>
<p>Task status is set to <code>handoff</code>.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Claude Code compresses context</strong>
<p>Auto-compact summarizes the context. Session ID <strong>does not change</strong>.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Continuation after compact</strong>
<p>The <code>/brewcode:start</code> skill calls <code>TaskList()</code>,
reads PLAN.md, and continues from the current phase. The lock file remains valid.</p>
</div>
</li>
</Steps>
<Callout type="tip" title="Compaction buffer">
The 33K token (16.5%) buffer is hardcoded: 20K (output cap) + 13K (safety).
The <code>CLAUDE_AUTOCOMPACT_PCT_OVERRIDE</code> variable can only shift the trigger threshold
earlier (values below 93%), but cannot increase the buffer.
</Callout>
## KNOWLEDGE.jsonl
KNOWLEDGE.jsonl is the knowledge accumulation file that is transferred between sessions
and injected into agent prompts.
### Entry format
```json
{"ts":"2026-01-26T14:00:00","t":"❌","txt":"Avoid SELECT * in JOOQ queries","src":"reviewer"}
{"ts":"2026-01-26T14:05:00","t":"✅","txt":"Use .fetchInto(Record.class) for type-safe queries","src":"developer"}
{"ts":"2026-01-26T14:10:00","t":"ℹ️","txt":"Project uses PostgreSQL 15 with JOOQ 3.19","src":"architect"}
```
### Entry types
| Type | Priority | Purpose |
|------|----------|---------|
| ❌ | Highest | Anti-patterns, mistakes, things to avoid |
| ✅ | Medium | Best practices, proven approaches |
| ℹ️ | Lowest | Facts, architectural decisions, context |
### Knowledge lifecycle
```
Agent completes work
|
v
bc-coordinator reads report -> extracts 3-10 entries
|
v
KNOWLEDGE.jsonl accumulates entries
|
v
pre-compact hook: deduplicate + trim to maxEntries
|
v
pre-task hook: compress to maxTokens -> inject as ## K block
|
v
/brewcode:rules -> sync to .claude/rules/ (avoid.md, best-practice.md)
|
v
bc-knowledge-manager: prune-rules (remove synced entries)
```
<Callout type="note" title="What is NOT recorded in KNOWLEDGE">
Only genuinely useful, reusable findings go into knowledge.
Phase notes, progress updates, obvious facts are not recorded.
Only anti-patterns (❌), best practices (✅), and architectural facts (ℹ️).
</Callout>
### Injection into agents
The `pre-task.mjs` hook compresses KNOWLEDGE.jsonl to `maxTokens` (default 500)
and adds a `## K` block to the agent prompt:
```
## K
❌ Avoid SELECT * in JOOQ queries
✅ Use .fetchInto(Record.class)
ℹ️ PostgreSQL 15 + JOOQ 3.19
```
System agents (Explore, Plan, general-purpose) do not receive the `## K` block.
## Artifact structure
```
.claude/tasks/{TS}_{NAME}_task/
├── PLAN.md # Phase Registry + metadata
├── SPEC.md # Task specification
├── KNOWLEDGE.jsonl # Accumulated knowledge
├── phases/ # Phase files (for agents)
│ ├── 1-create-entity.md
│ ├── 1V-verify-entity.md
│ ├── 1F1-fix-entity.md # Fix phase (created dynamically)
│ └── FR-final-review.md
├── artifacts/ # Agent reports
│ ├── FINAL.md # Final report
│ ├── 1-1e/ # Phase 1, Execution, iter 1
│ │ └── developer_output.md
│ ├── 1-1v/ # Phase 1, Verification, iter 1
│ │ └── reviewer_output.md
│ └── FR-1e/ # Final Review
│ └── reviewer_output.md
├── backup/ # Backups
└── .lock # Session lock file
```
### Artifact naming
| Pattern | Meaning |
|---------|---------|
| `{P}-{N}{T}` | Phase - Iteration + Type |
| `1-1e` | Phase 1, iteration 1, execution |
| `1-1v` | Phase 1, iteration 1, verification |
| `1-2e` | Phase 1, iteration 2, execution (after fix) |
| `FR-1e` | Final Review, iteration 1, execution |
## Finalization
After all phases are complete:
1. **bc-coordinator** generates `FINAL.md` with consolidated results
2. Task status -> `finished`
3. [`/brewcode:rules`](/brewcode/skills/rules/) extracts knowledge into `.claude/rules/`
4. `bc-knowledge-manager` removes synced entries from KNOWLEDGE.jsonl
5. Lock file is deleted on the next Stop
<Callout type="tip" title="After completion">
Use [/brewcode:convention](/brewcode/skills/convention/) to extract reference classes and patterns
from the written code. Results go into <code>.claude/convention/</code>
and help future tasks follow established conventions.
</Callout>
<CardGrid>
<Card title="Latest Release" icon="rocket" href="https://github.com/kochetkov-ma/claude-brewcode/releases/latest">
Download, changelog, and installation instructions.
</Card>
<Card title="View on GitHub" icon="link" href="https://github.com/kochetkov-ma/claude-brewcode/tree/main/brewcode/">
Source code, README, and configuration files.
</Card>
</CardGrid>
web/docs/src/content/docs/getting-started.mdx
+13 -13
View File
@@ -8,17 +8,17 @@ import { Card, CardGrid, Callout, InstallPrompt, Spoiler } from '../../component
# Introduction
**Brewcode** is a plugin suite for Claude Code that turns regular sessions into **infinite workflows**. A normal Claude Code session loses context on compaction — Brewcode keeps going. It saves knowledge, passes state across compaction cycles, and continues the task without a manual restart.
**Brewcode** is a plugin suite for Claude Code: spec authoring, semantic code search, multi-agent quorum review, and tooling to build your own skills, agents, and hooks. Recipes (skills) define what to brew, brewers (agents) do the work, and hooks manage prompt-time injection.
<Callout type="tip" title="One cycle — full task">
`spec` → `plan` → `start` — and the work runs to completion across however many compaction cycles it takes. You don't babysit it.
<Callout type="tip" title="Spec first, then build">
`spec` → `grepai` → `superreview` — author the spec, index the code for semantic search, then generate a project-tailored deep-review skill.
</Callout>
## Four plugins, one suite
<CardGrid>
<Card title="brewcode" icon="infinity" href="/brewcode/overview/">
The core: infinite tasks, codebase exploration, quorum review, skill/agent/hook creation, prompt optimization. 9 skills · 10 agents · 9 hooks.
The core: codebase exploration, quorum review, spec authoring, skill/agent/hook creation. 9 skills · 10 agents · 2 hooks.
</Card>
<Card title="brewtools" icon="settings" href="/brewtools/overview/">
Universal utilities: prompt optimization, AI artifact removal, secrets scanning, SSH, deploy. 10 skills · 3 agents.
@@ -34,17 +34,17 @@ import { Card, CardGrid, Callout, InstallPrompt, Spoiler } from '../../component
## What makes it different
<CardGrid>
<Card title="Infinite sessions" icon="infinity" href="/brewcode/overview/">
Automatic context handoff at ~90% capacity. The same session continues working
without manual restart. KNOWLEDGE.jsonl survives any number of compaction cycles.
<Card title="Spec authoring" icon="document" href="/brewcode/skills/spec/">
Writes a complete project specification via structured research and user interaction.
Outputs a SPEC.md with requirements, architecture decisions, and acceptance criteria.
</Card>
<Card title="Quorum code review" icon="search" href="/brewcode/skills/standards-review/">
<Card title="Quorum code review" icon="search" href="/brewcode/skills/superreview/">
3 to 5 parallel reviewers work independently. Results are merged via a
quorum algorithm (2/3 agreement). Deduplication and ranking by severity.
</Card>
<Card title="Knowledge accumulation" icon="brain" href="/brewcode/workflow/">
Agents record patterns, errors, and architectural facts in KNOWLEDGE.jsonl.
Knowledge is injected into every agent prompt and converted into project rules.
<Card title="Semantic code search" icon="brain" href="/brewcode/skills/grepai/">
grepai indexes the project with Ollama + bge-m3 embeddings. Self-installs its
hooks into the project and steers agents to grepai_search first.
</Card>
<Card title="Skills and agents" icon="settings" href="/brewcode/skills/">
25 skills and 14 specialized agents across 4 plugins. Built-in tools for
@@ -56,7 +56,7 @@ import { Card, CardGrid, Callout, InstallPrompt, Spoiler } from '../../component
| Plugin | Purpose | Skills | Agents | Hooks |
|--------|---------|--------|--------|-------|
| **brewcode** | Infinite tasks, codebase exploration, review, prompt optimization, skill/agent creation | 9 | 10 | 9 |
| **brewcode** | Spec authoring, codebase exploration, quorum review, skill/agent/hook creation | 9 | 10 | 2 |
| **brewtools** | Text utilities: prompt optimization, AI artifact removal, secrets scanning, SSH, deploy | 10 | 3 | — |
| **brewui** | UI / visual / creative tools (placeholder, empty) | 0 | 0 | — |
| **brewdoc** | Documentation tools: sync, generation, memory optimization, PDF conversion | 6 | 1 | — |
@@ -78,7 +78,7 @@ All four plugins install with a single marketplace and load automatically with e
Set up Brewcode in 2 minutes. One command to add the marketplace, one to install.
</Card>
<Card title="Quick Start" icon="play" href="/quickstart/">
Run your first infinite task in 15 minutes: [setup](/brewcode/skills/setup/), [spec](/brewcode/skills/spec/), [plan](/brewcode/skills/plan/), [start](/brewcode/skills/start/).
Get started in 15 minutes: [spec](/brewcode/skills/spec/), [semantic search](/brewcode/skills/grepai/), [deep review](/brewcode/skills/superreview/).
</Card>
</CardGrid>
web/docs/src/content/docs/installation.mdx
+7 -7
View File
@@ -90,10 +90,10 @@ import { Card, CardGrid, Callout, Steps, Tabs, TabItem, InstallPrompt } from '..
</li>
<li class="step step-primary">
<div>
<strong>Run project setup</strong>
<strong>Author a spec</strong>
<p>Inside a Claude Code session, run:</p>
<pre><a href="/brewcode/skills/setup/"><code>/brewcode:setup</code></a></pre>
<p>Setup automatically checks and installs prerequisites (brew, jq, coreutils) as Phase 0 before analyzing your project. Optionally installs ollama and grepai for semantic search.</p>
<pre><a href="/brewcode/skills/spec/"><code>/brewcode:spec</code></a></pre>
<p>Spec runs parallel codebase research and a refinement loop to produce a SPEC.md. For semantic search, run <a href="/brewcode/skills/grepai/"><code>/brewcode:grepai setup</code></a>, which checks/installs prerequisites (brew, ollama, grepai CLI) and self-installs its hooks.</p>
</div>
</li>
</Steps>
@@ -133,8 +133,8 @@ claude plugin update brewui@claude-brewcode
**`/brewcode:*` commands are not recognized**
-- Restart Claude Code after installation. Plugins only load at session start.
**<a href="/brewcode/skills/setup/">`/brewcode:setup`</a> reports missing `timeout`**
-- The setup skill creates a symlink automatically during Phase 0. If the issue persists: `brew install coreutils`.
**`timeout` command missing on macOS**
-- Brewcode's session-start hook expects `timeout`/`gtimeout`. Install it with `brew install coreutils`.
**`CLAUDECODE` environment variable interferes with launch**
-- Run `unset CLAUDECODE` before running `claude` in scripts (applies to v0.2.47+).
@@ -144,10 +144,10 @@ claude plugin update brewui@claude-brewcode
<CardGrid>
<Card title="Quick Start" icon="rocket" href="/quickstart/">
Run your first infinite task in 15 minutes: [setup](/brewcode/skills/setup/), [spec](/brewcode/skills/spec/), [plan](/brewcode/skills/plan/), [start](/brewcode/skills/start/).
Get started in 15 minutes: [spec](/brewcode/skills/spec/), [semantic search](/brewcode/skills/grepai/), [deep review](/brewcode/skills/superreview/).
</Card>
<Card title="Brewcode Overview" icon="building" href="/brewcode/overview/">
Architecture, concepts, all 25 skills, 14 agents, and 9 hooks.
Architecture, concepts, all 25 skills, 14 agents, and 4 hooks.
</Card>
</CardGrid>
web/docs/src/content/docs/quickstart.mdx
+41 -41
View File
@@ -1,6 +1,6 @@
---
title: "Quick Start"
description: "Your first 15 minutes with Brewcode: setup, spec, plan, start"
description: "Your first 15 minutes with Brewcode: spec, semantic search, deep review"
order: 3
---
@@ -8,8 +8,8 @@ import { Card, CardGrid, Callout, Steps, Badge } from '../../components/mdx';
# Quick Start
In 15 minutes you will go through the full Brewcode cycle: from project analysis to launching an infinite task.
By the end you will have a specification, a plan, and a running task with automatic context handoff.
In 15 minutes you will go through the core Brewcode workflow: from writing a structured specification
to indexing your codebase for semantic search and generating a project-tailored deep review skill.
<Callout type="note" title="Before you begin">
Make sure Claude Code and the brewcode plugin are installed.
@@ -21,68 +21,69 @@ By the end you will have a specification, a plan, and a running task with automa
<Steps>
<li class="step step-primary">
<div>
<strong>Project setup</strong> <Badge variant="primary" text="/brewcode:setup" />
<p>Open Claude Code in your project root and run:</p>
<pre><code>/brewcode:setup</code></pre>
<p><strong>What happens:</strong> the skill analyzes your project structure, tech stack,
test frameworks, and existing agents. It generates tailored templates based on the analysis.</p>
<p><strong>Result:</strong></p>
<ul>
<li><code>.claude/tasks/templates/PLAN.md.template</code> -- plan template</li>
<li><code>.claude/tasks/templates/SPEC.md.template</code> -- specification template</li>
<li><code>.claude/tasks/cfg/brewcode.config.json</code> -- configuration</li>
<li><code>.claude/skills/brewcode-review/</code> -- <a href="/brewcode/skills/review/">tailored code review skill</a></li>
</ul>
<p>This step runs once per project.</p>
<strong>Install brewcode</strong> <Badge variant="primary" text="plugin install" />
<p>Add the plugin from the marketplace and install it into your Claude Code session:</p>
<pre><code>claude plugin marketplace add https://github.com/kochetkov-ma/claude-brewcode
claude plugin install brewcode@claude-brewcode</code></pre>
<p>This step runs once per machine. Reload with <code>/reload-plugins</code> if already inside a session.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Creating a specification</strong> <Badge variant="primary" text="/brewcode:spec" />
<strong>Write a specification</strong> <Badge variant="primary" text="/brewcode:spec" />
<p>Describe the task in plain text or point to a requirements file:</p>
<pre><code>/brewcode:spec "Implement JWT authorization with role-based access"</code></pre>
<p>Or from a file:</p>
<pre><code>/brewcode:spec ./docs/requirements/auth.md</code></pre>
<p><strong>What happens:</strong> the plugin launches 5-10 research agents in parallel.
The agents analyze the codebase, find related code, dependencies, and patterns.
Then it asks clarifying questions (use the <code>-n</code> flag for autonomous mode without questions).</p>
<p><strong>Result:</strong> a <code>SPEC.md</code> file -- a structured specification with project context.</p>
<p><strong>What happens:</strong> the skill launches 5-10 research agents in parallel.
They analyze the codebase, locate related code, dependencies, and patterns,
then ask clarifying questions (use the <code>-n</code> flag for autonomous mode without questions).</p>
<p><strong>Result:</strong> a <code>SPEC.md</code> file — a structured specification with full project context.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Generating a plan</strong> <Badge variant="primary" text="/brewcode:plan" />
<pre><code>/brewcode:plan</code></pre>
<p><strong>What happens:</strong> a phased execution plan is generated from SPEC.md.
Each phase contains specific steps, expected artifacts, and completion criteria.</p>
<p><strong>Result:</strong> a <code>PLAN.md</code> file -- a plan with phases, ready for execution.</p>
<strong>Index for semantic search</strong> <Badge variant="primary" text="/brewcode:grepai setup" />
<pre><code>/brewcode:grepai setup</code></pre>
<p><strong>What happens:</strong> checks and installs Ollama and the CLI if missing, builds a vector
index of your codebase using the <code>bge-m3</code> embedding model, and self-installs grepai hooks
into the project.</p>
<p><strong>Result:</strong> all subsequent Claude Code sessions can query the project semantically —
faster and more precise context retrieval for every agent working on the spec.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Running the task</strong> <Badge variant="primary" text="/brewcode:start" />
<pre><code>/brewcode:start</code></pre>
<p><strong>What happens:</strong> Brewcode begins executing the plan phase by phase.
Agents work on the task, recording knowledge in KNOWLEDGE.jsonl.
When context reaches ~90%, the PreCompact hook fires:
knowledge is compressed, handoff state is written, and the session continues automatically.</p>
<p><strong>Result:</strong> the task runs to completion regardless of how many compaction cycles occur.
All artifacts are saved in `.claude/tasks/<ts>_<name>_task/artifacts/`.</p>
<strong>Generate a deep review skill</strong> <Badge variant="primary" text="/brewcode:superreview" />
<pre><code>/brewcode:superreview</code></pre>
<p><strong>What happens:</strong> analyzes your codebase and conventions, then generates a
project-tailored quorum review skill that merges code review and standards checking
into one reusable skill installed locally at <code>.claude/skills/</code>.</p>
<p><strong>Result:</strong> a custom review skill ready to invoke any time you need a deep,
project-aware code review.</p>
</div>
</li>
<li class="step step-primary">
<div>
<strong>Create a domain agent team</strong> <Badge variant="primary" text="/brewcode:teams — optional" />
<pre><code>/brewcode:teams</code></pre>
<p>Generates 5-20 project-specific agents with self-selection logic and performance tracking.
Skip this step if the default agent set is sufficient for your project.</p>
</div>
</li>
</Steps>
## What's next
<Callout type="tip" title="After the task completes">
The task is done. Here is what you can do with the results:
<Callout type="tip" title="After the workflow">
Your project is set up for iterative development. Here is what you can do next:
**[Code review](/brewcode/skills/standards-review/)** -- run a quorum review of the changes:
**[Deep review](/brewcode/skills/superreview/)** -- run a quorum review of any changes:
```bash
/brewcode:standards-review
/brewcode:superreview
```
**[Extract rules](/brewcode/skills/rules/)** -- convert accumulated knowledge into project rules:
**[Extract rules](/brewcode/skills/rules/)** -- convert knowledge into project-wide coding rules:
```bash
/brewcode:rules
```
@@ -97,8 +98,7 @@ By the end you will have a specification, a plan, and a running task with automa
<CardGrid>
<Card title="brewcode" icon="building" href="/brewcode/overview/">
9 skills for development: infinite tasks, review, semantic search,
skill and agent creation, dynamic teams.
9 skills: spec, semantic search, deep review, conventions, teams, and more.
</Card>
<Card title="brewtools" icon="wrench" href="/brewtools/overview/">
10 skills for text utilities: prompt optimization, AI artifact removal,
web/docs/src/utils/navigation.ts
-1
View File
@@ -54,7 +54,6 @@ export const navigation: NavSection[] = [
],
},
{ title: 'Hooks', slug: 'brewcode/hooks' },
{ title: 'Workflow', slug: 'brewcode/workflow' },
],
},
{