Setup Status

sonnet read-only probes 11 setups

Quick reference

Field	Value
Command	/brewcode:setup-status
Input	Free-text prompt; optional plugin or skill filter
Tools	Read, Bash, Glob, Grep, AskUserQuestion; no Write, Edit, or Agent
Covers	Eleven setups across brewcode, brewtools, and brewdoc
Writes	No setup artifacts; one optional, explicitly accepted task-tools settings key

What it does

Inspect an unfamiliar repository or check copied artifacts after updating the plugins. The skill reports installed, disabled, missing, partial, or stale setups and gives the exact next command. It compares artifact content versions, ownership, copied bytes, and required artifacts or wiring.

The probes are read-only. The skill runs no setup and spawns no agents. Its only write is an optional one-key settings merge after you accept the task-tools offer. There is no --run, --fix, or automatic setup mode.

When to use

Situation	Command
Inspect the whole project	/brewcode:setup-status
Inspect one plugin	/brewcode:setup-status brewtools
Inspect conventions and their loading rule	/brewcode:setup-status convention-setup
Inspect semantic-search wiring	/brewcode:setup-status semble-setup

Filters can occur in a free-text RU/EN prompt. Unrecognized text means a full report, not an error. Report prose follows your language; paths and commands remain literal.

Example

/brewcode:setup-status convention-setup

On a project without conventions, expect a missing row pointing to /brewcode:convention-setup install. On a deliberately disabled installation, expect disabled and an enable command; it will stay out of the run-list. An outdated document or loading-rule content version can produce stale even if the other convention files are current.

Workflow

Run-list order

Stage	Setups	Reason
1	agent-return	Inspect global drift before later subagent work
2	semble	Establish semantic search for repository analysis
3	convention	Extract architecture, coding, and testing patterns before agents consume them
4	teams	Create project agents and shared intent-guard
5	agent-router, manager, agent-deadline, superreview	Configure consumers of agents and intent-guard
6	task-board	Install the task process, methodology, and board
7	memory-sync, then docsync	Capture the completed configuration and track document drift

A stage without actionable rows is omitted. Conventions need no teams prerequisite. See Full Setup for installation examples.

The eleven setups and their anchors

Setup	Anchor
teams-setup	.claude/teams/*/team.md
semble-setup	.claude/rules/semble-first.md
convention-setup	.claude/rules/convention.md
superreview-setup	Emitted .claude/skills/superreview/SKILL.md
task-board-setup	.claude/features/board.md
agent-deadline-setup	Deadline guard, project or global scope
agent-router-setup	.claude/hooks/agent-router.mjs
manager-setup	.claude/brewtools/manager/state.json
agent-return-setup	Return guard, project or global scope
memory-sync-setup	Emitted .claude/skills/memory-sync/SKILL.md
docsync-setup	.claude/docsync/config.json

Recurring tools such as agents, skills, rules, text-optimize, and publish do not have setup installation state and stay out of this roster. Convention-setup installs persistent documents and loading guidance, so it belongs in the roster.

Verdicts

Evaluated in this order:

Verdict	Meaning
n/a	The owning plugin is unavailable
disabled	The setup’s off-switch confirms deliberate disablement
missing	Anchor and secondary artifacts are absent in live and disabled spellings
partial	Incomplete installation, half-applied toggles, unresolved metadata, or wrong ownership
stale	Content-version, copied-byte, absence, or wiring signals need refresh
installed	Required artifacts and applicable checks agree

The report distinguishes changed content, legacy stamps, unstamped artifacts, and byte drift. A missing plugin comparison source is an unknown validation result, never proof of stale content. Shared files from another setup do not prove installation.

Conventions: documents and reversible loading

The convention row checks .claude/rules/convention.md and all three generated documents: reference-patterns.md, testing-conventions.md, and project-architecture.md. Their generated_by must identify convention-setup, and their content versions are checked against its current source. These documents are project-specific; no byte comparison with a template can establish their freshness.

Parking the loader as convention.md.disabled reports disabled. Upgrade refreshes conventions and owned loader metadata while preserving local wording and the parked state. Accepted coding rules and manual CLAUDE.md references remain active when convention loading is disabled.

Off-switches

Setup group	Disabled signal
teams	All team roster members parked as .disabled; shared intent-guard stays
superreview, memory-sync	Generated SKILL.md parked as .disabled
task-board	Every deployed machinery entry parked; task data stays
convention	Loading rule parked; generated documents and accepted rules stay
semble	State enabled is false
manager	State hard is not true
agent-deadline, agent-return	Config enabled is not true; missing key means off
agent-router, docsync	Config enabled is false; missing key means on

Disabled takes precedence over missing, partial, and stale, so a deliberate pause never enters the repair list.

Technical signals

Signal	Question
content_version	Has this artifact’s own tracked content changed?
version	Which plugin release produced this installation?
generated_by	Did the setup that owns this path write it?
Applicable cmp comparison	Was a copied asset refreshed, or has its body drifted?

The content version is compared with the plugin source for that same artifact, not the release-wide plugin version. A release with unchanged artifact content does not make an installation stale. Legacy version-only artifacts use the source’s fallback rules.

The source’s STAMPS table contains 23 carrier entries: 7 brewcode, 13 brewtools, and 3 brewdoc. Conventions contribute four: the loader and three documents. Deadline and return each contribute guard/config pairs for project and global scopes; router contributes hook/config, and manager contributes state plus guard fallback. Counts describe carrier entries, not setup totals. TOTAL and per-plugin SEEN/WANT assertions reject missing entries.

Copied-byte checks cover the assets named by the roster: semantic-search hooks and rule, deadline hooks, router hook, manager guard, docsync hooks, supported memory-sync references, team tracer, and return hook modules. Semble’s wiring is checked independently, including stale settings entries and missing registrations. Its customized root .sembleignore is excluded from byte comparison.

Emitted superreview metadata is read from its live SKILL.md, never from unresolved placeholders in the pristine baseline. A changed baseline may require manually porting template changes. Hand-edited references or search rules can also require a diff review rather than a blind upgrade.

Related