agentleFS
ddtcorexddtcorex/maestro-skills/skills/govard-magento/SKILL.md

govard-magento

Skill5 starsChanged 3 months agoMITView on GitHub

This skill should be used when the user asks to "clear Magento cache", "flush redis cache", "run Magento CLI", "run bin/magento commands", "deploy static content", "setup:di:compile", "reindex catalog", "run indexer commands", "enable/disable modules", "start frontend sync", "run browser-sync", "set up live reload for Hyva/Luma", or "govard frontend". Provides Magento-specific Govard shortcuts and commands. DEPENDENT on govard-toolbox for base commands.

The file

Files
SKILL.md

Govard Magento Commands

Magento-specific shortcuts and commands for Govard environments. Run govard <cmd> --help for the current flag set and govard capabilities for what the installed binary supports.

Related Skills

REQUIRED BACKGROUND: Load govard-toolbox first. This skill only covers Magento-specific shortcuts layered on top of Govard's base environment commands.

Stack commands need Docker running. govard audit run --checks integrity is the container-free exception (see govard-toolbox, "Host Without Docker").

This skill covers only container/CLI shortcuts. For module architecture, DI, and security patterns, see magento2-dev-core and magento2-backend-dev; for code quality and performance checks, see magento2-linter, magento2-security-scan, and magento2-performance-audit.

Code Quality Audit

govard audit run --checks lint is the native, persistent lint gate for this project. Target-mode resolution, the PHP matrix, provider rules, and caching/rerun identity are covered in magento2-linter ("Govard-Native Lint Audit Is the Real Gate"); this skill does not duplicate that policy. To re-check the exact same session (for example after a fix) instead of starting a fresh, non-comparable run:

govard audit rerun --session SESSION_ID

Optionally capture a page-profile artifact in the same run:

govard audit run --checks lint,profiler --url 'https://<domain>/'

The profiler reads the CSV Magento writes to var/log/profiler.csv. If the run fails with "collect Magento profiler CSV ... No such file", Magento never wrote the file for that request, so the capture is unusable; try another uncached HTML URL. Policy details live in magento2-linter.

Gotchas:

  • An audit refuses to run while Xdebug is enabled, to protect timing fidelity. Pass --allow-xdebug to accept the overhead, or turn Xdebug off with govard debug off. (The error text suggests govard config set stack.features.xdebug false, but config set rejects that key.)
  • audit rerun reuses the session's stored checks. Passing a different --checks set than the session recorded can fail with "no persisted ... settings"; start a new audit run instead.

Unit Tests

Magento's project root ships no phpunit.xml, so a bare vendor/bin/phpunit there prints usage and exits 1. That exit means "no configuration found", never "tests failed". The unit-suite configuration lives at dev/tests/unit/phpunit.xml.dist; always pass it explicitly with -c and scope the run with --filter (module or class name) so unrelated core suites never execute:

govard sh -c 'vendor/bin/phpunit -c dev/tests/unit/phpunit.xml.dist --filter "MyModuleBatchLoader|ExampleTest"'

For a direct path run (single module, no filter syntax), pair --no-coverage with the unit bootstrap:

govard sh -c 'vendor/bin/phpunit --no-coverage --bootstrap dev/tests/unit/framework/bootstrap.php app/code/Acme/Label/Test/Unit'

Never run the full suite bare inside review or automated checks: Magento core fixtures can conflict when loaded together and kill the run for reasons unrelated to the change under review. A healthy scoped run reports OK with its test/assertion counts (PHPUnit warnings about duplicate suite files are noise from Magento's own config). Treat a Fatal error naming a vendor/ _files fixture as a pre-existing repo condition, not a regression. When piping output through tail, capture the real status via ${PIPESTATUS[0]} (or set -o pipefail); a trailing echo "EXIT:$?" after a pipe reports tail's status and masks PHPUnit failures as passes.

Magento CLI

govard tool magento <args> runs bin/magento in the PHP container. Run govard tool magento list to see what the installed Magento provides, and do not assume a command exists (for example there is no generate:* namespace and no dev:js:* / dev:css:* namespace in stock Magento).

# Cache management
govard tool magento cache:flush
govard tool magento cache:clean full_page      # takes cache TYPES (see cache:status), not cache tags
govard tool magento cache:enable layout block_html
govard tool magento cache:disable config

# Module management (module:disable fails if enabled modules depend on it)
govard tool magento module:enable Vendor_Module
govard tool magento module:disable Vendor_Module
govard tool magento module:status

# Setup commands
govard tool magento setup:di:compile
govard tool magento setup:static-content:deploy -f
govard tool magento setup:upgrade --keep-generated
govard tool magento setup:db:status

# Deploy mode
govard tool magento deploy:mode:set developer
govard tool magento deploy:mode:show

# Admin user
govard tool magento admin:user:create

Indexer Commands

govard tool magento indexer:status
govard tool magento indexer:reindex                       # all
govard tool magento indexer:reindex catalog_product_price # one (dependent indexers may rebuild too)
govard tool magento indexer:set-mode schedule             # or realtime

Cron Commands

govard tool magento cron:run
govard tool magento cron:run --group=default
govard tool magento cron:install   # writes the container user's crontab
govard tool magento cron:remove

Development Tools

govard tool magento dev:template-hints:enable    # dev only; also :disable / :status
govard tool magento dev:query-log:enable         # also :disable

Frontend Development (BrowserSync / LiveReload)

Requires stack.features.frontend_sync: true in .govard.yml (Magento family only). Without it, govard frontend ... exits with "frontend sync is disabled". govard env up never starts this; it is a separate, explicit, on-demand lifecycle:

govard env up                    # app/db/etc. up, no frontend services yet
govard frontend start            # renders + starts BrowserSync/LiveReload + watchers
govard frontend logs -f          # sync/injector container
govard frontend logs watch-<theme> -f
govard frontend stop             # removes only frontend services, keeps their volumes

Prerequisites (Govard never creates or edits these files):

Let govard frontend start be the discovery oracle: it validates the setup and fails fast with the reason. Only read theme files to fix a specific failure; do not pre-verify by hand.

  • Hyva: exactly one scripts.browser-sync owner under app/design/frontend/<Vendor>/<Theme>/web/tailwind, with a committed package-lock.json. The theme's browser-sync.config.js must read GOVARD_FRONTEND_SYNC_TARGET (and port) from the environment instead of hard-coding a host.
  • Luma: root Gruntfile.js, package.json, package-lock.json (copy Magento's .sample files, then govard tool npm install). No BrowserSync config needed.
  • Hyva and Luma discovery are exclusive: start fails unless exactly one setup is valid.

Switching the active Hyva theme: move the scripts.browser-sync entry (plus its browser-sync.config.js and package-lock.json) from the old theme's web/tailwind/package.json to the new theme, then run govard frontend start again.

SymptomFix
frontend start reports success but the page redirects to a different host or drops the sessionThe theme's browser-sync.config.js proxy settings rewrite the origin or cookie domain; keep both unchanged
Discovery fails ("requires either exactly one Hyva ... or root Gruntfile.js and package.json")Confirm exactly one theme owns scripts.browser-sync, or that the Luma root files exist, and not both
frontend start fails with "container ... health is unhealthy"Read govard frontend logs; the sync container could not serve the project (for example Luma themes not yet compiled)

Database Operations

# Direct MySQL shell
govard db connect

# Run SQL. Table names here have no prefix; if this project uses one (db.table_prefix in
# app/etc/env.php), prepend it, e.g. m2_core_config_data instead of core_config_data
govard db query "SELECT * FROM core_config_data WHERE path LIKE '%template%'"

# Import a local dump with a clean reset (drops and recreates the database)
govard db import --file backup.sql --drop

# Stream a remote dump straight into the local database
govard db import --stream-db -e staging --drop

# Export from remote into the local var/ directory, excluding noise and PII tables
govard db dump -e staging --no-noise --no-pii --local

A failed remote dump can leave a tiny stub file in var/; delete it before retrying.

Configuration

govard tool magento config:show system/smtp/host
govard tool magento config:set web/secure/base_url https://local.test/

# Show a value in every scope it is set (default/website/store) in one call;
# bin/magento config:show only reads a single scope at a time
govard tool magerun config:store:get web/secure/base_url

govard tool magento app:config:status

Do not run app:config:dump on a Govard project casually. It writes the dumped paths into app/etc/config.php and locks them. Afterwards govard config auto fails with "The value you set has already been locked" when it tries to set the same paths (for example Varnish or search settings). Dump only if the project really manages config through files, and use app:config:import to apply it.

Diagnostics

# Environment health check: base URL and cookie domain settings, required
# PHP extensions, MySQL InnoDB, missing files/folders
govard tool magerun sys:check

# System snapshot: Magento version/edition, app mode, cache backend,
# search engine, module count
govard tool magerun sys:info

Multi-Website / Multi-Store Setup

Register additional store domains in .govard.yml under store_domains, then let Govard wire up the vhost/DNS side:

domain: "primary.test"
store_domains:
  brand-b.test:
    code: base
    type: website
govard domain add brand-b.test   # records an extra domain in .govard.yml
govard env up                    # applies it
govard config auto
govard tool magento cache:flush

Store codes are also selectable via URL path (/fr/, /admin/) without a separate domain; reserve store_domains for genuinely separate hostnames/websites.

Redis Cache

govard redis flush   # flush all keys
govard redis cli
govard redis info

govard valkey is the equivalent when the cache service is Valkey.

Varnish (only when stack.features.varnish: true)

govard varnish ban '/.*'   # purge everything (quote the pattern)
govard varnish stats
govard varnish ps

Without the feature, these fail with "no such service: varnish" or "Varnish container ... is unknown". For tag-based invalidation use Magento's own cache flow (cache:clean full_page / cache:flush).

Logging

For agents, prefer bounded tail reads (no -f follow, it never returns):

govard sh -c "tail -n 50 var/log/system.log"
govard sh -c "tail -n 50 var/log/exception.log"
govard sh -c "tail -n 50 var/log/debug.log"

Common Issues & Solutions

SymptomFix
"There are no commands defined" after pulling codegovard tool magento setup:di:compile
Static assets not updatinggovard tool magento setup:static-content:deploy -f + cache:flush, then hard-refresh the browser
Database connection refusedgovard ps (is the DB container up?), govard logs db, then govard down && govard up if needed
Container won't startgovard doctor, then govard logs
Storefront 5xx and exception.log shows NoNodesAvailableExceptionThe search service is still starting or down; govard ps, then govard elasticsearch _cluster/health (or govard opensearch ...)
govard config auto fails with "value ... already been locked"app/etc/config.php locks that path (see app:config:dump above); remove it from the file or use Magento's --lock-env flow
Xdebug not connectinggovard debug on, confirm the IDE is listening on port 9003; see govard-toolbox for the IDE setup

Template-only changes do not need setup:di:compile, only setup:static-content:deploy.

Common Workflows

After Pulling Code

govard tool magento setup:upgrade --keep-generated
govard tool magento setup:static-content:deploy -f
govard tool magento cache:flush

After Database Sync

govard config auto   # re-applies local DB, search, Redis, Varnish, base URLs and developer mode
govard tool magento cache:flush

Production Deployment Prep

govard tool magento maintenance:enable
govard tool magento setup:upgrade
govard tool magento setup:di:compile
govard tool magento setup:static-content:deploy -f --theme=Vendor/Theme
govard tool magento cache:flush
govard tool magento maintenance:disable

Govard Deploy (conditional migrate)

govard deploy <remote> probes setup:db:status before the downtime block: exit 0 skips maintenance/workers/config-import/migrate, 1-2 runs them. govard deploy plan <remote> prints the resolved task list without executing anything (it needs a configured branch or --branch; the synthetic sandbox remote resolves the current branch by itself). See govard-toolbox (Deployment) for the full gate, verdict, and resume semantics.

Rehearsing it against govard sandbox (confirmed live on a real Magento project):

  • Seed needs an installed origin. sandbox up dumps the running origin database and rewrites core_config_data base URLs. An origin with an empty database (no bin/magento setup:install yet) fails the seed with "Table ... core_config_data doesn't exist". Install the origin first.
  • Database profile. --profile full follows stack.db_version. A MariaDB series that ships no mysqladmin binary (the newest series) never answers the readiness probe, so up fails with exit 127 on mysqladmin ping. Pass --db mariadb:<older series> for the rehearsal; a database volume written by another series is refused until sandbox down --purge.
  • Track source only. The sandbox mirrors the git repository, so ignore pub/static/*, generated/*, var/*, pub/media/* and app/etc/env.php. A checkout that tracks built pub/static (dangling symlinks into the origin path) makes build:assets fail with "cannot be copied" warnings.
  • Shared files. app/etc/env.php and pub/media are shared and come from the seed, not from git. sandbox reset removes them: afterwards build:assets fails with "The default website isn't defined" until sandbox up --reseed restores env.php.
  • Migrate step. With an up-to-date schema the maintenance, worker-pause, config-import and migrate tasks are reported as skipped ("db up-to-date"). A pending schema change (for example a new module with db_schema.xml) runs maintenance, setup:upgrade (it completed on the sandbox without a search service), then maintenance off before the symlink swap. A failure after maintenance is enabled keeps the lock, the release directory and the maintenance flag: the live site answers 503 until deploy --resume completes or deploy rollback is run. deploy unlock alone does not clear maintenance, and refuses a young lock without --force.
  • --db-backup. The task takes a plain mariadb-dump/mysqldump of the database named in app/etc/env.php, with the credentials in a private option file (stock Magento disables setup:backup, so it is not used). A target with no dump client is refused up front as a configuration error (exit 4). The dump lands in the release's backup area; deploy rollback --with-db restores the dump recorded by the release being replaced and creates var/backups if it is missing. The flag is not remembered across --resume, repeat it.
  • Pulling files. A pull does not follow symlinks that point outside the synced path (rsync --safe-links), so sync -s <remote> --file --path app/etc leaves the local env.php alone when the remote one is a symlink to a shared path. Add --resolve-symlinks only when you want the target's content copied.
  • Doc roots. remote exec <remote> -- <cmd> fails with "cd: ... No such file or directory" until the first release is live on a symlink sandbox. govard sandbox ssh -- <cmd> runs one command without a tty and returns its exit status (a command without -- is a usage error); remote exec still needs the first deploy.
  • Open targets. govard open admin|sftp -e sandbox resolves the synthetic remote and uses the published port.
  • Local audits. audit run refuses while stack.features.xdebug is on unless --allow-xdebug is passed; the lint rows report real findings when core code is tracked in the checkout.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.