MCPInstall

metabase-full-app-to-modular-embedding-upgrade

Migrates a project from Metabase Full App / Interactive (iframe-based) embedding to Modular (web-component-based) embedding. Use when the user wants to replace Metabase iframes with Modular embedding web components.

Installs
358
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/metabase/agent-skills --skill metabase-full-app-to-modular-embedding-upgrade
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads metabase-full-app-to-modular-embedding-upgrade automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

Execution contract

Follow the workflow steps in order — do not skip any step. Create the checklist first, then execute each step and explicitly mark it done with evidence. Each step's output feeds into the next, so skipping steps produces wrong migrations.

If you cannot complete a step due to missing info or tool failure, you must:

  1. record the step as ❌ blocked,
  2. explain exactly what is missing / what failed,
  3. stop (do not proceed to later steps).

Required output structure

Your response should contain these sections in this order:

  1. Step 0: Metabase Version Detection
  2. Step 0.1: Migration Plan Checklist
  3. Step 1: Project Scan
  4. Step 2: iframe Analysis & Web Component Mapping
  5. Step 3: Migration Plan
  6. Step 4: Applied Code Changes
  7. Step 5: Validation
  8. Step 6: Final Summary

Each step section should end with a status line:

  • Status: ✅ complete or Status: ❌ blocked

Steps are sequential — do not start a step until the previous one is ✅ complete.

Evidence requirements

  • Step 0: Metabase version detected (source: Docker tag, env var, or user answer).
  • Step 1: every matched file path, every iframe location, SSO endpoint, layout/head file, Metabase config variables.
  • Step 2: per iframe — parsed URL, content type, ID, hash params, mapped web component with attributes.
  • Step 3: the complete file-by-file change plan with exact old/new code.
  • Step 4: per file — what was changed and exact diffs applied.
  • Step 5: each validation check's pass/fail result with evidence.

Architectural conformance

Follow the app's existing architecture, template engine, layout/partial system, code style, and route patterns. Do not switch paradigms (e.g., templates to inline HTML or vice versa). If the app has middleware for shared template variables, prefer that over duplicating across route handlers.

Performance

  • Maximize parallelism within each step. Use parallel Grep/Glob/Read calls in a single message wherever possible.
  • Do not use sub-agents for project scanning — results need to stay in the main context for cross-referencing in later steps.
  • Do not parse repo branches, commits, PRs, or issues.

Scope

This skill converts Full App / Interactive embedding (iframe-based) to Modular embedding (web-component-based via embed.js).

The consumer's app may be written in any backend language (Node.js, Python, Ruby, PHP, Java, Go, .NET, etc.) with any template engine. Keep instructions language-agnostic unless a specific language is detected in Step 1.

What this skill handles

  • Replacing <iframe> elements pointing to Metabase with appropriate web components (e.g. <metabase-question>, <metabase-dashboard>)
  • Adding the embed.js script tag (exactly once at app layout level)
  • Adding window.metabaseConfig setup code (exactly once at app layout level)
  • Modifying SSO/JWT endpoints to support modular embedding's JSON response format
  • Mapping iframe URL customization parameters to theme config and component attributes

What this skill does not handle

  • Migrating from Static (signed/guest) embedding

Allowed documentation sources

Fetch the version-specific llms-embedding-full.txt using this URL:

https://www.metabase.com/docs/v0.{VERSION}/llms-embedding-full.txt

The version in the URL uses the format v0.58 (normalize: strip leading v or 0., drop patch — e.g., 0.58.1 → 58 → URL uses v0.58). This single file contains all embedding documentation for that version, optimized for LLM consumption.

Other constraints:

  • No GitHub PRs/issues or npm pages
  • Do not follow changelog links to GitHub or guess URLs

AskUserQuestion triggers

Use AskUserQuestion and halt until answered if:

  • The Metabase instance URL cannot be determined from project code or environment variables
  • Always ask for the Metabase instance version — do not rely solely on code detection
  • An iframe URL pattern does not match any known resource type (dashboard, question, collection, home)
  • No SSO/JWT endpoint can be identified in the project
  • No layout/head file can be identified (unclear where to inject embed.js)
  • Multiple layout files exist and it is unclear which one(s) to use
  • The backend language cannot be determined
  • Multiple iframes specify different locale values (ask user which locale to set in window.metabaseConfig)

Pre-workflow steps

Migration Plan Checklist

Create a checklist to track progress. In Claude Code, use TaskCreate/TaskUpdate tools:

  • Step 0: Detect Metabase version
  • Step 1: Scan project + fetch target version docs
  • Step 2: Analyze iframes and map to web components (using docs)
  • Step 3: Plan migration changes
  • Step 4: Apply code changes
  • Step 5: Validate changes
  • Step 6: Final summary

Workflow

Step 0: Detect Metabase instance version

Always AskUserQuestion for the Metabase instance version — even if a version appears in Docker tags or env vars, confirm it with the user. Abort if v52 or older (modular embedding was introduced in v53).

Then fetch llms-embedding-full.txt for the confirmed version (see "Allowed documentation sources" for URL format).

Before anything else, determine the Metabase version. Grep the project for Docker image tags (metabase/metabase:v, metabase/metabase-enterprise:v), METABASE_VERSION, or version references. If undetected, AskUserQuestion (options: v52 or older, v53, v54–v58, v59+). Abort if v52 or older (modular embedding not available — it was introduced in v53). Record the version — it controls jwtProviderUri placement in later steps.

Step 1: Scan the project + fetch docs no sub-agent)

Perform the project scan and doc fetch concurrently — they are independent. Use parallel tool calls within a single message wherever there are no dependencies.

1a: Fetch target version docs

Fetch llms-embedding-full.txt for the target version (see "Allowed documentation sources" for URL format). These docs are the authoritative source for web component attributes, window.metabaseConfig options, and SSO endpoint behavior for the target version. Use them in Step 2 for mapping instead of relying on hardcoded tables alone.

Launch this concurrently with the project scan steps below.

1b: Identify backend language and framework

  • Check for dependency/build files (package.json, requirements.txt, Gemfile, pom.xml, go.mod, composer.json, etc.).
  • Identify the template engine and record the language and framework.

1c: Find ALL Metabase iframes

Use Grep to search for all of these patterns (in parallel):

  • <iframe in all template/HTML/JSX/view files
  • iframe in all server-side code files (JS/TS/Python/Ruby/Go/Java/PHP) — catches iframes built via string concatenation or template literals
  • auth/sso adjacent to iframe or src attributes. Note: the SSO URL may be constructed in a separate variable or function and passed to the iframe src — if the iframe src is a variable, trace its definition to check for auth/sso.

For each file with a match, read the entire file.

1d: Find SSO/JWT authentication code

Use Grep to search for all of these patterns (in parallel):

  • /auth/sso
  • /sso/metabase or similar SSO route patterns
  • jwt.sign or jwt.encode or JWT or jsonwebtoken or PyJWT or jose
  • JWT_SHARED_SECRET or METABASE_JWT_SHARED_SECRET
  • return_to (Metabase SSO redirect parameter)
  • redirect near auth/sso (catches the SSO redirect logic)

For each matching file, read the entire file.

1e: Find the layout/head file(s)

Find the single file (or common code path) where the HTML <head> section is defined — this is where embed.js and window.metabaseConfig will be injected.

Search for:

  • <head> or <!DOCTYPE or <html in template/view files
  • Layout/wrapper patterns: include('head'), <%- include, {% extends, {% block, layout, base.html, _layout, application.html
  • If the app builds HTML via inline strings in server code (e.g., res.send(...)), identify where the <head> content is generated

1f: Find Metabase configuration

Grep for METABASE_ and MB_ prefixed variables. Record every Metabase-related variable name and where it is read.

Output: Structured Project Inventory

Compile all findings into:

Backend: {language}, {framework}, {template engine}
Metabase config:
  - Site URL variable: {name} (read at {file}:{line})
  - Dashboard path variable: {name} (read at {file}:{line})
  - JWT secret variable: {name} (read at {file}:{line})
  - Other variables: ...
Layout/head file: {path}:{line range} (or "inline HTML in {file}:{line range}")
Iframes found: {count}
  - {file}:{line} — {brief description}
  - ...
SSO endpoint: {file}:{line} — {route} ({method})

Step 2: Analyze iframes and map to web components (only after Step 1 ✅)

Use the documentation fetched in Step 1a as the authoritative reference for web component attributes, window.metabaseConfig options, and SSO endpoint behavior. The hardcoded tables below are fallbacks — if the docs describe additional attributes or different behavior for the target version, prefer the docs.

For each iframe found in Step 1:

2a: Parse the iframe URL

Extract from the iframe src attribute (which may be a template expression, variable, or literal):

  • Metabase base URL: may come from env var, constant, or be hardcoded
  • Resource path: the path after the base URL, e.g., /dashboard/1, /question/entity/abc123, /collection/5
  • Resource type: dashboard, question, collection, or home (if path is /)
  • Entity ID or numeric ID: the resource identifier in the path.
    • An ID may be:
      • a numeric id, e.g. 123
      • a numeric id + slug, e.g. 123-slug. You need to remove the slug completely; including the slug will prevent the resource from loading.
      • an entity id — URLs with pattern /{resource_type}/entity/{entity_id} use entity IDs
  • URL hash/query parameters used for UI customization (e.g., #logo=false&top_nav=false)
  • SSO wrapping: whether the iframe goes through an SSO endpoint first (e.g., /sso/metabase?return_to=...)

2b: Map content type to web component

Full App iframe path patternModular Web ComponentRequired Attribute
/dashboard/{id} or /dashboard/entity/{entity_id}<metabase-dashboard>dashboard-id="{id or entity_id}"
/question/{id} or /question/entity/{entity_id}<metabase-question>question-id="{id or entity_id}"
/model/{id} or /model/entity/{entity_id}<metabase-question>question-id="{id or entity_id}"
/collection/{id} or /collection/entity/{entity_id}<metabase-browser>initial-collection="{id or entity_id}"
/ (Metabase home / root)<metabase-browser>initial-collection="root"

If the iframe path is built dynamically from a variable, the web component attribute should use the same variable/expression.

If an iframe path does not match any known pattern → AskUserQuestion.

2c: Map URL customization parameters

Parameters to drop (not applicable — modular web components do not include Metabase application chrome):

Full App ParameterWhy it is dropped
top_navWeb components have no Metabase top navigation bar
side_navWeb components have no Metabase sidebar
logoWeb components have no Metabase or whitelabel logo
searchWeb components have no Metabase search bar
new_buttonNo + New button (use with-new-question / with-new-dashboard on <metabase-browser> if applicable)
breadcrumbsWeb components have no Metabase breadcrumbs

Parameters that map to web component attributes:

Full App ParameterModular Equivalent
header=falsewith-title="false" on the component
action_buttons=falsedrills="false" on the component

Parameters that map to window.metabaseConfig:

Full App ParametermetabaseConfig Property
locale={code}locale: "{code}"

Locale migration rules:

  • If one locale value is found across all iframes → add locale: "{code}" to window.metabaseConfig automatically
  • If multiple different locale values are found across iframes → AskUserQuestion to let the user decide which single locale to set in window.metabaseConfig (modular embedding supports only one global locale)

2d: Output Migration Mapping Table

For each iframe, output:

iframe #{n}: {file}:{line}
  Old: {full iframe HTML or code}
  Content type: {dashboard|question|collection|home}
  ID: {static value or variable expression}
  Dropped params: {list}
  Mapped attributes: {list}
  New: {exact replacement web component HTML}

Step 3: Plan migration changes (only after Step 2 ✅)

Create a complete file-by-file change plan covering all areas below. Every change should be specified with the target file, the old code, and the new code.

3a: embed.js script injection — exactly once per app

  • Target: the layout/head file identified in Step 1e
  • Location: inside <head> (or as close as possible to other <script> tags)
  • Code to add:
    <script defer src="{METABASE_SITE_URL}/app/embed.js"></script>
    
  • {METABASE_SITE_URL} should be rendered dynamically using the project's existing template expression syntax.
  • If the Metabase URL variable is only available in specific routes, pass it to the layout via middleware or template context.
  • Verify this will appear exactly once in the rendered HTML regardless of which page the user visits — if it loads twice, the SDK reinitializes and breaks auth state.

3b: metabaseConfig — exactly once per app

Modular embedding reads its configuration from window.metabaseConfig. There is no defineMetabaseConfig() function — assign the config object directly.

  • Target: same layout/head file as 3a
  • Location: before the embed.js script tag (the config must be set before embed.js loads, otherwise the SDK has no config to read)
  • Code to add (minimum — add auth fields only if the fetched docs list them for this version):
    <script>
      window.metabaseConfig = {
        instanceUrl: "{METABASE_SITE_URL}",
        // Add auth fields here only if supported by the confirmed version's docs
      };
    </script>
    
  • Locale: If a locale parameter was found on any iframe in Step 2c, add locale: "{code}" to the config object. If multiple iframes had different locale values, the user will have already been asked which one to use (per AskUserQuestion trigger).
  • instanceUrl (and jwtProviderUri if supported) should be rendered dynamically using the project's template expression syntax.
  • Auth config fields: Consult the docs fetched in Step 0 to determine which fields window.metabaseConfig supports for the confirmed version. For example, jwtProviderUri may or may not be available. If the docs list it, include it as a full absolute URL (e.g., http://localhost:9090/sso/metabase) — relative paths don't work. If the docs don't list it, the JWT Identity Provider URI must be configured in Metabase admin settings instead (see Step 3f).
  • window.metabaseConfig should be set exactly once — if it appears in per-iframe code instead of the layout, each component will re-initialize the SDK.

3c: SSO endpoint modification

The existing SSO endpoint currently redirects the browser to Metabase's /auth/sso?jwt={token}&return_to={path}.

For modular embedding, the embed.js SDK sends a fetch request to the JWT Identity Provider URI and expects a JSON response. The endpoint should be converted to return JSON only — do not keep a fallback to the old redirect-based auth flow.

This is a full migration, not a gradual one. The old iframe-based embedding is being completely replaced, so the redirect behavior is no longer needed.

Consult the auth docs fetched in Step 0 for the expected SSO endpoint response format for the confirmed version.

Constraints:

  • Do not modify the JWT signing logic — only change how the response is delivered
  • Remove the old redirect behavior entirely — the endpoint should only return JSON
  • The JSON response body should be exactly { "jwt": "<token>" } — no other fields, because the SDK parses this exact shape
  • Remove any code that builds the redirect URL (e.g., new URL("/auth/sso", ...), searchParams.set("return_to", ...)) as it is now dead code

3d: iframe replacement plan

For each iframe from Step 2d's Migration Mapping Table:

  • Specify: file path, exact old code to replace, exact new code
  • The new web component should preserve any dynamic ID expressions from the original iframe URL
  • If the iframe had explicit width/height attributes or inline style, apply them directly to the web component element (e.g., <metabase-dashboard dashboard-id="1" style="width:800px;height:600px">) — do not wrap in a <div>
  • If the iframe was styled via CSS classes, apply those classes directly to the web component
  • If the iframe was inside a container element with styles, keep that container
  • Remove any server-side SSO URL construction that was used only for the iframe src (e.g., building /sso/metabase?return_to=...). But do not remove the SSO endpoint itself — it is still needed for modular embedding auth.
  • If the iframe src was built via a server-side route handler that sends inline HTML (e.g., Express res.send('<iframe ...')), replace the iframe HTML within that handler's response string

3e: Dead code removal

After replacing iframes and converting the SSO endpoint, identify and remove:

  • Variables that built the iframe src URL (e.g., iframeUrl, mbUrl) if they are no longer used anywhere
  • URL parameter/modifier strings that were appended to iframe URLs (e.g., mods = "logo=false") if they are no longer referenced anywhere (check the SSO endpoint — if the redirect logic was removed, these strings may now be dead code too)
  • Redirect-related code removed from the SSO endpoint (e.g., URL construction for /auth/sso, return_to parameter handling) — this is already handled as part of Step 3c
  • Helper functions that constructed Metabase iframe URLs if they are no longer called
  • Do not remove: the SSO endpoint itself, JWT signing function, environment variable reads, or any code that is used by other parts of the application

3f: Metabase admin configuration notes (manual steps for the user)

List these as part of the plan — they will be included in the final summary:

  1. Enable modular embedding: Admin > Embedding > toggle "Enable modular embedding"
  2. Configure CORS origins: Admin > Embedding > Modular embedding > add the host app's domain (e.g., http://localhost:9090)
  3. Configure JWT Identity Provider URI: Admin > Authentication > JWT > set to the full URL of the SSO endpoint (e.g., http://localhost:9090/sso/metabase). Check the fetched docs to determine whether this is required or optional for the confirmed version (it depends on whether window.metabaseConfig supports a JWT provider field).
  4. JWT shared secret: No change needed — reuse the existing shared secret from Full App embedding setup

Step 4: Apply code changes (only after Step 3 ✅)

Apply all changes from Step 3 in this order (backend changes first to minimize the window where things are broken):

  1. First: Modify the SSO endpoint to return JSON (Step 3c) — this is backend-only
  2. Second: Add window.metabaseConfig assignment and embed.js script tag to the layout/head file (Step 3b + 3a, config before embed.js)
  3. Third: Replace each iframe with its web component (Step 3d), one file at a time
  4. Fourth: Remove dead code (Step 3e)

Constraints:

  • Use the Edit tool with precise old_string / new_string for every change
  • Do not add new package dependencies — modular embedding requires only the embed.js script served by the Metabase instance
  • Do not change environment variable names
  • If a file requires multiple edits, apply them top-to-bottom to avoid offset issues

Step 5: Validate changes (only after Step 4 ✅)

Perform all of these checks. Checks 5a–5c can run in parallel (all are independent grep searches). Check 5d and 5e require reading specific files. Each check should have an explicit pass/fail result.

5a: No remaining Metabase iframes

Use Grep to search for <iframe and iframe across all project files (excluding node_modules, .git, lockfiles). Verify that no Full App / Interactive Embedding iframes pointing to Metabase remain. Non-Metabase iframes should be untouched. Also leave any guest embedding (formerly "static embedding") or public embedding iframes untouched — those use different URL patterns (e.g., /embed/ or /public/) and are not part of this migration.

Pass criteria: zero Full App / Interactive Embedding iframes found (guest/public embed iframes are excluded).

5b: embed.js appears exactly once

Use Grep to search for /app/embed.js across all project files (excluding node_modules, .git). This pattern is specific to Metabase's embed script URL and avoids false positives from other tools that may use a generic embed.js filename. Pass criteria: exactly one occurrence in the layout/head file.

5c: window.metabaseConfig is set exactly once

Use Grep to search for window.metabaseConfig across all project files (excluding node_modules, .git). Pass criteria: exactly one occurrence (the assignment in the layout/head file).

5d: SSO endpoint returns JSON only

Read the SSO endpoint file. Verify:

  • The endpoint returns a JSON response with { jwt: token }
  • The old redirect logic (res.redirect, new URL("/auth/sso", ...), return_to) has been fully removed
  • No conditional check for response=json exists (since JSON is the only response format now)

Pass criteria: endpoint returns JSON only, no redirect fallback remains.

5e: Spot-check modified files

Read each modified file and verify:

  • Web components have required attributes (dashboard-id, question-id, or initial-collection)
  • Template syntax is valid (no unclosed tags, correct expressions)
  • Dead-code variables identified in Step 3e have been removed

Pass criteria: all checks pass.

If any check fails:

  • Fix the issue immediately
  • Re-run the specific check
  • If unable to fix after 3 attempts, mark Step 5 ❌ blocked and report which check failed and why

Step 6: Output summary

Organize the final output into these sections:

  1. Changes applied: list every file modified and a one-line description of each change
  2. Web component mapping: table showing each old iframe → new web component
  3. Dropped parameters: list of Full App iframe parameters that were dropped, with brief explanation of why they don't apply to modular embedding
  4. Theme configuration: any theme/appearance settings mapped into window.metabaseConfig
  5. Manual steps required (Metabase admin configuration from Step 3f):
    • Enable modular embedding
    • Configure CORS origins
    • Configure JWT Identity Provider URI
    • Any other admin steps identified
  6. Behavioral differences the user should be aware of:
    • Users can no longer navigate between dashboards/questions/collections within a single embed (each web component is standalone)
    • The Metabase application shell (nav, sidebar, search) is no longer present
    • Any iframe parameters that could not be mapped

Retry policy

Doc fetching:

  • If fetching llms-embedding-full.txt returns 404, verify the Metabase version number and retry. If still failing, mark Step 1 ❌ blocked.

Validation:

  • If any validation check in Step 5 fails after 3 fix attempts, mark Step 5 ❌ blocked and report which check failed and why.
  • If AskUserQuestion is not answered, remain blocked on that step — do not guess or proceed with assumptions.

FAQ

What does the metabase-full-app-to-modular-embedding-upgrade skill do?

Migrates a project from Metabase Full App / Interactive (iframe-based) embedding to Modular (web-component-based) embedding. Use when the user wants to replace Metabase iframes with Modular embedding web components.

How do I install metabase-full-app-to-modular-embedding-upgrade?

Run npx skills add https://github.com/metabase/agent-skills --skill metabase-full-app-to-modular-embedding-upgrade in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/metabase-full-app-to-modular-embedding-upgrade/ (personal) or .claude/skills/metabase-full-app-to-modular-embedding-upgrade/ (project).

Which AI agents can use metabase-full-app-to-modular-embedding-upgrade?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is metabase-full-app-to-modular-embedding-upgrade safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains metabase-full-app-to-modular-embedding-upgrade?

It is published by metabase in metabase/agent-skills. It has 358 recorded installs.

More from metabase/agent-skills

View all →

rde

metabase/agent-skills

Expert data engineering with Metabase at the center of any company's stack: explore and profile raw warehouse tables, build clean tables as Metabase transforms pinned by transform tests, build the semantic layer (models, metrics, measures, segments, metadata), design dashboards, answer questions with checked numbers, reconcile against a reference, change delivered definitions safely. Use for data work, not one mb command (metabase-cli): "make sense of my data", "model this raw schema", "set up analytics for X", "build a data model", "define MRR / active customers officially", "build a semantic layer", "go from raw tables to a dashboard", "does this number match finance", "our numbers look wrong", "two cards disagree", "test this transform", "why does this transform get case X wrong", "make Metabot answer questions about X", "explain this table", "change this definition", "build a data app", "donor retention", "event attendance", "quarterly board report", "be my data engineer / anal

Data & Databases949 installs

rde-data-apps

metabase/agent-skills

Build or change a Metabase data app on an rde machine: settles what the metabase-data-apps skill needs from the local setup (the remote-sync repository, the Metabase URL, the API key rde stores, how the app reaches Metabase), makes sure everything the app reads is Library content already synced into that repository, confirms every sync step with the user, and hands the build to it. Use only when an app is asked for explicitly: "build me an app for X", "create a data app", "make an internal tool / portal in Metabase", "add a page / a form / a filter to my data app". A dashboard, a question, or a metric is not an app; those stay with the rde skill.

Backend & APIs877 installs

metabase-cli

metabase/agent-skills

Drive a Metabase instance from the terminal via the mb CLI: auth profiles, databases and schemas, schema sync and field-values rescan, tables, fields, cards (questions, models, metrics) run as JSON/CSV/XLSX, dashboards and dashcards, collections and their tree, snippets, segments, measures, transforms and transform-jobs, settings, search, git-sync. Use for any Metabase operation from the terminal: "log into metabase", "what profiles do I have", "list cards", "run card 42 as CSV", "create a transform", "list dashboards", "move a dashcard", "list collections", "what's in collection 4", "show the collection tree", "list snippets", "create a segment", "archive a measure", "search metabase for X", "import the latest changes", "add a directory to git sync", "set a setting", "what schemas are in this database", "trigger a sync", "rescan field values", or any mb . Whole data jobs go to rde.

Data & Databases784 installs

metabase-database-metadata

metabase/agent-skills

Understands the Metabase Database Metadata Format — a YAML-based on-disk representation of databases, tables, and fields synced from a Metabase instance. Use when the user needs to read, edit, or understand metadata files produced by @metabase/database-metadata, or when reasoning about a project's schema (columns, types, FK relationships) through the .metadata/databases folder.

Data & Databases564 installs

metabase-representation-format

metabase/agent-skills

Understands the Metabase Representation Format — a YAML-based serialization format for Metabase content (collections, cards, dashboards, documents, segments, measures, snippets, transforms, actions, and data apps). Use when the user needs to create, edit, understand, or validate Metabase representation YAML files, or when working with Metabase serialization/deserialization (serdes). Covers entity schemas, MBQL and native queries, visualization settings, parameters, and folder structure.

AI & Agents535 installs

metabase-semantic-checker

metabase/agent-skills

Open the skill to read its SKILL.md instructions.

AI & Agents432 installs

Popular ai & agents skills

View category →

google-agents-cli-publish

google/agents-cli

This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).

AI & Agents513.9K installs

google-agents-cli-eval

google/agents-cli

This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).

AI & Agents513.5K installs

skill-creator

anthropics/skills

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

OfficialAI & Agents416.4K installs

codex-pet

prime-skills/runcomfy-agent-skills

Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.

AI & Agents402.1K installs

full-output-enforcement

leonxlnx/taste-skill

Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.

AI & Agents401K installs

executing-plans

obra/superpowers

Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available

AI & Agents245.7K installs