Cerb Skills for Claude Code

Claude Code skills for working with Cerb, the automation and workflow platform for teams.

Installation

Skills are installed by placing them in the .claude/skills/ directory of your project.

Claude Code (CLI / Web)

From your project root:

# Clone into your project's skills directory
git clone https://github.com/cerb/claude-skills.git .claude/skills

# Or, if .claude/skills/ already exists, clone and copy the skill directories
git clone https://github.com/cerb/claude-skills.git /tmp/cerb-skills
cp -r /tmp/cerb-skills/cerb-automations .claude/skills/cerb-automations
cp -r /tmp/cerb-skills/cerb-dev .claude/skills/cerb-dev
cp -r /tmp/cerb-skills/cerb-search .claude/skills/cerb-search

Restart Claude Code or start a new session. The skills will be automatically discovered.

Verify the skill is available by typing /cerb-automations or checking:

ls .claude/skills/cerb-automations/SKILL.md

Claude Desktop

1. Click Customize in the left sidebar.
2. Select Skills from the menu.
3. Click the + button.
4. Upload the skill as a ZIP file (download from releases or create from the repository).

Optional: Cerb Docs MCP Server

For live access to Cerb documentation, add the MCP server in your Claude Code or Claude Desktop settings:

• URL: https://api.cerb.cloud/docs/mcp
• No authentication required

Skills

cerb-automations

Create and modify Cerb automations, workflows, and event listeners using KATA syntax.

Trigger: /cerb-automations or when the user asks to create/update an automation.

Capabilities:

• Write automations with any trigger type (automation.function, interaction.worker, webhook.respond, etc.)
• React to events (record.changed, mail.received, mail.route, etc.)
• Use all automation commands: set:, decision:, repeat:, while:, http.request:, record.create/get/search/update/delete:, llm.chat/agent:, queue.push/pop:, metric.increment:, and more
• Build workflows that package automations, event listeners, and configuration together
• Create and send email via draft records (transactional, compose, reply, forward)
• Work with all Cerb record types and their field schemas
• Construct search queries with filters, deep search, and boolean groups
• Configure toolbar interactions and form elements
• Apply automation policies (callers, commands, time limits)

Reference files included:

• kata.md — KATA language syntax, annotations, Twig scripting functions and filters
• automations.md — all commands, triggers, events, policies, and structure
• record-types.md — record type field schemas
• search-queries.md — search query syntax and per-record-type filter fields
• toolbars.md — toolbar and form element configuration
• workflows.md — workflow KATA schema
• icons.md — available icon names

Guides included:

• guide-record-dictionaries.md — record dictionaries and key expansion
• guide-drafts.md — creating draft records to send email
• guide-record-changed.md — building record.changed event automations
• guide-custom-fields.md — working with custom fields in automations

---

cerb-dev

Work on Cerb core and plugin code using the Devblocks PHP framework.

Trigger: /cerb-dev or when the user asks to add record types, write migrations, create extensions, or debug platform internals.

Capabilities:

• Add new record types with full DAO/Model/Context/View/SearchFields boilerplate (via code generator)
• Write database migration patches
• Add fields to existing DAOs and models
• Implement card widgets, cron jobs, and search index backends
• Register extensions in plugin.xml and translations in strings.xml
• Work with worklist subtotals, peek/edit templates, and form handling patterns
• Register and increment platform metrics

Reference files included:

• architecture.md — directory layout, plugin structure, naming conventions, context system, extension points, template paths
• dao-pattern.md — DAO class structure, database operations, events/deltas, form handling, migration patch authoring
• extensions.md — card widget, cron job, and search index extension patterns
• plugin-xml.md — plugin.xml manifest structure, extension points, class loaders
• new-record-type.md — complete step-by-step guide for creating a new record type
• adding-dao-fields.md — adding fields to an existing DAO/model/context
• peek-edit-patterns.md — Smarty gotchas, checkbox groups, dynamic rows, flat lookup sets
• worklist-subtotals.md — adding subtotals to View_ classes
• metrics.md — registering and incrementing metrics
• rerun-patch.md — forcing a database patch to re-run in development

Tools included:

• tools/gen-dao.py — Python generator that writes all PHP and Smarty boilerplate for a new record type from a table name and field list

---

cerb-search

Construct Cerb search queries for any record type.

Trigger: /cerb-search or when the user asks to filter records by field values in a search bar, worklist, automation, or saved search.

Capabilities:

• Build search queries for any Cerb record type (tickets, messages, workers, orgs, contacts, tasks, etc.)
• All filter types: text, fulltext, numeric, boolean, date, chooser, record/deep search, links, watchers, null
• Deep search across related records to any depth
• Boolean groups (AND, OR, NOT with parentheses)
• Sorting and result limiting
• Safe query parameter injection for automation use (record_query_params:)

Reference files included:

• search-queries.md — complete search query syntax, all operators, and filter fields for every record type

Note: cerb-search shares its reference data with cerb-automations. If installing skills individually rather than cloning the full repo, install both together to avoid duplication drift.

Cerb Documentation

The official Cerb documentation is at https://cerb.ai/docs/.

An MCP server is available at https://api.cerb.cloud/docs/mcp for searching and fetching docs from Claude Desktop or Claude Code.

A public docs API is also available (no auth required):

# Semantic search
curl --silent -X POST "https://api.cerb.cloud/docs/search" \
  -H "Content-Type: application/json" \
  -d '{"query":"your search query here"}'

# Fetch pages as LLM-friendly Markdown (comma-separated doc IDs from search results)
curl --silent "https://api.cerb.cloud/docs/fetch/index,pricing"