AI agent skills
Masthead provides a collection of pre-built AI agent skills that allow agents, such as Google Antigravity or Claude Code, to analyze and triage issues programmatically. These skills help agents interpret BigQuery dataset schemas, lineage information, and cost optimization recommendations more efficiently.
All skills are available in the for-agents GitHub repository and ship as one plugin, masthead-agent-tools, for Claude Code, OpenAI Codex, and Google Antigravity. The plugin bundles the skills, two workflow commands, and the Masthead MCP server connection.
Installation
Section titled “Installation”Install the plugin in your coding agent, restart it, then sign in to the masthead MCP server once. The repository README is the source of truth for these steps; the Agent tools button in the Masthead app copies a setup prompt that walks your agent through them.
| Client | Install | MCP sign-in |
|---|---|---|
| Claude Code | /plugin marketplace add masthead-data/for-agents then /plugin install masthead-agent-tools@masthead-data |
/mcp, authenticate masthead |
| OpenAI Codex | codex plugin marketplace add masthead-data/for-agents then codex plugin add masthead-agent-tools@masthead-data |
codex mcp login masthead |
| Google Antigravity | Clone the repository into .agents/plugins/masthead-agent-tools or ~/.gemini/config/plugins/masthead-agent-tools |
Agent Settings, Customizations, masthead, Authenticate |
Plugins load at startup, so restart the client after installing. A 401 Unauthorized from any Masthead tool means the sign-in step is still missing.
Workflow commands
Section titled “Workflow commands”| Workflow | Claude Code | Codex | Antigravity |
|---|---|---|---|
| Incident triage: open incidents, lineage, blast radius, mitigation plan | /masthead-agent-tools:triage |
$masthead-agent-tools:masthead-triage |
/masthead-triage |
| FinOps audit: dead-end and unused tables, dataset billing models, pipelines, workload assignments | /masthead-agent-tools:savings |
$masthead-agent-tools:masthead-savings |
/masthead-savings |
Both accept an optional scope, for example ... savings tables in project my-project or ... triage incident <uuid>.
The FinOps skills read the insights table from your Masthead BigQuery export. They resolve the dataset through the get_tenant_settings MCP tool and stop with a link to request access if the export isn’t enabled for your tenant. The incident triaging skill needs only the MCP server.
Observability and incident management
Section titled “Observability and incident management”Incident triaging skill
Section titled “Incident triaging skill”Triage, analyze downstream impact, and mitigate data quality and pipeline incidents using the Masthead MCP server.
- Associated resources: Masthead MCP server tools, including incidents and project metadata
- Skill definition: masthead-incident-triaging
Asset prioritization skill
Section titled “Asset prioritization skill”Suggest which tables and BI assets (Looker dashboards and looks, Looker Studio reports, Google Sheets) to raise to PRIORITY or CRITICAL monitoring, or lower back to REGULAR, from Masthead asset scores, with a reason per asset taken from the score metrics, and apply the change on confirmation.
- Associated resources: Masthead MCP server tools
list_asset_scores,update_table_priority, andupdate_bi_asset_priority - Skill definition: masthead-asset-prioritization
The in-app prompt runs this skill in recommendation mode; apply tier changes only after confirming in your agent session.
Cost optimization
Section titled “Cost optimization”These skills help AI agents interpret the insights table to recommend cost-saving operations across workload assignments, pipelines, datasets, and tables. They never execute a change: the agent prepares commands and a review list for you to run.
Compute savings with workload assignments
Section titled “Compute savings with workload assignments”Optimize BigQuery compute costs by assigning data models (Dataform, dbt, Airflow) and principals (service accounts, users) to slot reservations or on-demand compute based on Masthead recommendations.
- Associated resources: The
insightstable in the Masthead BigQuery dataset, specifically compute recommendations of typeCompute costsfor data models and principals - Skill definition: masthead-compute-savings-with-workload-assignments
Compute savings with pipelines
Section titled “Compute savings with pipelines”Optimize BigQuery compute costs by identifying and pausing/disabling unused, dead-end, or inefficient pipelines that consume compute resources.
- Associated resources: The
insightstable in the Masthead BigQuery dataset, specifically compute recommendations for scheduled pipelines - Skill definition: masthead-compute-savings-with-pipelines
Storage savings with datasets
Section titled “Storage savings with datasets”Optimize BigQuery storage costs at the dataset level by switching storage billing models and setting expiration policies.
- Associated resources: The
insightstable in the Masthead BigQuery dataset, specifically dataset-level storage billing recommendations - Skill definition: masthead-storage-savings-with-datasets
Storage savings with tables
Section titled “Storage savings with tables”Optimize BigQuery storage costs by identifying and removing dead-end and unused tables.
- Associated resources: The
insightstable in the Masthead BigQuery dataset, specifically storage recommendations for dead-end and unused tables - Skill definition: masthead-storage-savings-with-tables