Skip to content

README.md

Latest commit

 

History

History
80 lines (59 loc) · 2.49 KB

README.md

File metadata and controls

80 lines (59 loc) · 2.49 KB

1P BigQuery Skill Sample

This sample demonstrates the 1P (first-party) Skills pattern: combining a raw toolset (BigQueryToolset) with a curated skill (SkillToolset) to give the agent both tools and guided workflows.

What This Shows

  • BigQueryToolset provides raw tools: execute_sql, list_dataset_ids, get_table_info, etc.
  • SkillToolset provides skill discovery and loading: list_skills, load_skill, load_skill_resource, run_skill_script.
  • The bigquery-data-analysis skill ships pre-packaged with ADK and follows the agentskills.io specification.

The agent can discover the skill at runtime, load its instructions for guided workflows, and access reference materials on-demand.

Setup

  1. Install ADK with BigQuery extras:

    pip install google-adk[bigquery]

  2. Set up OAuth credentials:

    • Create OAuth 2.0 credentials in the Google Cloud Console.
    • Update agent.py with your client_id and client_secret.

  3. Run the sample:

    adk web contributing/samples

  4. Select 1p_bigquery_skill from the agent list.

How It Works

User Query
    |
    v
LlmAgent (gemini-2.5-flash)
    |
    +-- BigQueryToolset tools (direct data access)
    |     list_dataset_ids, list_table_ids, get_table_info,
    |     execute_sql, forecast, detect_anomalies, ...
    |
    +-- SkillToolset tools (guided workflows)
          list_skills         -> discovers "bigquery-data-analysis"
          load_skill          -> loads step-by-step instructions
          load_skill_resource -> loads sql_patterns.md, etc.
          run_skill_script    -> executes skill scripts

Progressive Disclosure

The skill uses three levels of content:

  1. L1 - Metadata (always available): skill name and description shown via list_skills.
  2. L2 - Instructions (on activation): full workflow steps loaded via load_skill.
  3. L3 - References (on demand): detailed SQL patterns, schema exploration guides, and error handling loaded via load_skill_resource.

This keeps the agent's context efficient while making deep knowledge available when needed.

Extending This Pattern

Other toolsets can follow the same pattern:

  1. Create a spec-compliant skill directory under integration/<toolset>/skills/.
  2. Add a get_*_skill() convenience loader.
  3. (Optional) Add an alias in tools/<toolset>/ for backward compatibility.
  4. Users add both the toolset and SkillToolset to their agent's tools.