CtrlK
BlogDocsLog inGet started
Tessl Logo

supervisor-api

Replace the client-side agent loop with Databricks Supervisor API (hosted tools + client-side function tools). Use when: (1) User asks about Supervisor API, (2) User wants Databricks to run the agent loop server-side, (3) Connecting Genie spaces, UC functions, agent endpoints, or MCP servers as hosted tools, (4) Mixing client-side function tools (Python callables your app executes) with hosted tools.

72

Quality

87%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

Quality

Content

75%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable with a clear multi-step workflow and good cross-skill navigation, but loses conciseness to duplicated tracing code that could be consolidated. No bundle files exist (references/scripts/assets absent), so progressive disclosure is scored on the in-file structure and signaled references.

Suggestions

Consolidate the tracing setup: Step 3 already shows `_get_trace_destination`, `_extra_body`, and the instrumented `responses.create` call — in 'Enabling Tracing', reference that rather than re-listing the full function and call, keeping only the new rationale (how spans link, the UC storage requirement).

Move the lengthy 'MCP Server Tools: Multi-Turn Approval Flow' detail into a reference file (or the sibling skill) and keep a short summary plus the follow-up input example inline, tightening the main SKILL.md.

Add an explicit validate/verify checkpoint after the deploy step (e.g., confirm the app responds and traces appear in MLflow) to lift workflow clarity toward a full feedback loop.

DimensionReasoningScore

Conciseness

Mostly efficient and largely domain-specific, but the tracing setup is duplicated — `_get_trace_destination` and the `responses.create(...)` call appear in full in Step 3 and again (a stricter variant) in 'Enabling Tracing' — which could be tightened; fits 'mostly efficient but includes some unnecessary explanation' rather than 4's 'minor instances'.

3 / 5

Actionability

Provides copy-paste-ready, executable guidance throughout — pyproject dependencies, the full TOOLS list, complete `agent.py` invoke/stream handlers, app.yaml scopes, permission tables, and deploy commands — covering the common cases.

5 / 5

Workflow Clarity

A clear Step 1–5 sequence with a local-test step (`uv run start-app`) and a fail-fast `.env`/experiment checkpoint gives most checkpoints; not 5 because there is no explicit validate-then-fix loop for the deploy/tracing config path, and not 3 since the sequence and key checkpoints are present.

4 / 5

Progressive Disclosure

Clear section headers and well-signaled one-level references to sibling skills ('See the **supervisor-api-client-function-calling** skill', 'supervisor-api-background-mode', 'add-tools') plus the external MLflow docs; minor gap is that the sizable tracing and MCP-approval content is fully inlined rather than split out, so it does not reach 5.

4 / 5

Total

16

/

20

Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, complete, and distinct: it states the concrete capability and provides four explicit 'Use when' triggers with natural terminology. It uses imperative/third-person voice with no first- or second-person phrasing, so no voice penalty applies.

DimensionReasoningScore

Specificity

Lists multiple concrete actions — 'Replace the client-side agent loop', 'Connecting Genie spaces, UC functions, agent endpoints, or MCP servers as hosted tools', 'Mixing client-side function tools' — with comprehensive coverage, matching the top anchor.

5 / 5

Completeness

Clearly answers both 'what' (replace the client-side agent loop with the Supervisor API, hosted + client-side function tools) and 'when' (four explicit numbered trigger clauses), matching the anchor-5 example structure.

5 / 5

Trigger Term Quality

An explicit 'Use when: (1)...(4)' list surfaces natural terms users would say — 'Supervisor API', 'run the agent loop server-side', 'Genie spaces', 'MCP servers' — giving comprehensive keyword coverage; not below 5 since the domain has no file extensions/synonyms to add.

5 / 5

Distinctiveness Conflict Risk

A clear niche — Databricks Supervisor API with specific hosted-tool types — makes it highly distinguishable from other skills with minimal conflict risk.

5 / 5

Total

20

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
databricks/app-templates
Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.