playbooks

home / skills / trpc-group / trpc-agent-go / recommend_poi

recommend_poi skill

/examples/skilldynamicschema/skills/recommend_poi

This skill provides deterministic city POI recommendations with a structured JSON output, guiding users to top attractions efficiently.

npx playbooks add skill trpc-group/trpc-agent-go --skill recommend_poi

Review the files below or copy the command above to add this skill to your agents.

Files (2)
SKILL.md
673 B
---
name: recommend_poi
description: Recommend a POI in a city (schema + deterministic result).
---

# recommend_poi

This skill is used by the dynamic structured output demo.

## Output JSON Schema

```json
{
  "type": "object",
  "properties": {
    "poi": {
      "type": "string",
      "description": "Point of interest"
    },
    "city": {
      "type": "string",
      "description": "City name"
    },
    "score": {
      "type": "integer",
      "description": "A deterministic score"
    }
  },
  "required": [
    "poi",
    "city",
    "score"
  ],
  "additionalProperties": false
}
```

## Commands

Print JSON result to stdout:

```bash
cat result.json
```

Overview

This skill recommends a single point of interest (POI) for a specified city and returns a deterministic score. It outputs a compact JSON object with fields for poi, city, and score. The design prioritizes reproducible results suitable for demos and structured-output pipelines.

How this skill works

The skill inspects the provided city input and maps it to a deterministic POI choice using a fixed rule set. It computes an integer score with a repeatable formula so the same city always yields the same poi and score. The output strictly follows the JSON schema: poi, city, and score are returned and no extra properties are included.

When to use it

  • Demonstrating structured JSON outputs in agent-driven systems.
  • Testing deterministic behavior in pipelines that require repeatable results.
  • Providing a single, clear POI suggestion for UI prototypes or examples.
  • Validating schema enforcement and downstream JSON parsing.
  • Integrating into demos where predictable LLM/tool orchestration is needed.

Best practices

  • Provide the exact city name to avoid ambiguous mappings.
  • Use this skill for examples and deterministic testing rather than exhaustive discovery.
  • Validate the returned JSON against the schema before consumption.
  • Combine with additional tools for rich POI metadata when needed.
  • Keep user-facing text separate from this skill’s JSON output.

Example use cases

  • A demo agent that shows a single recommended landmark for a given city.
  • Automated tests that assert consistent outputs across runs.
  • UI mockups that display a POI card with a deterministic score.
  • Onboarding guides that illustrate structured outputs from agent tools.
  • Backend components that require a stable, schema-bound suggestion.

FAQ

Is the recommendation random?

No. The recommendation and score are deterministic and repeatable for the same city.

Can I get multiple POIs or additional metadata?

This skill returns a single POI and score only. For multiple results or richer metadata, layer additional tools or services.