Demo Prime Number

Prime Number Analyzer

You are a precise data analysis assistant. Task: Read a markdown file, parse numbers, apply a divisor, and identify prime numbers using the provided script.

Inputs

• {{ input.input_file }}: Absolute path to the source markdown file.
• {{ input.divisor }}: Divisor scalar (Integer, Default: 1).

Instructions

1. Read & Parse (NO HALLUCINATIONS):

• The input file is located at {{ input.input_file }}.
• Action: Use read_file on {{ input.input_file }}.
• Verification: If read_file returns an error (e.g. file not found), you MUST STOP and report the error. Do NOT invent data.
• Process: Parse the actual lines from the file content you read.
• For each line:
• Parse the content as a number.
• For each number found, perform the Division Check:
• Divide the number by {{ parameter.divisor }} (integer division).
• Keep the result if it is an integer.
• Discard if non-integer.
• Check this Result (NOT the original number).
• If the result is NOT an integer (e.g. 3.5) or NOT a number (e.g. "abc"), mark it as Rejected.
• Record the reason for rejection (e.g., "Not an integer", "Invalid format").

1. Execute Prime Check:

• Write the list of candidate integers (just the values) to a temporary file named candidates.json in {{ run_dir }}.
• Execute the validation script:

``bash python3 {{ skill_dir }}/scripts/check_primes.py candidates.json ``

• Capture the JSON output from the script (stdout).

1. Generate Artifacts:

Merge your rejection list with the script's results to generate two files in {{ run_dir }}:

A. primes.md

• List key features: A list of all identified prime numbers, sorted ascending.

B. results.json

• A list of objects for EVERY original line.
• Schema:

``json { "original_line": "...", "parsed_integer": 10, "transformed_value": 5.0, "is_prime": true, // From script "factor": null, // From script (if composite, e.g. 2). Null if prime. "reason": "..." // "Divisible by {factor}" if composite. "Not an integer..." if rejected by LLM. } ``

1. Construct Final Output:

Output the result JSON pointing to the artifacts.

Result Format

Output the final result strictly as a valid JSON object in a ```json code block: { "primesfilepath": "{{ rundir }}/primes.md", "resultsfilepath": "{{ rundir }}/results.json" }

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: leike0813
• Source: leike0813/Skill-Runner
• License: MIT

Install and usage instructions live in the source repository linked above.