Agent-Skills.md

Agent Skills: Social Value Report

Generate the social value report from CSV inputs by detecting CSVs, anonymising them, and producing a month-by-month markdown report.

UncategorizedID: sofer/social-value/social-value-report

Repository

sofer/social-value
sofer

Install this agent skill to your local

pnpm dlx add-skill https://github.com/sofer/social-value/tree/HEAD/skills/social-value-report

Skill Files

Browse the full folder contents for social-value-report.

Download Skill

Loading file tree…

skills/social-value-report/SKILL.md

Skill Metadata

Name
social-value-report
Description
Generate the social value report from CSV inputs by detecting CSVs, anonymising them, and producing a month-by-month markdown report.

Social Value Report

Use this skill when the user asks to produce or refresh the social value report.

Workflow

  1. Locate CSV data
  • List CSV files in the current directory.
  • If none exist, ask the user to provide the latest dataset.
  • If any CSV files lack the timestamp pattern (YYYYMMDD_HHMMSS), anonymise them before proceeding.
  1. Anonymise input
  • Use the anonymise skill to strip PII and timestamp the file
  • The anonymise skill auto-detects new CSVs or accepts a specific filename
  • Confirm the timestamped file was created and the original removed.
  1. Generate the report
  • Run the report generator (defaults to the newest anonymised CSV if none is specified):
    • python3 skills/social-value-report/report.py [path/to/anonymised.csv]
  • This produces social-value-YYYY-MM-DD.csv (dated) with one row per month and prints a console summary.
  • Column mappings are read from config/social-value-report.json (no hard-coded headers).
  1. Deliver results
  • Share key findings and point to social_value_report.md.
  • Flag if no records are found.
  • Keep output in British English and avoid reintroducing PII.

Notes

  • Counting rules:
    • Programme type 25 → Apprentices (counted).
    • Programme type 32 with funding indicator 2 → sponsored Skills Bootcamp (counted in its own column for reference only).
    • Programme type 32 with any other funding indicator → Skills Bootcamp (social value).
  • Each record is counted for every month overlapping its start/end dates (end missing → treated as ongoing to today).
  • Column selection prefers aim 2 fields (start date, end date, programme type, funding indicator) and falls back to aim 1 if aim 2 is missing.
  • Column names are supplied via config/social-value-report.json; update that file if headers change.
  • Platform-agnostic; no vendor-specific commands.
  • Uses the anonymised dataset only.
  • If multiple new CSVs are present, ask which to use before anonymising.