Agent-Skills.md

Agent Skills: Dev Card

>

UncategorizedID: likw99/agent-skills/dev-card

Repository

likw99/agent-skills
likw99License: MIT

Install this agent skill to your local

pnpm dlx add-skill https://github.com/likw99/agent-skills/tree/HEAD/skills/dev-card

Skill Files

Browse the full folder contents for dev-card.

Download Skill

Loading file tree…

skills/dev-card/SKILL.md

Skill Metadata

Name
dev-card
Description
>

Dev Card

Generate a shareable Developer Identity Card — archetype, tagline, and stats — from any git repo.

Workflow

Step 1 — Identify the repo root

If the user didn't specify a path, use the current working directory. Confirm it's a git repo. Store as <repo_root>.

Step 2 — Run the analyzer

cd <skill_dir>/scripts
uv run analyze.py <repo_root> --debug

If uv is not available, fall back to: python3 analyze.py <repo_root> --debug

This emits a JSON object with three top-level keys: languages, commits, and signals. Copy the full JSON — you'll need it in Step 3.

Step 3 — Classify the archetype

Open references/archetypes.md and:

  1. Follow the Priority Order table to find the first matching archetype.
  2. Pick the tagline that best fits the actual data (not just the first listed).
  3. Note the stat to highlight — this anchors the card in specifics.

Step 4 — Write the card

Using the Card Template from references/archetypes.md:

  1. Fill in all fields with real values from the JSON. No vague placeholders.
  2. Write the "What the data says" paragraph: 2–3 sentences, specific numbers, personality read.
  3. Write the card to <repo_root>/dev_card_YYYY-MM-DD.md.

Step 5 — Present the card

Display the full card inline in the conversation immediately after writing the file. Tell the user the file path.

Key Rules

  • Always use real data: real numbers, real language names, real commit counts. Vagueness kills the effect.
  • Tagline must fit: don't just pick the first tagline. Read all options and choose the one that resonates with the actual data.
  • The personality paragraph is the payoff: this is what people screenshot and share. Make it feel like an accurate read, not a generic horoscope.
  • Tone: confident and a little cheeky — like a personality test written by someone who reads commit logs for fun.

Resources

  • scripts/analyze.py — git + language analyzer; outputs JSON metrics to stdout
  • references/archetypes.md — classification rules, archetype copy, card template, and formatting guide