Agentype

Analyze local AI-agent usage across Claude Code, Codex, OpenCode, pi-agent, Gemini CLI, OpenClaw, Nanobot, and configured Nanobot-compatible roots to produce...

Agentype summarizes a user's local AI-agent history into one deterministic usage overview. Persona discovery is performed by the triggering agent using its own LLM — Agentype itself makes no LLM calls by default.

When to Use

Use this skill when the user asks:

• "what is my agentype?"
• "analyze my agent usage"
• "show my AI usage stats"
• "which agents or models do I use most?"
• "what persona am I based on my AI workflow?"
• /agentype

Do not use it for billing estimates. Agentype reports tokens and local usage signals, not provider invoices.

What It Reads

Agentype collects local session and token metadata from supported agents where available:

• Claude Code
• Codex
• OpenCode
• pi-agent
• Gemini CLI
• OpenClaw
• Nanobot
• Nanobot-compatible JSONL roots configured through AGENTYPE_NANOBOT_ROOTS

Agentype is fully local in this skill workflow. It reads agent history from disk and prints a terminal summary. Handle persona inference on the agent side rather than asking the CLI to contact external model services.

Run

If Agentype is installed:

agentype

If working from a source checkout:

uv run agentype

For users without uv, prefer installing the published CLI:

pipx install agentype-cli
agentype

Custom Local Paths

If a user's agent history lives outside the default locations, ask for the relevant root and configure it before running Agentype. Nanobot-compatible JSONL roots can be added with AGENTYPE_NANOBOT_ROOTS:

AGENTYPE_NANOBOT_ROOTS="/path/to/workspace:/path/to/another/root" agentype --json-out

For unsupported agent layouts, tell the user the collector paths live in src/agentype/paths.py and source adapters live in src/agentype/sources/, so they can add their own local path or adapter before publishing private stats.

Output Modes

• Default: poster-style terminal overview with AGENTYPE/persona first, then token usage, breakdowns, and trends. No LLM calls by default.
• -v: adds detailed tables for statistics, discovered themes, and data confidence.
• --json-out: writes output/agentype.json with the full analysis.
• --json-in PATH: renders a previously written Agentype JSON file. Use this after filling top-level persona fields.
• --png-out: writes output/agentype.png, a shareable poster-style summary for chat or IM environments.

Agent Instructions

When the user invokes this skill:

1. Run agentype --json-out (or uv run agentype --json-out from a source checkout) to collect deterministic local usage into output/agentype.json.
2. Read output/agentype.json.
3. Using your own LLM when needed, infer a persona from the aggregate signals: top projects, agents, models, skill metadata, and usage patterns. Fill these top-level JSON fields: archetype, description, keywords, and comment. Keep the comment to 2-3 concise evidence-grounded sentences starting with "You are a...".
4. Run agentype --json-in output/agentype.json --png-out (or uv run agentype --json-in output/agentype.json --png-out) to render the final terminal result and shareable PNG from the filled JSON.
5. Relay the persona and top usage stats to the user in a compact summary. Attach output/agentype.png when the environment supports files or images.
6. If the user asks for debugging or validation, rerun as agentype -v --json-out or uv run agentype -v --json-out.
7. Do not expose raw session files, prompts, or private transcripts.
8. In chat or IM environments, summarize the terminal result compactly and prefer the PNG for share-oriented requests.