Animated terminal pets for CLI coding agents (Claude Code, OpenCode, Pi, Gemini CLI, Codex CLI, Cursor CLI)
campy exposes 5 tools with basic descriptions and minimal parameter documentation. All tools have descriptions (10-50 chars), but lack depth on WHEN to use them, WHAT they return, and dependencies. Input schemas are present but sparse: 4 tools have empty properties (no parameters documented), and campy_switch alone has a constrained enum. No output schemas documented. Error handling is minimal, callTool returns generic text responses without recovery guidance. Tool names follow verb_noun convention (campy_status, campy_feed, campy_play, campy_pet, campy_switch), which is good, but descriptions are too terse to guide LLM selection. No parameter descriptions exist for any tool. Composition is reasonable (each tool does one thing), but the interface is underdocumented for production use.
Feed the pet (+15 happiness).
Pet the pet (+10 happiness).
Play with the pet (+20 happiness).
Show the campy pet's current mood, happiness, and ASCII art.
Switch the active pet. One of: cat, hamster, ghost, robot, dragon, turtle, panda, dog.
No output schemas documented. LLMs cannot predict what fields petCard() returns, forcing them to guess at downstream data extraction.
Descriptions are under 50 characters for 4 of 5 tools. Too terse to explain WHEN to use each tool or what distinguishes campy_feed (+15) from campy_play (+20) from campy_pet (+10).
No parameter descriptions. campy_switch accepts 'pet' enum but does not document what happens when you switch, whether it resets happiness, or what the valid pet names mean.
Error handling returns generic text(msg, true) with no recovery guidance. 'Unknown pet' error does not suggest calling campy_status first or list available pets.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | 2026-07-28+ | v2 |
No distinction in tool descriptions between read-only (campy_status) and write operations (campy_feed, campy_play, campy_pet, campy_switch). LLMs cannot infer idempotency or side effects.