catime — AI-Generated Hourly Cat Images 🐱

One-liner: Fetch the latest (or any) AI-generated cat image and send it to the user with a caption and story.

Every hour, catime generates a unique cat artwork using Google Gemini. This skill allows an AI agent to retrieve cat images by number, date, or recency and send them directly to the user via the message tool.

---

Installation

CODEBLOCK0

After installation the catime CLI is available. Verify:

CODEBLOCK1

---

Quick Start for AI Agents

Step 1 — Retrieve a cat image with catime

Run the CLI to get image metadata. The most common command:

CODEBLOCK2

Exact output:

CODEBLOCK3

Step 2 — Parse the output and send the image to the user

Extract the URL from the line starting with URL:. Then use the message tool (or equivalent) to send the image along with a caption.

Example message tool payload (YAML):

CODEBLOCK4

Example message tool payload (JSON):

CODEBLOCK5

---

Command Reference

catime latest

Fetch the most recently generated cat.

Input:
CODEBLOCK6

Output:
CODEBLOCK7

Parse guidance:

• - Line 1 — Header: INLINECODE4
• Line 2 ( URL:) — The image URL. This is the most important line.
• Line 3 ( Idea:) — Short concept in Chinese.
• Line 4 ( Prompt:) — Full image generation prompt in English.
• Line 5 ( Story:) — Narrative story for the cat in Chinese.

catime today

Fetch all cats generated today (UTC). Returns multiple cat entries.

Input:
CODEBLOCK8

Output (excerpt):
CODEBLOCK9

Parse guidance: The first line says Found N cat(s) for 'today':. Then each cat entry follows the same format as catime latest. To send the newest, pick the last entry.

catime <number>

Fetch a specific cat by its sequential number.

Input:
CODEBLOCK10

Output:
CODEBLOCK11

Parse guidance: Same format. Short entries (like numbered lookups) may only show the header and URL lines without Idea/Prompt/Story.

catime --list

List all cats (number + URL, compact format).

Input:
CODEBLOCK12

Output (first 10 lines):
CODEBLOCK13

Parse guidance: Each cat takes 2 lines. Odd lines are headers, even lines are URLs. Use --list | tail -2 to get the latest, or --list | wc -l to count cats (divide by 2).

catime <date>

Fetch cats from a specific date (YYYY-MM-DD format).

Input:
CODEBLOCK14

catime view

Opens the web gallery in the default browser. Not useful for agents — use latest or a number instead.

---

Agent Workflow: Send a Cat Image to User

Follow these steps exactly:

1. Decide which cat to fetch

User request	Command to run
"Show me a cat" / "Send me a cat"	INLINECODE19
"Show me today's cats"

catime today | | "Show me cat number 42" | catime 42 | | "Show me cats from Jan 30" | catime 2026-01-30 | | "Show me a random cat" | catime $(shuf -i 1-241 -n 1) |

2. Run the command and capture stdout

CODEBLOCK15

3. Parse the URL from stdout

Look for the line that starts with URL: . Extract everything after URL: .

Parsing rule (pseudocode):
CODEBLOCK16

Parsing rule (bash):
CODEBLOCK17

4. Parse optional metadata

CODEBLOCK18

5. Send the image to the user

Use the message tool with the extracted URL and metadata:

CODEBLOCK19

6. Fallback behavior

• - If URL is missing: Tell the user the cat data was not available and suggest trying catime latest or a different number.
• If catime is not installed: Run pip install catime first, then retry.
• If network error: The CLI fetches catlist.json from GitHub. If offline, tell the user the cat gallery requires internet access.
• If the user wants multiple cats: Use catime today and parse all entries (each entry separated by a blank line). Send the first or let the user choose.

---

Output Format Reference

Every catime command outputs cat entries in this format:

CODEBLOCK20

Field-by-field guide:

Field	Line prefix	Always present?	Description
Number	INLINECODE32 (header)	✅ Yes	Sequential cat ID (1, 2, 3, …)
Date/Time

header | ✅ Yes | UTC timestamp of generation |
| Model | model: (header) | ✅ Yes | AI model used (e.g. gemini-3-pro-image-preview) |
| URL | URL: | ✅ Yes | Direct link to the image file (.webp or .png) |
| Idea | Idea: | Sometimes | Short concept (usually Chinese) |
| Prompt | Prompt: | Sometimes | Full English prompt used for image generation |
| Story | Story: | Sometimes | Narrative story for the cat (usually Chinese) |

Key notes for parsing:

• - The URL line is always present and is the critical field for sending images.
• INLINECODE40, Prompt, and Story appear in detailed views (latest, today, specific number with detail) but may be absent in --list mode.
• All URLs point to GitHub Releases — they are publicly accessible, no auth needed.
• Image formats are .webp (most common) or .png (older cats).

---

Tips for AI Agents

1. 1. Always use catime latest as the default when the user just wants "a cat."
2. The URL is on the line starting with URL: — this is all you need to send an image.
3. Stories are in Chinese — you can translate them for English-speaking users or include them as-is for bilingual charm.
4. Images are hosted on GitHub Releases — they load fast and don't require authentication.
5. New cats appear every hour — if the user wants a fresh one, latest always works.
6. For random selection, pick a random number between 1 and the latest cat number.
7. --list is efficient for browsing — it only shows 2 lines per cat (header + URL).
8. Do not use catime view — it opens a browser, which is not useful in agent/CLI context.
9. Include the story in your caption when available — it adds personality and delight.
10. The Prompt field is useful if the user is curious about how the image was generated.

---

About catime

• - 🎨 AI-generated cat images every hour using Google Gemini
• 📚 103+ art styles in the style library
• 🐱 Each cat has a unique story and personality
• 🌐 Gallery: yazelin.github.io/catime
• 📦 PyPI: INLINECODE53
• ⭐ GitHub: github.com/yazelin/catime

---

Note: A helper script for automated cat-sending workflows can be written if needed, but the CLI commands above are sufficient for all agent operations.