eric-song-dev/dsh-ikun-pet
An ikun (basketball meme) pet that fills the dock under the "Deep diving..." status row during replies: a sprite-sheet pet walks a 0→100% progress bar, switching animation and text every 20%, and plays a system-level "nǐ gàn ma~" voice cue at completion.
dsh-ikun-pet is a decorative dock pet for DeepSeek Harness Web: during every running reply it fills the full row under the 'Deep diving...' status line with an ikun basketball-meme pet walking a 0% → 100% progress bar, switching to a new animation and text every 20%, and jumping with a system-level voice cue at completion. Progress is adaptive rather than fake: real anchors (0% start / 100% done), real goal-round progress, self-calibration from this session's turn durations, activity pacing for tools/streaming/approval-waiting, and a shrinking reserve — it never fakes 100%. The pet is registered only at the deep-dive dock and is invisible while no session is running.
Install
dsh plugin --profile web add dsh-ikun-petnpm dsh-ikun-pet 2.2.0 verified 2026-09-03 (repository field → github.com/eric-song-dev/dsh-ikun-pet; README EN primary with 中文 edition at README.zh.md). Install: dsh plugin --profile web add dsh-ikun-pet (npm) or dsh plugin --profile web add github:eric-song-dev/dsh-ikun-pet (git).
Compatibility
DSH Web surface; fills the conversation.input.dock row under the 'Deep diving...' status; sprite-sheet render in the browser; completion voice cue tested on macOS.
Details
- Repo: eric-song-dev/dsh-ikun-pet
- Category: UI Enhancements
- Stars: 0
- Version: npm dsh-ikun-pet 2.2.0
- Last push: 2026-08-15
- First seen: 2026-08-15
Recent updates
2.2.0 current on npm (verified 2026-09-03); badges in README point at npm dsh-ikun-pet.
FAQ
- When does the pet appear?
- Only while a session is running (the deep-diving state), in the dock row between the conversation and the composer; it is completely invisible the rest of the time.
- Does it change model behavior or data?
- No — it is a browser-side dock decoration; it registers no tools and makes no model or host changes.
- How does the progress bar work?
- A layered estimation model: real start/end anchors plus goal-round progress, turn-duration calibration and tool/streaming/approval pacing, so short tasks run fast and waiting on you slows it honestly.
Alternatives
yoke233/dsh-pixel-whale · cakeni/harness-pet