メニュー
Use when the current Agent LLM cannot process images directly and visual analysis is needed — bridges images through KimiCode CLI print mode to a multimodal Kimi model for text description
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/Dqz00116/skill-lib --skill kimicode-vision-bridgeコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
SOC 職業分類に基づく
SKILL.md を表示中
| name | kimicode-vision-bridge |
| description | Use when the current Agent LLM cannot process images directly and visual analysis is needed — bridges images through KimiCode CLI print mode to a multimodal Kimi model for text description |
| version | 1 |
Routes images through KimiCode CLI's print mode (kimi --print) to give a non-vision Agent LLM visual understanding. Kimi is a multimodal model; print mode provides a non-interactive, programmatic pipe — the Agent submits an image + question via CLI, Kimi returns a text description on stdout.
Agent (no vision) KimiCode CLI Kimi multimodal model
│ │ │
│ image + question │ │
├───────────────────────────────→│ │
│ kimi --print -p "..." │ │
│ ├─────────────────────────────→│
│ │ API call with image │
│ │←─────────────────────────────┤
│ │ text response │
│←───────────────────────────────┤ │
│ text on stdout │ │
▼ ▼ ▼
Print mode docs: https://www.kimi-cli.com/en/customization/print-mode.html
Use this skill when:
Do NOT use when:
Both checks must pass. If either fails, stop and report the failure.
kimi) is installedWindows (PowerShell):
Get-Command kimi -ErrorAction SilentlyContinue | Select-Object -ExpandProperty Source
macOS / Linux:
which kimi || command -v kimi
| Result | Action |
|---|---|
| Found | Pass |
| Not found | Fail. "KimiCode CLI not found. Install from https://www.kimi-cli.com" |
Locate the config file and verify an API key and a vision model are set.
Config locations (try in order):
| Windows | macOS | Linux |
|---|---|---|
%APPDATA%\kimi\config.json | ~/Library/Application Support/kimi/config.json | ~/.config/kimi/config.json |
%USERPROFILE%\.kimi\config.json | ~/.kimi/config.json | ~/.kimi/config.json |
Verify:
apiKey, token, api_key, or under providers)model, defaultModel) — must be vision-capable$KIMI_API_KEY), verify those are set| Result | Action |
|---|---|
| API key + vision model set | Pass |
| API key missing | Fail. "No API key configured. Run kimi config set or edit the config file." |
| Model missing or text-only | Fail. "No vision model configured. Set a multimodal model via kimi config set model <name> (e.g., moonshot-v1-vision, gpt-4o, claude-3.5-sonnet)." |
Once prerequisites pass, the bridge has two steps.
The Agent already has an image (from the user, from a file, from a prior tool call). Combine it with a clear instruction and pipe it through kimi --print.
Approach 1 — File reference (recommended)
Point Kimi at the image file on disk. KimiCode's file-read tool loads and analyzes it.
kimi --quiet -p "Describe every visible element in this image in detail: /path/to/image.png"
Approach 2 — Inline base64 via JSONL stdin (fully programmatic)
Pipe the image as base64 directly — no temp file needed.
BASE64=$(base64 -w 0 /path/to/image.png)
echo "{\"role\":\"user\",\"content\":[{\"type\":\"image_url\",\"image_url\":{\"url\":\"data:image/png;base64,$BASE64\"}},{\"type\":\"text\",\"text\":\"Describe every visible element in this image in detail.\"}]}" \
| kimi --print --input-format=stream-json --output-format=stream-json
Windows PowerShell equivalent:
$base64 = [Convert]::ToBase64String([IO.File]::ReadAllBytes("C:\path\to\image.png"))
$msg = '{"role":"user","content":[{"type":"image_url","image_url":{"url":"data:image/png;base64,' + $base64 + '"}},{"type":"text","text":"Describe every visible element in this image in detail."}]}'
$msg | kimi --print --input-format=stream-json --output-format=stream-json
Effective prompts for the image question:
| Scenario | Prompt |
|---|---|
| Generic | "Describe every visible element — text, layout, colors, positions, any errors or warnings." |
| UI/dialog | "Read every label, button text, input field value, and error message in this screenshot." |
| Code | "Read all visible code including line numbers, syntax highlighting, and any squiggly/error indicators." |
| Chart/graph | "Describe the chart type, axes labels, data ranges, trend direction, and any annotations." |
| Document | "Transcribe the visible text exactly. Note any formatting (bold, headers, tables)." |
| Flags | Output format | How to parse |
|---|---|---|
--quiet | Plain text, final message only | Read directly |
--output-format=stream-json | JSONL, one JSON per line | grep '"role":"assistant"' | tail -1 | jq -r '.content' |
Inject into Agent context:
[Vision bridge: KimiCode print mode]
<text from stdout>
[End vision bridge]
IMAGE="/path/to/image.png"
RESULT=$(kimi --quiet -p "Describe every visible element in $IMAGE in detail" 2>/dev/null)
echo "[Vision bridge: KimiCode print mode]"
echo "$RESULT"
echo "[End vision bridge]"
The bridge works with any image the Agent can reference. Common sources:
kimi --quiet -p "Describe ~/Downloads/screenshot.png"screencapture, Windows: PowerShell GDI+, Linux: ImageMagick/gnome-screenshot)Screenshot capture is not part of the bridge skill itself — use platform-native tools.
"Focus only on reading every text string in the dialog box.""Compare these two screenshots and describe what changed."| Mistake | Fix |
|---|---|
| Using a text-only model | Run kimi config set model <vision-model> to switch |
| Image path contains spaces or special chars | Wrap path in quotes or use a temp file with a simple name |
| Expecting KimiCode to take screenshots | Use OS tools; KimiCode is the analysis pipe, not the capture tool |
| Forgetting to verify prerequisites first | Always run Check 1 and Check 2 before attempting the bridge |
| Not checking exit codes | 0=success, 1=permanent error (auth/config), 75=transient (rate limit, retry) |
| Symptom | Likely cause | Action |
|---|---|---|
kimi: command not found | CLI not on PATH | Install from https://www.kimi-cli.com |
| Exit code 75 | Rate limit / transient | Wait 10s, retry |
| Exit code 1 | Auth or config error | Run kimi config to check API key and model |
| Empty or nonsensical output | Model may not be vision-capable | Verify the model supports multimodal input |
| "Image input not supported" in output | Model is text-only | Switch to a vision model in KimiCode config |
| Base64 too large for stdin | Image too big | Use file-reference approach (Approach 1) |
| Output appears truncated | Long response | Use --output-format=stream-json for reliable capture |