Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add xiincs/claude-code-vision-skill --skill visiongit clone --depth 1 https://github.com/xiincs/claude-code-vision-skillWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/xiincs/claude-code-vision-skill/vision)<a href="https://agentmods.dev/skills/xiincs/claude-code-vision-skill/vision"><img src="https://agentmods.dev/badge/skills/xiincs/claude-code-vision-skill/vision/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/xiincs/claude-code-vision-skill/vision"><img src="https://agentmods.dev/badge/skills/xiincs/claude-code-vision-skill/vision.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00049 | $0.01261 |
| Opus 5.5 | $0.00020 | $0.00504 |
| Sonnet 5.5 | $0.00010 | $0.00252 |
| Haiku 4.5 | $0.00005 | $0.00126 |
Grade A, and why
vision scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
Copies of this mod
1 near-identical copy found in the catalogue:
- vision — 89% identical, 25 lines differ
How it starts
The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.
vision
Multi-provider vision tool. Call various vision models to describe images. Feed it a prompt + image path, get back a text description.
When to use this tool
If you can already see and understand the image yourself (native multimodal model), skip this tool — analyze it directly.
A SessionStart hook normally announces this session's routing status up front. If that context isn't visible (e.g. compacted out of a long conversation, or the hook isn't installed), check before calling this tool:
python vision.py --check-routing
native→ you already have native image understanding this session; don't call this tool.external(default) → proceed with the quick start below.
Quick start
python vision.py [--provider <name>] <image_path> <prompt>
When --provider is omitted, the provider is resolved by: --provider flag > VISION_PROVIDER env > first API key found.
Providers
doubao (Volcengine Ark)
- API key:
DOUBAO_API_KEY - Default model:
doubao-seed-2-0-pro-260215 - Custom endpoint:
DOUBAO_BASE_URL
qwen (DashScope)
- API key:
DASHSCOPE_API_KEY - Default model:
qwen-vl-max - Custom endpoint:
DASHSCOPE_BASE_URL - Available models:
qwen-vl-max,qwen-vl-plus,qvq-max
deepseek (DeepSeek)
- API key:
DEEPSEEK_API_KEY - Default model:
deepseek-v4-flash-vision-exp - Custom endpoint:
DEEPSEEK_BASE_URL - Only
deepseek-v4-flash-vision-expaccepts images —deepseek-v4-flashanddeepseek-v4-proare text-only and reject image input with an error.
openai (GPT-4o)
- API key:
OPENAI_API_KEY - Default model:
gpt-4o - Custom endpoint:
OPENAI_BASE_URL - Also works with any OpenAI-compatible endpoint.
anthropic (Claude)
- API key:
ANTHROPIC_API_KEY - Default model:
claude-sonnet-5 - Custom endpoint:
ANTHROPIC_BASE_URL - Requires the
anthropicpackage (pip install anthropic); it's imported lazily so other providers work without it.
any custom provider
Any --provider name outside the built-in ones is resolved dynamically from
environment variables named after it — no code changes needed:
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 128 lines · 49 tokens per session scan A 9ca142d965d6
vision is a skill published in the GitHub repository xiincs/claude-code-vision-skill (165 stars, last pushed 2mo ago), licensed MIT. It adds 49 tokens to every session and 1,261 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
blip-2-vision-language
Vision-language pre-training framework bridging frozen image encoders and LLMs. Use when you need image captioning, visual question answering, image-text retrieval, or multimodal chat with state-of-the-art zero-shot performance.
llamaindex
Data framework for building LLM applications with RAG. Specializes in document ingestion (300+ connectors), indexing, and querying. Features vector indices, query engines, agents, and multi-modal support. Use for document Q&A, chatbots, knowledge retrieval, or building RAG pipelines. Best for data-centric LLM…
moe-training
Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scaling model capacity without proportional compute increase. Covers MoE…
nemo-curator
GPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication (16× faster), quality filtering (30+ heuristics), semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with RAPIDS. Use for preparing high-quality training datasets, cleaning web data, or…
sentence-transformers
Framework for state-of-the-art sentence, text, and image embeddings. Provides 5000+ pre-trained models for semantic similarity, clustering, and retrieval. Supports multilingual, domain-specific, and multimodal models. Use for generating embeddings for RAG, semantic search, or similarity tasks. Best for production…
clawrouter
Hosted-gateway LLM router — save 84% on inference costs. A local proxy that forwards each request to the blockrun.ai gateway, which routes to the cheapest capable model across 90 models from OpenAI, Anthropic, Google, DeepSeek, xAI, Z.AI, and more. Free open-weight models included. Also exposes realtime market data…