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 UCSC-VLAA/VisualClaw --skill video-action-identificationgit clone --depth 1 https://github.com/UCSC-VLAA/VisualClawWrote 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/ucsc-vlaa/visualclaw/video-action-identification)<a href="https://agentmods.dev/skills/ucsc-vlaa/visualclaw/video-action-identification"><img src="https://agentmods.dev/badge/skills/ucsc-vlaa/visualclaw/video-action-identification.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00035 | $0.00257 |
| Opus 5 | $0.00017 | $0.00129 |
| Sonnet 5 | $0.00007 | $0.00051 |
| Haiku 4.5 | $0.00003 | $0.00026 |
Grade A, and why
video-action-identification 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 8d 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.
What it actually says
Identify Core Video Action vs. Secondary Details
- Watch the video mentally and identify the MAIN activity (e.g., painting, pruning, playing cards).
- For each option, separate:
- Core process: The central repeated action
- Secondary details: One-time interactions, posture changes, or peripheral activities
- Select the option that captures the core process WITHOUT excessive secondary details.
- Reject options that add deviations or tangential actions (e.g., "interacts with a boy and stands up from a chair").
- Reject options that misidentify the activity (e.g., "harvesting" when the action is "pruning").
Example: If the video shows someone pruning plants, reject "harvesting" or "watering" even if those words appear plausible. Reject descriptions that add "also interacts with a boy" unless that's the core activity.
Anti-pattern: Selecting an option just because it contains some correct details mixed with incorrect secondary actions. Always prioritize accuracy of the PRIMARY action.
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.
- 8d ago First seen · 20 lines · 35 tokens per session scan A 9b2143ccab9c
video-action-identification is a skill published in the GitHub repository UCSC-VLAA/VisualClaw (55 stars, last pushed 2mo ago), licensed MIT. It adds 35 tokens to every session and 257 once invoked, about $0.0002 per session on Opus 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
continuous-learning
Pattern extraction, confidence-scored evaluation, skill creation, organization, versioning, and cross-project export pipeline.
moai-workflow-jit-docs
Enhanced Just-In-Time document loading system that discovers, loads, and caches relevant documentation based on user intent and project context. Use when users need specific documentation on demand.
openakita/skills@youtube-summarizer
Summarize YouTube videos by extracting transcripts and generating structured notes. Use when the user wants to summarize a YouTube video, extract key points from a talk, create study notes from a lecture, or get timestamps for important moments. Supports multiple URL formats and languages.
openakita/skills@baidu-scholar
Baidu Scholar academic search skill. Search academic papers, journals, and knowledge resources. Use when user needs academic literature, research papers, or scholarly information.
openakita/skills@baidu-video-notes
AI Video Notes skill for video content analysis and note generation. Extract key information from videos for learning, meetings, and content summarization. Use when user wants to analyze video content or generate notes from videos.
goga-discover
Interview the user relentlessly about a decision until every branch of the design tree is resolved, then record the result as a short ADR.