Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/calesthio/generative-media-skillsnpx agentmods add skills/calesthio/generative-media-skills/tencent-hunyuanvideoWrote 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/calesthio/generative-media-skills/tencent-hunyuanvideo)<a href="https://agentmods.dev/skills/calesthio/generative-media-skills/tencent-hunyuanvideo"><img src="https://agentmods.dev/badge/skills/calesthio/generative-media-skills/tencent-hunyuanvideo/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/calesthio/generative-media-skills/tencent-hunyuanvideo"><img src="https://agentmods.dev/badge/skills/calesthio/generative-media-skills/tencent-hunyuanvideo.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 342 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00077 | $0.07370 |
| Opus 5 | $0.00039 | $0.03685 |
| Sonnet 5 | $0.00015 | $0.01474 |
| Haiku 4.5 | $0.00008 | $0.00737 |
Grade A, and why
tencent-hunyuanvideo scanned grade A with 2 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 9d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
from urllib.error import HTTPError, URLError Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
subprocess.run(["ffprobe","-v","error","-show_streams","-show_format",str(part)], check=True) How it starts
The opening of the file, as written. The whole thing — 393 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Tencent HunyuanVideo
Separate every request into one of three contracts before taking action:
- TokenHub
hy-video-1.5— current managed Tencent Cloud T2V/I2V interface for new integrations. - Legacy Tencent Cloud
vclmAPI 3.0 — older TC3-signedSubmitHunyuanToVideoJobsurface retained for existing customers while sales and model capabilities migrate to TokenHub. - HunyuanVideo-1.5 open weights — the official 8.3B local T2V/I2V repository, checkpoints, dependencies, infrastructure, and community license.
Do not infer that the managed model is byte-identical to a public checkpoint. Do not apply the open-weight license to cloud output, or Tencent Cloud terms to downloaded weights.
Gate execution first
Before a paid request, asset upload, model download, environment installation, or training job:
- State the surface, exact model/action or checkpoint family, region/territory, mode, input, target resolution/duration, audio plan, disclosure, estimated charge or infrastructure footprint, output paths, and QA gates.
- Require explicit approval. For hosted work require a monetary cap; for local work require approved disk/network/GPU scope and license acceptance.
- Default every supplied example to plan-only. Never silently enable postpaid billing, purchase credits/concurrency, download the roughly 372 GB current model repository, or contact a prompt-rewrite service.
- Persist provider request/job IDs or local revision hashes before long-running work. A submit timeout with no job ID is an unknown state; do not blindly resubmit because neither hosted surface documents an idempotency key.
Choose the correct family
| Family | Purpose | Key boundary |
|---|---|---|
| TokenHub HY-Video-1.5 | Managed text or image to 5 s video | Custom /v1/api/video/submit and /query; model hy-video-1.5; current action schema exposes only 720p |
Legacy vclm Hunyuan |
Existing Tencent Cloud integration | TC3-HMAC-SHA256, API version 2024-05-23, region ap-guangzhou, action names rather than a model field |
| HunyuanVideo-1.5 | Current lightweight local foundation model | 8.3B; 480p/720p base generation; separate SR to 720p/1080p; T2V/I2V |
| HunyuanVideo (original) | Legacy local 13B foundation model | Separate checkpoints/code; approximately 45 GB VRAM at 544×960×129f and 60 GB at 720×1280×129f in official tests |
| HunyuanVideo-Avatar | Audio-driven human animation | Separate repository/model; not an audio switch on HunyuanVideo-1.5 |
| HunyuanCustom | Subject customization and separate image/audio/video-conditioned workflows | Separate repository/checkpoints and license review |
| HunyuanVideo-Foley | Generate synchronized audio from video/text | Separate video-to-audio model; it does not generate picture |
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.
- 9d ago First seen · 393 lines · 77 tokens per session scan A 9210731cba55
tencent-hunyuanvideo is a skill published in the GitHub repository calesthio/generative-media-skills (170 stars, last pushed 2mo ago), licensed MIT. It adds 77 tokens to every session and 7,370 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 2 findings (makes network calls, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
artifact-deploy
One-click deploy a user's pre-built app/artifact into their OWN AWS account and get a global public HTTPS link (Vercel-like), with a default TTL and promote-to-persistent. Use when the user says "deploy this", "ship this demo", "give me a public link", "share this externally", or "deploy to AWS".
Tencent Cloud Infra
A command-line guide for managing Tencent Cloud services, including servers, networks, domains, certificates, storage, permissions, and monitoring. Tencent Cloud is a cloud provider, and tccli is its command-line tool.
tcapi
A helper for managing Tencent Cloud resources through Tencent Cloud's command-line API tool. Tencent Cloud is a cloud-services provider offering products such as virtual servers, storage, networks, and containers.
vigilante-issue-implementation-on-terraform
Implement a GitHub issue end-to-end when Vigilante dispatches work for a Terraform repository with fmt, validate, and secret-safe infrastructure guidance.
aws
Skill "aws" from dhaupin/vant, covering aws, when to use, what to do, 1. services and 2. cli.
vLLM
Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.