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 javded-itres/Holix --skill media-gengit clone --depth 1 https://github.com/javded-itres/HolixWrote 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/javded-itres/holix/media-gen)<a href="https://agentmods.dev/skills/javded-itres/holix/media-gen"><img src="https://agentmods.dev/badge/skills/javded-itres/holix/media-gen/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/javded-itres/holix/media-gen"><img src="https://agentmods.dev/badge/skills/javded-itres/holix/media-gen.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00026 | $0.00524 |
| Opus 5.5 | $0.00010 | $0.00210 |
| Sonnet 5.5 | $0.00005 | $0.00105 |
| Haiku 4.5 | $0.00003 | $0.00052 |
Grade A, and why
media-gen 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 16d 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
Media generation
When the user asks to нарисуй, сгенерируй картинку, imagine, make a video, сделай ролик, оживи фото, перерисуй, сделай из этих фото:
- Call
generate_imageorgenerate_videowith a clear prompt (language of the user). - Reference photos: if the user already sent images (this turn or earlier — paths in «Вложения из Telegram/MAX»), pass those paths in
references. Typical flow: they upload first, then say what to do. Do not ask them to send the files again. Do notread_filebinary photos. - The tool writes a file under workspace
media/and returns afile://Open link. In TUI, put that markdown link in the reply so the user can click it. - In TUI, never call
send_chat_files. The Open link is already in the transcript. - In Telegram/MAX, the file is sent when auto_send is on. If the tool result does not contain
Sent N file(s), callsend_chat_fileswith the saved path. - Never paste base64. Never claim the user received the file unless a messenger send was reported.
Hard rule: do not assemble video yourself
Video comes only from generate_video (the configured model). The tool returns immediately and keeps running in the background with no timeout; a 502 is retried, not a failure. In TUI, wait for the Open link. In Telegram/MAX, the file is sent for you. Do not:
- install or call
ffmpeg/moviepy/opencvto stitch frames - turn a sequence of
generate_imagestills into an mp4 - write a Python/shell script that encodes video
- fake a clip from screenshots
Stills are generate_image. Motion is generate_video with the same prompt and references if the user sent photos.
Leave provider empty. Each model has a note in holix-media settings. A plain generate uses the text-only model. Pass references when the user attached a picture or asks to change, add, or interact with one; the tool then picks the model that accepts a reference.
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.
- 16d ago First seen · 29 lines · 26 tokens per session scan A 9b6795ab670c
media-gen is a skill published in the GitHub repository javded-itres/Holix (17 stars, last pushed today), licensed MIT. It adds 26 tokens to every session and 524 once invoked, about $0.0001 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 2026-10-01.
Other skills, from other repositories
mnemosyne
Persistent cross-session memory via Mnemosyne — store, recall, and consolidate facts, preferences, and context.
mnemosyne-memory-override
Hard rule override that forces Mnemosyne for all durable memory storage. The legacy memory tool is DEPRECATED for user preferences, credentials, and project conventions. Use memory ONLY for ephemeral session state.
hermes-tweet
Use Xquik in Hermes Agent for public X research, monitoring, thread summaries, creator discovery & approved actions. Not affiliated with X Corp. Use when the user requests X data or a named action. Trigger with "search X", "monitor X", "post tweet", or "X trends".
vrc-monitor-agent
Use for VRChat friend queries (online status, who played with whom, activity timelines, online patterns) and VRChat social actions (boop, invite, join world, friend management, group operations, image uploads) via the vrc-monitor MCP server on port 8799.
review-workflow
A review workflow for GitHub pull requests and issues, including judging whether changes can be merged or an issue can be closed. It requires read-only review, REST API access, and testing changes in practice.
vrchat-social-queries
VRChat social queries: online, companions, patterns, boops.