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 agentmods add skills/hec-ovi/observer-skill/observernpx skills add hec-ovi/observer-skill --skill observergit clone --depth 1 https://github.com/hec-ovi/observer-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/hec-ovi/observer-skill/observer)<a href="https://agentmods.dev/skills/hec-ovi/observer-skill/observer"><img src="https://agentmods.dev/badge/skills/hec-ovi/observer-skill/observer.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 | $0.00110 | $0.01584 |
| Opus 5 | $0.00055 | $0.00792 |
| Sonnet 5 | $0.00022 | $0.00317 |
| Haiku 4.5 | $0.00011 | $0.00158 |
Grade A, and why
observer 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 3d 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.
This is a copy
100% identical to observer — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.
observer
The user is about to watch something hard. Your job runs in two halves: prepare everything before they press play, then answer instantly while they watch. The whole point of the first half is that the second half has no waiting in it.
Every tool below is mcp__observer__<name>. Each result tells you the phase you are in, so
you never have to guess where you are.
Start
| The user | You |
|---|---|
| pastes a link and wants it explained | open with the url |
| already has the page open and gives you a session id | status with that id |
| says "wait, what did he mean there" mid-session | you are already in the loop; see Session |
open takes hasAds, extraKnowledge, and toolkit. If the user did not say, ask in one
short line before opening: does the video carry sponsor reads, do you want research, do you
want visuals. Then open and do not ask again.
If they said what they want from the video ("I only care about the training setup"), pass it
as userPrompt. It outranks everything you would otherwise decide.
Prepare
1. Transcribing. Call status until the phase moves on. Nothing else to do; do not
narrate the wait. A three-hour podcast with captions takes seconds, without captions it
takes minutes and the progress is real.
2. Read all of it. transcript in pages until you have the whole thing. Read it
before you write anything. When status reports source.hasAds, read
references/ads.md first: it is how to tell a sponsor read from the
talk, and it holds for every step after this one. Then read
references/study-plan.md and write the concept list with
concepts.
3. Research, only when settings.extraKnowledge is on. Read
references/research.md. Use your own search tools, and attach what
you find with note. Look specifically for what moved after the recording date: that is
where you add value the video cannot.
4. Visuals, only when settings.toolkit is on. Read
references/visual-plan.md to decide what earns a picture, then
references/artifacts.md for how to write one. build compiles
it, runs it in the real page, and hands you back either line-accurate errors or a PNG path.
Read the PNG. If it looks wrong, fix it and build again with the same id. link binds it
to its concept.
What ships with it
8 files 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.
- 3d ago First seen · 133 lines · 110 tokens per session scan A ab0ccabb6034
observer is a skill published in the GitHub repository hec-ovi/observer-skill (0 stars, last pushed 29d ago), licensed MIT. It adds 110 tokens to every session and 1,584 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to observer, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
gather-screenshots
Generate screenshot capture checklists for Product Demo lessons based on readiness assessment.
consulting-deck
Create, revise, and audit executive-ready strategy-consulting presentations as native, editable PowerPoint files. Use when turning topics, text, Markdown, Word, PDF, spreadsheets, CSV data, URLs, research, or an existing PPTX into a board, client, strategy, transformation, market-analysis, or McKinsey-style consulting…
baoyu-comic
Knowledge comic creator supporting multiple art styles and tones. Creates original educational comics with detailed panel layouts and batch-capable image generation. Use when user asks to create "知识漫画", "教育漫画", "biography comic", "tutorial comic", or "Logicomix-style comic".
blog-post
Author a new entry for the Semiotic blog. Use this skill whenever the user asks for a blog post, release summary, chart explainer, or case study to publish at /blog/SLUG.
chatgpt-app-submission
Inspect a ChatGPT Apps MCP server codebase and generate chatgpt-app-submission.json with app info suggestions, tool hint justifications, test cases, and negative test cases, then report review-check findings and outputSchema warnings for submission review.
semiotic-charts
Generate trustworthy, accessible, agent-legible charts with Semiotic. Use whenever you produce a data visualization in a React/TypeScript app, stream a chart as generative UI, or pick a chart type for a dataset. Emit a validated config and run the trust loop — never hand-write chart JSX that breaks on first paint.