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 instructions/teunlao/vidin/agents-mdgit clone --depth 1 https://github.com/teunlao/vidinWrote 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/instructions/teunlao/vidin/agents-md)<a href="https://agentmods.dev/instructions/teunlao/vidin/agents-md"><img src="https://agentmods.dev/badge/instructions/teunlao/vidin/agents-md.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 | $0.01075 | $0.01075 |
| Opus 5 | $0.00537 | $0.00537 |
| Sonnet 5 | $0.00215 | $0.00215 |
| Haiku 4.5 | $0.00108 | $0.00108 |
Grade A, and why
vidin AGENTS.md 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.
How it starts
The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
vidin turns a screen recording into a small set of annotated keyframes an agent can read, exposed as a CLI and an MCP server. The design rationale — why frames are selected on events instead of sliced on an interval — is in README.md; read it before touching the selector.
Commands
bun test # full suite, ~17s; needs ffmpeg, builds the fixture on first run
bunx tsc --noEmit # typecheck (development runs .ts directly, no build)
bun run build # tsc → dist/ — what npm ships; runs on plain Node ≥ 20
bun run vidin <file> # CLI: analyse a video into ./.vidin/<name>/
bun run mcp # MCP server on stdio
bun run fixture # regenerate fixtures/screen-bug.mp4 (synthetic, scripted timeline)
Toolchain: Bun ≥ 1.1 (development only) and ffmpeg/ffprobe on PATH.
yt-dlp is needed only for URL input. The published package is
@teunlao/vidin (plain vidin is blocked by npm's similarity check against
vision); it targets Node ≥ 20 via npx, and end users never need Bun.
Hard rules
- No Bun APIs in
src/orbin/. The npm package runs under whatevernpxresolves to, so runtime code usesnode:APIs exclusively. Bun (which implements them) is the dev runner;Bun.*is allowed only inscripts/andtest/. - No image libraries. All drawing (change boxes, labels, contact sheets) is
ffmpeg filters in
src/core/render.ts. Do not add sharp/canvas/jimp — keeping ffmpeg the only native dependency is a design goal, not an accident. - Runtime dependencies stay at two:
@modelcontextprotocol/sdkandzod. Adding one is a design decision to raise, not a convenience. - Imports carry explicit
.tsextensions; ESM only.tsconfig.build.jsonrewrites them to.jswhen emittingdist/.
Selector invariants (src/core/select.ts)
Breaking either one is a correctness bug, not a tuning choice;
test/regressions.test.ts pins them:
- The diff baseline advances only by emitting. No code path may fold a changed screen into the reference without keeping a frame — every later diff would then measure zero against a state the reader never sees.
- Annotations describe the surviving neighbour. After budget pruning, each
frame's deltaPct and regions are recomputed (
relink) against the frame that actually precedes it in the final list.
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 · 92 lines · 1,075 tokens per session scan A f762a7b3994f
vidin AGENTS.md is an instructions file published in the GitHub repository teunlao/vidin (2 stars, last pushed 1mo ago), licensed MIT. It adds 1,075 tokens to every session, about $0.0054 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 2026-08-31.
Other instructions, from other repositories
Orkas-VideoStudio AGENTS.md
Instructions for Orkas-AI/Orkas-VideoStudio, covering working with orkasvideostudio (for coding agents), skills and capabilities.
agents-remember AGENTS.md
Instructions for Foxfire1st/agents-remember, covering agents remember source checkout instructions, start here — route by role, memory and onboarding, memory retrieval strategies and source layout.
odds-api AGENTS.md
Instructions for odds-api/odds-api, covering agent instructions for odds-api/odds-api, preferred integration paths, canonical files, runtime and repository and release safety.
next-devtools-mcp CLAUDE.md
Claude Code instructions for vercel/next-devtools-mcp, covering claude.md, project overview, build and development commands, install dependencies and watch mode for development.
TestTrout AGENTS.md
AGENTS.md instructions for DrDroidLab/TestTrout, covering notes for coding agents, the short version, rules and working on the tool itself.
ContextFlux AGENTS.md
AGENTS.md instructions for divyanshu-iitian/ContextFlux, a project described as: Task-adaptive repository context engine for coding agents: local MCP, exact token budgets, and dependency-aware retrieval.