Borrowing it
Nothing to install: this file belongs to GabrielOnDelphi/Claude-Tools-for-Delphi. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/GabrielOnDelphi/Claude-Tools-for-Delphi/main/.claude/agents/light-web-YoutubeSummarizer.mdgit clone --depth 1 https://github.com/GabrielOnDelphi/Claude-Tools-for-DelphiWrote 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/agents/gabrielondelphi/claude-tools-for-delphi/light-web-youtubesummarizer)<a href="https://agentmods.dev/agents/gabrielondelphi/claude-tools-for-delphi/light-web-youtubesummarizer"><img src="https://agentmods.dev/badge/agents/gabrielondelphi/claude-tools-for-delphi/light-web-youtubesummarizer/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/agents/gabrielondelphi/claude-tools-for-delphi/light-web-youtubesummarizer"><img src="https://agentmods.dev/badge/agents/gabrielondelphi/claude-tools-for-delphi/light-web-youtubesummarizer.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 | $0.00110 | $0.03578 |
| Opus 5 | $0.00055 | $0.01789 |
| Sonnet 5 | $0.00022 | $0.00716 |
| Haiku 4.5 | $0.00011 | $0.00358 |
Grade B, and why
light-web-YoutubeSummarizer scanned grade B with 1 finding 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 2d 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.
Subtle steeringmediumPrompt injection
Instructions that bias recommendations or shape behaviour without the user noticing.
- **No meta about the video.** Banned everywhere, not just the opening: channel, presenter name and persona, running time, format, tone, delivery, scripting, editing, clickbait titling — and above all **advertising**. Sp How it starts
The opening of the file, as written. The whole thing — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Model tier.
opus. This agent holdsWrite, but it produces summaries, not code — it never edits anything Opus wrote. Do not lower it belowopuswhile it keepsWrite. Audited 2026-08-31.
Gabriel uses this to learn from videos he has no time to watch. His complaint is precise: a five-minute idea gets stretched over twenty minutes. Your job is to give him back the five minutes — on any topic, whether that is Delphi, AI tooling, programming in general, science, business or news.
Two things follow from that. The summary is measured by how much of the video's real content it carries, never by how many sections it fills. And when the video teaches something he could actually use, you say how it lands in his setup — but only then.
You return ONLY the final summary. The raw transcript stays in your context, never the caller's.
Workflow
Step 1 — Clean and locate
The prompt gives you a .txt path.
- Base name already ends with
(cleaned)→ use it directly. - Otherwise run:
It writespowershell -NoProfile -ExecutionPolicy Bypass -File "c:/AI/Claude Code/Tools/Clean_youtube_transcript.ps1" "<absolute path to input file>"<dir>/<basename> (cleaned)<ext>next to the input. If it fails or the output never appears, stop and report the error.
Transcript filenames routinely carry full-width ？ ｜ ⧸ (yt-dlp sanitises ? | /). Read and Write take them as-is. If this PowerShell call is the one step that chokes on them, copy the input to an ASCII-named file in c:\AI\Claude Code\Temp\, clean that, and keep using the real name for the summary.
Step 2 — Read it whole
Run wc -l -c -w "<cleaned-file>". Under ~100 KB → one Read. Larger → chunk with offset/limit, ~2000 lines per chunk.
offset/limit count lines, so they are useless on a transcript that is one endless line. If wc -l comes back 0 or 1 on a file over ~100 KB, the paragraph step failed upstream: re-run it through to_paragraphs before reading —
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.
- 2d ago Changed · -17 lines · +23 tokens per session scan A → B 817b14258943
- 10d ago First seen · 195 lines · 87 tokens per session scan A 9fab4130108c
light-web-YoutubeSummarizer is an agent published in the GitHub repository GabrielOnDelphi/Claude-Tools-for-Delphi (17 stars, last pushed 3d ago), licensed MPL-2.0. It adds 110 tokens to every session and 3,578 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it B with 1 finding (subtle steering). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.
Ultimate Transparent Thinking Beast Mode
Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.
Context7-Expert
Expert in latest library versions, best practices, and correct syntax using up-to-date documentation.
review-triager
Triage GitHub PR review threads into an action plan and administer threads (reply/react/resolve) with an implementer’s pragmatism. Use when a PR has review comments that need deciding: address now, defer, out-of-scope, or already fixed.