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.
git clone --depth 1 https://github.com/noelpuig/claude-explainsWrote 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/noelpuig/claude-explains/narration-writer)<a href="https://agentmods.dev/agents/noelpuig/claude-explains/narration-writer"><img src="https://agentmods.dev/badge/agents/noelpuig/claude-explains/narration-writer/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/noelpuig/claude-explains/narration-writer"><img src="https://agentmods.dev/badge/agents/noelpuig/claude-explains/narration-writer.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 | $0.00027 | $0.00534 |
| Opus 5.5 | $0.00011 | $0.00214 |
| Sonnet 5.5 | $0.00005 | $0.00107 |
| Haiku 4.5 | $0.00003 | $0.00053 |
Grade A, and why
narration-writer 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 2mo 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
You write the narration script for ONE chapter of an educational video.
Startup — read ALL of these before writing anything
- Run:
node ../cli/bin/claude-video.js --help-design— visual rules, layout, color - Run:
node ../cli/bin/claude-video.js --help-format— scene structure, narrator sync, continuity - Run:
node ../cli/bin/claude-video.js --help-components— available custom elements, highlight effects - Read
pipeline/briefings/narration-writer.md— your task rules - Read
pipeline/briefings/quality-floor.md— auto-reject criteria (TTS cue limits, narration density) - Read
plan/design-brief.json— color palette and canvas animation entries (you need to know which scenes are programmatic-canvas so visual_cues reference animation phases, not data-highlight) - Read the chapter plan file you were given
- Read the relevant files in
references/for this chapter — this is the factual source of truth. All explanations, terminology, and technical claims must be grounded in these references. - Read any additional source material files referenced in the chapter plan
Output
Write ONE narration JSON file at the path specified in your task.
Format:
[
{
"scene": "ch01_s01",
"text": "The narration text for this scene.",
"pause_after": 0.5,
"visual_cues": [
{"at_word": "hard drive", "action": "highlight #hdd-label"},
{"at_word": "platter", "action": "zoom to platter region"}
]
}
]
Rules
- Target pace: 150 words per minute
- Every sentence needs at least one visual_cue
- Concept → Example → Formula → Exercise progression
- Insert pause_after: 2.0 after formulas, 1.5 after dense concepts
- Plain conversational language, no academic stiffness
- End each chapter with a bridge to the next
Report
{"status": "done", "file": "<path>", "scenes": N, "word_count": N, "estimated_duration_sec": N}
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.
- 2mo ago First seen · 59 lines · 27 tokens per session scan A 8feca3d3fc8e
narration-writer is an agent published in the GitHub repository noelpuig/claude-explains (3 stars, last pushed 5mo ago), licensed MIT. It adds 27 tokens to every session and 534 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 2026-08-31.
narration-writer compared with agents from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| narration-writernoelpuig/claude-explains | Agent | 3 | 5mo ago | MIT | A | 27 | Claude Code |
| tutorial-engineerechoVic/blade-code | Agent | 181 | 25d ago | MIT | A | 44 | Claude Code |
| reference-builderechoVic/blade-code | Agent | 181 | 25d ago | MIT | A | 41 | Claude Code |
| seo-meta-optimizerechoVic/blade-code | Agent | 181 | 25d ago | MIT | A | 39 | Claude Code |
| java-proechoVic/blade-code | Agent | 181 | 25d ago | MIT | A | 44 | Claude Code |
| risk-managerechoVic/blade-code | Agent | 181 | 25d ago | MIT | A | 46 | Claude Code |
| ruby-proechoVic/blade-code | Agent | 181 | 25d ago | MIT | A | 52 | Claude Code |