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 epicsagas/Velith --skill book-technicalgit clone --depth 1 https://github.com/epicsagas/VelithWrote 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/epicsagas/velith/book-technical)<a href="https://agentmods.dev/skills/epicsagas/velith/book-technical"><img src="https://agentmods.dev/badge/skills/epicsagas/velith/book-technical/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/epicsagas/velith/book-technical"><img src="https://agentmods.dev/badge/skills/epicsagas/velith/book-technical.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00049 | $0.01268 |
| Opus 5.5 | $0.00020 | $0.00507 |
| Sonnet 5.5 | $0.00010 | $0.00254 |
| Haiku 4.5 | $0.00005 | $0.00127 |
Grade A, and why
book-technical 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 1mo 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 — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Technical
The reader of a technical book is trying to become someone who can do the thing. Every chapter must move them closer to doing it themselves, and the surest sign of a book that fails is a reader who can follow along but cannot deviate.
Structure
Default: Intro (10-15%) → Foundations (25%) → Practice (30%) → Advanced (20%) → Reference (10%). Within that:
- Running project. The reader builds one thing across the book. Each chapter adds a capability. By the end they have something that works and that they understand every line of. Books without a running project produce readers who have read about the thing.
- Concept gradient: each concept is introduced once, at the first chapter that needs it, with the minimum needed then, and deepened later where it becomes necessary. The outline's dependency DAG enforces this; the architect verifies there are no forward references.
- Difficulty is monotonic non-decreasing within a part; the first chapter of a part can step back slightly.
- Reference material (API tables, option lists) goes to the back or an appendix, never inside a teaching chapter.
Failure-first teaching
The single largest gap between AI technical writing and good human technical writing: humans show the thing going wrong.
- Show the naive version first. Run it. Show the actual error message or wrong output.
- Explain why it fails in terms of the mental model, not just the fix.
- Then the correct version. Then what else would have broken it.
- Debugging is a chapter topic, not an afterthought. How would the reader know something is wrong? What would they look at first?
- Performance, security, and edge cases appear where the reader would first hit them, not in a separate "advanced considerations" chapter.
Code standards
- Every code block has a language tag and, when it is a file, a filename comment on line 1.
- Runnable, or explicitly labeled
# pseudocode. Imports included. Expected output shown after. - Lengths: 10-30 lines when introducing; 30-80 when building; 80+ only for a complete listing, and then with the key section explained separately.
- Version pinned in
PRD.mdand stated once in the introduction; not repeated per chapter. Breaking changes and security-relevant versions are called out inline. - Prefer "save the file" to "click File → Save"; UI paths change, concepts do not.
- The fact-checker runs code blocks where the environment allows. Write code that can be run.
- Diagrams: Mermaid for flow/sequence, ASCII fallback for EPUB readers that do not render Mermaid, tables for comparisons. At most three per chapter.
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.
- 1mo ago Changed · +54 lines · +24 tokens per session 758a43607869
- 1mo ago First seen · 21 lines · 25 tokens per session scan A bbf767c3c325
book-technical is a skill published in the GitHub repository epicsagas/Velith (39 stars, last pushed 28d ago), licensed Apache-2.0. It adds 49 tokens to every session and 1,268 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
paper-explainer
A research-content workflow that turns an arXiv paper or a local PDF into an explanation for a video or written article. arXiv is a public online archive of research papers.
intro-example
Minimal MDA source demonstrating relationship-graph footnotes and metadata.mda. MDA-extended fields. Use as a reference fixture when learning the MDA source format.
ai-native-software-dev
Use when the user asks to build, change, modernize, or reconstruct non-trivial software with an AI-native workflow, including greenfield development, work in an existing codebase, or authorized reference-based product reconstruction; also use when lifecycle planning, architecture, implementation, testing, review…
learn-anything
This skill should be used when the user wants a beginner-to-expert learning path, curriculum, study plan, or training progression for any topic — academic, physical, creative, linguistic, or professional — built by researching how that domain is optimally taught, then emitting only the first progression tier through a…
capture-learning
Use when a real outcome just landed and the lesson should be banked so the OS stops repeating mistakes - an eval result, a customer reply, a metric move, a failed or won launch, a Share-of-Model move - when the founder says "we just learned", "that worked", "that failed", "log this", "note this for next time"…
p13d-intuition-training
A training framework for turning intuition into compact mental models and pattern recognition. Intuition is a quick sense about a situation formed from experience or learned patterns.