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 navikt/copilot --skill deliberate-ai-usegit clone --depth 1 https://github.com/navikt/copilotWrote 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/navikt/copilot/deliberate-ai-use)<a href="https://agentmods.dev/skills/navikt/copilot/deliberate-ai-use"><img src="https://agentmods.dev/badge/skills/navikt/copilot/deliberate-ai-use/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/navikt/copilot/deliberate-ai-use"><img src="https://agentmods.dev/badge/skills/navikt/copilot/deliberate-ai-use.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 | $0.00084 | $0.00946 |
| Opus 5.5 | $0.00034 | $0.00378 |
| Sonnet 5.5 | $0.00017 | $0.00189 |
| Haiku 4.5 | $0.00008 | $0.00095 |
Grade A, and why
deliberate-ai-use 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 — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Deliberate AI Use
The always-on rules (explain choices, mark red-zone code, ask before delegating fully) are in deliberate-ai-use.instructions.md. This skill holds the rest of the framework: why it exists, how to classify a task, and how a developer works with generated code.
Why
Research shows that how you use AI matters more than whether you use it. In Anthropic's 2026 study, developers using AI assistance averaged 50 % on a knowledge quiz, against 67 % for those who coded by hand. Those who delegated everything scored under 40 %. Those who asked questions to understand the code scored 65 % or more (Anthropic).
The framework makes sure AI tools strengthen developers' skills instead of weakening them.
Sources:
- Anthropic: How AI assistance impacts coding skills (2026)
- INNOQ: AI Coding Patterns Through Cognitive Load Theory (2026)
- METR: AI experienced OS dev study (2025)
- Stray et al.: Developer Productivity With and Without GitHub Copilot (HICSS-59, 2026), a study of Nav IT
- MIT/Microsoft: The Effects of Generative AI on High-Skilled Work (2025)
- Nav developer survey 2026: 59 % are worried about losing skills
Green and red zone
Classify a task before using AI on it.
🟢 Green zone: suited to AI
Tasks where AI adds the most value without weakening understanding:
- Boilerplate and repetitive code (Nais manifest, CRUD endpoints, Dockerfile)
- Technology the developer already knows well
- Configuration and infrastructure
- Refactoring with a known goal (rename, extract, move)
- Test data and fixtures
🔴 Red zone: code by hand first
What ships with it
1 file 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 Changed 4e52ba35ebcc
- 5d ago Changed 60f1bc5b6eb0
- 17d ago Changed 4927691141f3
- 18d ago First seen · 73 lines · 84 tokens per session scan A 5a2a6c24a50c
deliberate-ai-use is a skill published in the GitHub repository navikt/copilot (55 stars, last pushed today), licensed MIT. It adds 84 tokens to every session and 946 once invoked, about $0.0003 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 2026-09-28.
Other skills, from other repositories
extract-wisdom
Extract wisdom, insights, and actionable takeaways from YouTube videos, blog posts, articles, or text files. Use when asked to extract wisdom or key insights from a given content source.
ml-llm-wiki
Use when answering questions from this machine-learning knowledge base. Triggers: questions about transformers, attention cost and efficiency, and long-context scaling; 'what do we know about attention', 'check the ML wiki'. Read-only querying of compiled knowledge; to add, update, supersede, lint, audit, or critique…
appraise-gemstone
Appraise gemstone value using the four Cs (color, clarity, cut, carat), origin assessment, treatment detection, and market factor analysis. Advisory educational guidance only — not a certified appraisal. Use when understanding factors that determine a gemstone's value, pre-screening stones before a professional…
assess-holistic-health
Conduct temperament-based health assessment from Hildegard von Bingen's Causae et Curae. Evaluates the four temperaments (sanguine, choleric, melancholic, phlegmatic), elemental correspondences (air, fire, earth, water), and provides dietary and lifestyle recommendations for rebalancing. Use when understanding…
explain
Explain a concept, system or piece of code as a diagram-first terminal answer, or as an interactive HTML page.
learning-opportunities
Facilitates deliberate skill development during AI-assisted coding. Offers interactive learning exercises after architectural work (new files, schema changes, refactors). Use when completing features, making design decisions, or when user asks to understand code better. Triggers on "learning exercise", "help me…