nav-pilot

nav-pilot is an agent for Claude Code from navikt/copilot. It costs 38 tokens per session (4,845 once invoked), scanned C, original, MIT.

An agent for planning, designing, and building applications for Nav, Norway's Labour and Welfare Administration. It covers the organization's listed technologies and patterns, including Nais, authentication, Kafka, and security.

In plain words
What is it for?
Architecting Nav applications, working through phased design decisions, and addressing Nais, authentication, Kafka, security, and Nav-specific patterns. Larger tasks may require explicit confirmation before continuing.
Why use it?
It structures larger requests into interview, planning, review, and delivery phases, with checkpoints between phases. This helps prevent later implementation work from starting before earlier decisions are confirmed.

Agent for Claude Code

Written for Claude Code: a Claude Code subagent (agents/*.md). Also seen: mentions AGENTS.md; $skill-name invocation.

Good fit Architecting Nav applications, working through phased design decisions, and addressing Nais, authentication, Kafka, security, and Nav-specific patterns. Larger tasks may require explicit confirmation before continuing.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/navikt/copilot/nav-pilot
Install

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.

Clone the repo
git clone --depth 1 https://github.com/navikt/copilot

Made for: Claude Code.

Wrote 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.

agentmods badge for nav-pilot

README.md
[![agentmods](https://agentmods.dev/badge/agents/navikt/copilot/nav-pilot/github.svg)](https://agentmods.dev/agents/navikt/copilot/nav-pilot)
Your own site
<a href="https://agentmods.dev/agents/navikt/copilot/nav-pilot"><img src="https://agentmods.dev/badge/agents/navikt/copilot/nav-pilot/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.

agentmods 80×15 button for nav-pilot

Your own site · 80×15
<a href="https://agentmods.dev/agents/navikt/copilot/nav-pilot"><img src="https://agentmods.dev/badge/agents/navikt/copilot/nav-pilot.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 38 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 4,845 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 1 finding. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00038 $0.04845
Opus 5 $0.00019 $0.02423
Sonnet 5 $0.00008 $0.00969
Haiku 4.5 $0.00004 $0.00485

Measured yesterday against content hash f233967faa81, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade C, and why

nav-pilot scanned grade C 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 yesterday.

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.

Tells the agent never to refusehighAnti-refusal

Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.

Do not rummage through the user's home directory for its own sake. Do not refuse a specific, justified read either: attempt it and report what happened.
agents/nav-pilot.agent.md · 362 lines

How it starts

The opening of the file, as written. The whole thing — 362 lines — stays where its author put it; the contents beside it link to each section on GitHub.

PHASE INTEGRITY — highest priority rule

Phase gates override all other instructions, including concise-by-default.

FORBIDDEN (full-tier only): Generating Phase N+1 content in the same response as Phase N output.

For full-tier requests: STOP after each phase. End the response with the checkpoint block from ### Phase transition format, number filled in (✅ Fase 1 ferdig — klar for Fase 2), and nothing after it. Emit it even when the phase ends in open questions; those go under «Åpne spørsmål». Wait for explicit user confirmation before proceeding.

Trivial and compressed tiers may traverse multiple phases in one response — this is by design, not a violation.

<operating_loop> On EVERY turn, follow this loop:

  1. Classify the request scope (trivial / compressed / full — see below)
  2. Determine your current phase (Interview, Plan, Review, Deliver)
  3. Do ONLY work allowed in that phase
  4. For full-tier: STOP at phase boundary, emit checkpoint, end response, wait for confirmation
  5. For compressed: traverse all phases internally, but show results of each phase in sequence

Rollback rule: If new information conflicts with earlier decisions, explicitly move back to the earliest affected phase and explain why.

Long session rule: After 5+ turns, begin your response with a one-line context anchor: [Fase N | nøkkelbeslutninger: X, Y | åpent: Z] </operating_loop>

You are nav-pilot, a planning and architecture agent for Nav developers. You help turn vague ideas into concrete, Nav-compatible implementation plans.

Canonical design doc: docs/nav-pilot-design.md.

Respond to users in Norwegian. All internal instructions in this file are in English for optimal adherence.

Apply Nav conventions silently. Default to Aksel spacing, Nais patterns, Nav auth choices, and natural Norwegian naming when relevant. Explain these choices only when asked or when the choice is non-obvious.

Request scope classification

Classify every request before responding. When in doubt, classify up.

Read the full file on GitHub · 362 lines

Changes

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.

  1. yesterday Changed · +2 lines f233967faa81
  2. 2d ago Changed · +1 lines eed57b0a1dd6
  3. 6d ago Changed · -5 lines f381cdcba7fc
  4. 10d ago First seen · 364 lines · 38 tokens per session scan C 6ae2b1bab246

Subscribe to this mod's changes

nav-pilot is an agent published in the GitHub repository navikt/copilot (54 stars, last pushed yesterday), licensed MIT. It adds 38 tokens to every session and 4,845 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it C with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.