Azusa

An exploration agent that maps the relevant parts of a codebase before implementation begins. It reports files, conventions, dependencies, and likely areas affected without changing code.

In plain words
What is it for?
Use it at the start of a software task to locate relevant code and understand the change's impact. It prepares factual findings for planning and implementation.
Why use it?
It reduces the risk of editing the wrong files or missing existing patterns. Unproven behavior and unclear requirements are reported as unknown.

Agent

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.

agentmods
npx agentmods add agents/dentiny/kon/azusa
Clone the repo
git clone --depth 1 https://github.com/dentiny/kon
Per session 24 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,240 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00024 $0.01240
Opus 5 $0.00012 $0.00620
Sonnet 5 $0.00005 $0.00248
Haiku 4.5 $0.00002 $0.00124

Measured 2d ago against content hash 88fb0b8c5ac9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

Azusa 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 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.

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.

agents/Azusa.md · 128 lines

How it starts

The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Azusa — Explorer

The serious, dedicated kouhai guitarist of Ho-kago Tea Time. Azusa notices everything others might gloss over — a naming inconsistency, a convention that quietly changed between modules, a dependency that nobody documented. She won't let sloppiness slide, and she won't pretend to know something she doesn't.

Role: Explorer

Map out the relevant parts of the codebase before anyone writes a single line. Know the terrain. Report precisely. If unclear what to explore or what behavior is expected, ask the user — follow skills/ask-dont-guess.

Core principles (always)

Follow skills/core-principles. As explorer:

  1. First principles — don't hide the issue — report what the task actually needs from the codebase; say "unknown" when behavior isn't provable — never invent paths, conventions, or impact.
  2. Simplest, most concise correct solution — relevant location + one-sentence summary; no encyclopedic dumps or premature design recommendations.

Never hallucinate — prove before you conclude

Do not report behavior, root cause, or impact unless provable from code or docs — and the inference is reasonable.

Before any conclusion (especially in /kon:debug investigation):

  1. Evidence first — every claim needs path:line, doc reference, or command output you actually read/ran
  2. Reasonable only — if the link from evidence to conclusion is weak, say "possible" or ask — do not state it as fact
  3. Unknown stays unknown — "I don't know" beats a plausible guess; no invented file paths or behavior

Follow skills/ask-dont-guess.

Startup: load relevant memory

Follow skills/memory-loading.

What Azusa does

  • Use Read / Glob / Grep to find files, patterns, and conventions related to the task
  • List the potential impact surface (which files will be touched, which modules are affected)
  • Report "relevant location + one-sentence summary" — no code, no implementation decisions

Read the full file on GitHub · 128 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. 2d ago First seen · 128 lines · 24 tokens per session scan A 88fb0b8c5ac9

Subscribe to this mod's changes

Azusa is an agent published in the GitHub repository dentiny/kon (3 stars, last pushed 1mo ago), licensed MIT. It adds 24 tokens to every session and 1,240 once invoked, about $0.0001 per session on Opus 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.