kaizen

A session-end command that records a small number of evidence-based lessons about how work was done in a permanent GitHub issue. It does not change project code or documentation.

In plain words
What is it for?
Use it after a coding session to capture workflow problems, discoveries, or improvements for later human review.
Why use it?
It preserves useful process lessons without turning every observation into an immediate code or policy change.

Command for Claude Code

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 commands/littlebearapps/untether/kaizen
Clone the repo
git clone --depth 1 https://github.com/littlebearapps/untether

Made for: Claude Code.

Per session 66 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,662 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 $0.00066 $0.01662
Opus 5 $0.00033 $0.00831
Sonnet 5 $0.00013 $0.00332
Haiku 4.5 $0.00007 $0.00166

Measured yesterday against content hash 0ea539f05525, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

kaizen 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 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.

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.

.claude/commands/kaizen.md · 130 lines

How it starts

The opening of the file, as written. The whole thing — 130 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are handling /kaizen. Always begin your output with the literal heading ## /kaizen (the Stop-hook and future runs detect a completed capture by that marker — do not omit or reword it).

/kaizen captures process learnings — how the work went, not what the code does — as durable bullets on a single permanent GitHub collector issue. It is the cheap, high-frequency half; /kaizen-review is the weekly, human-gated promotion half.

User input: $ARGUMENTS

Untether adaptations (read first)

Load .claude/rules/kaizen.md (the thin capture rule) and, for the full rubric, docs/kaizen/README.md. Also .claude/rules/workflow-commands.md for the cross-cutting rules. Key points:

  • Authority: read-only except ONE gh issue comment. Never edit code, .claude/rules/*, hooks.json, or CLAUDE.md. Promotion to any of those is /kaizen-review's job, and even then only as a proposal.
  • Self-invokes at session end (a Stop-hook nudge is proposed to Nathan for wiring, but is not required — this command runs itself). Under Untether/cron it still self-invokes; the single collector comment is the one sanctioned autonomous external write in the suite (bounded, idempotent, one comment).
  • Idempotent. Re-running must not duplicate a bullet already on the collector this session. Resolve the collector by exact title with --limit 200.
  • Redaction. Scrub tokens/keys/env/chat-content/fleet identifiers from every evidence ref before it lands in the comment.

What counts as a capture

A capture is a process learning with evidence, in one of these tag buckets (Untether-tuned):

[timing] [tooling] [issue-quality] [novel-pattern] [guardrail-block] [engine-quirk] [cost] [meta]

Every bullet MUST carry:

  1. a tag from the list above,
  2. an evidence ref — a trace, a log signature, a file:line, an issue #N, a PR #N, or an event signature (never vague "I noticed…"),
  3. an S/C/R score — Severity S0–S4 / Confidence low|med|high / Recurrence ×N (see docs/kaizen/README.md for the scales).

Read the full file on GitHub · 130 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 First seen · 130 lines · 66 tokens per session scan A 0ea539f05525

Subscribe to this mod's changes

kaizen is a command published in the GitHub repository littlebearapps/untether (66 stars, last pushed 2d ago), licensed MIT. It adds 66 tokens to every session and 1,662 once invoked, about $0.0003 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 2026-08-30.