improve

A procedure for improving an existing system against a measurable target such as speed, size, latency, or score.

In plain words
What is it for?
Running measured optimization attempts, judging results, and preserving or rolling back changes when the metric does not improve.
Why use it?
It requires trustworthy measurements and controlled attempts so an apparent improvement is not just a misleading proxy result.

Skill for Claude CodeCodex

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 skills/newbie0413/simply/improve
Any agent
npx skills add NEWBIE0413/simply --skill improve
Clone the repo
git clone --depth 1 https://github.com/NEWBIE0413/simply

Made for: Claude Code, Codex.

Per session 52 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 734 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 $0.00052 $0.00734
Opus 5 $0.00026 $0.00367
Sonnet 5 $0.00010 $0.00147
Haiku 4.5 $0.00005 $0.00073

Measured yesterday against content hash d4ebe7668d82, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

improve 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.

pi-skills/improve/SKILL.md · 54 lines

How it starts

The opening of the file, as written. The whole thing — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.

improve - measurable improvement rail

Default: do not use it

Most work belongs in the standard ledger -> criteria -> verify loop. This rail adds repeated measurement plus Git snapshot and rollback operations, so enter it only when the user explicitly requests a measurable improvement and the metric is trustworthy or can be justified against the ledger outcome. A vague request to "make it better" requires outcome clarification, not this rail.

attempt_begin and attempt_judge may run git add, git commit, git reset --hard, and git clean -fd. Those operations are blocked by default. Enter the rail only when this run has an explicit --destructive-rails or SIMPLY_DESTRUCTIVE_RAILS=1 opt-in. --unsafe-workers controls worker permissions and does not grant this approval.

The governing risk is Goodhart's law: the agent optimizing a proxy will find its gaps without intending to. Measurement and optimization therefore remain separate. attempt_judge, not the agent's impression, decides whether an attempt is kept.

0. Classify the regime

Record the classification in the rail description before entering:

  • A - direct metric: a clear machine metric exists. Declare the rail.
  • B - human judgment: visual quality, aesthetics, or UX. Do not use this rail; ask for human visual QA at checkpoints.
  • C - justified proxy: a reasonable proxy exists. Explain in prose how it predicts the ledger outcome, then declare the rail.
  • D - unjustifiable: no defensible metric or proxy. Do not use the rail; escalate the ambiguity.

If the justification cannot be written clearly, the regime is D, not C.

1. Declare and freeze the rail

Use rail_declare:

  • metric_cmd must print one number on its final line. The baseline is measured at declaration.
  • holdout_cmds run only at rail_close. Do not run them during optimization or target their expected result; seeing the holdout early defeats it.
  • budget is the maximum attempt count and the loop's hard exit.
  • Existing criteria become invariants. Acceptance is a Pareto gate: the metric improves and every existing criterion remains satisfied.

Read the full file on GitHub · 54 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 · 54 lines · 52 tokens per session scan A d4ebe7668d82

Subscribe to this mod's changes

improve is a skill published in the GitHub repository NEWBIE0413/simply (1 stars, last pushed 10d ago), licensed MIT. It adds 52 tokens to every session and 734 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 2026-08-31.