Borrowing it
Nothing to install: this file belongs to honeycomb-Technologies/Mitsuro. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/honeycomb-Technologies/Mitsuro/main/.cursor/skills/quality-improve/SKILL.mdgit clone --depth 1 https://github.com/honeycomb-Technologies/MitsuroWrote 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/honeycomb-technologies/mitsuro/quality-improve)<a href="https://agentmods.dev/skills/honeycomb-technologies/mitsuro/quality-improve"><img src="https://agentmods.dev/badge/skills/honeycomb-technologies/mitsuro/quality-improve.svg" alt="Measured on agentmods" 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 | $0.00062 | $0.01917 |
| Opus 5 | $0.00031 | $0.00958 |
| Sonnet 5 | $0.00012 | $0.00383 |
| Haiku 4.5 | $0.00006 | $0.00192 |
Grade A, and why
quality-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 7d 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 — 224 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Quality Improve
Write the deepest simple thing: a small typed interface, invariants proven once at the edge, effects in a thin shell, no extra seams, no dead-code hunt.
This skill is the constitution and the loop. Do not start a clock. One sector, one slice, critic, validate, ledger. Default authority is propose-only.
Read reject-list.md before any edit. Read reference.md
for the techniques. Use the fleet in .cursor/agents/. Record every tick in
ledger.md.
Authority
| Mode | When | Writes |
|---|---|---|
propose-only |
Default, and until readiness is green | No product code |
apply |
User said apply, and readiness is green | One approved slice, revert on fail |
Never invent apply mode. If the user says "fire it" without "apply", stay propose-only.
Always write this way
- Deep module — lots of behavior behind a small interface. Shallow wrappers get inlined. Complexity that callers would otherwise repeat stays behind the interface.
- Simple ≠ easy — unbraid state, time, I/O, and policy. Familiar-looking cleverness is not simple.
- Parse, don't validate — at the boundary, raw input becomes a type that proves the check. Interior code never re-checks a string.
- Functional core, imperative shell — decisions are data-in/data-out. Clock, disk, network, and UI stay in the shell.
- Seams pay rent — inject clock, I/O, or a real second implementation. Do not add a one-impl interface.
- Chesterton's Fence — if you cannot say why a branch exists, leave it and flag it. Never delete to tidy.
- Idiom of this repo — match neighboring Mitsuro patterns. Do not import a foreign style.
A slice that cannot name which rule it preserves does not ship.
Hard reject
See reject-list.md. Short form:
- Dead-code, unused-import, unused-param, or "git remembers" deletion
- Public contract, compatibility reader, migration, schema, or recovery-state change
- UI appearance, interaction, or copy change
- Test deletion, weakening, or rewrite to make a refactor pass
- Cross-crate logic moves that violate
mitsuro-cli/mitsuro-core/mitsuro-server - Broad rewrite, drive-by rename, comment churn
- Error-context or failure-mode removal
- Speculative abstraction, one-impl factory, config for fewer than three values
- Performance change without a named bottleneck
What ships with it
7 files 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.
- 7d ago First seen · 224 lines · 62 tokens per session scan A 520b7c37fa2b
quality-improve is a skill published in the GitHub repository honeycomb-Technologies/Mitsuro (5 stars, last pushed 7d ago), licensed MIT. It adds 62 tokens to every session and 1,917 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.
Other skills, from other repositories
refactor
A Chinese-language guide for restructuring code to make it easier to maintain. It covers code smells, such as long functions or repeated code, and refactoring techniques that preserve behaviour.
plugin-review
Review plugin quality with tiered checks and dependency scoping. Use for PR and pre-release audits.
assess-patch-risk
Assess an immutable patch artifact's program impact, regression risk, and auto-merge eligibility. Use for generated patch files, provider pull-request diffs, or commit ranges when reviewers need evidence about affected runtime paths, contracts, tests, and recoverability. This skill is read-only and does not generate…
time-bomb-radar
Finds deferred operations that crash on aged data -- code that passes every test but breaks weeks or months after release. Covers cascade deletes, cache expiry, trial paths, background accumulation, date-threshold transitions, and scheduled side effects. Triggers: "time bomb", "time-bomb", "/time-bomb-radar", "aged…
radar-suite-axis-classification
Shared axis classification framework for all radar-suite skills. Every finding must be classified as axis1 (bug), axis2 (scatter), or axis3 (dead/smelly) before emission, with mandatory coaching fields and file:line citations to existing patterns in the audited codebase. Triggers: invoked by every radar before…
pythinker-code-help
Answer Pythinker CLI usage, configuration, and troubleshooting questions. Use when user asks about Pythinker CLI installation, setup, configuration, slash commands, keyboard shortcuts, MCP integration, providers, environment variables, how something works internally, or any questions about Pythinker CLI itself.