wise-simplify-auto · diff

git:20260914.d768486 to git:20260915.2471a1b

31 added, 20 removed. Audit A to A.

---

name: wise-simplify-auto

description: >-

- Autonomously simplify recently-modified code and commit it — dispatches

- the `code-simplifier` agent over the working tree (cleanup only,

- behaviour preserved: clarity, consistency, dead-code/redundancy

- removal), then drafts a Conventional-Commits subject and commits. The

+ Autonomously simplify recently-modified code and commit it — runs the

+ simplify pass over the working tree (the `code-simplifier` agent on

+ Claude Code when installed, otherwise the same cleanup inline on the

+ current model; behaviour preserved: clarity, consistency,

+ dead-code/redundancy removal), then drafts a Conventional-Commits

+ subject and commits. Works on every harness. The

lightweight per-commit tier of the plugin's two-tier quality model, as a

standalone building block. No routine prompts, never pushes.

Invoked as `/wise-simplify-auto` (bare alias) or

`/wise:wise-simplify-auto` (canonical). Use when the user says "simplify

and commit", "clean up and commit", "run a simplify pass", or types

`/wise-simplify-auto`.

argument-hint: "[--on <harness>[:<model>[:<effort>]] | --on ask]"

allowed-tools: Task, Read, Bash(git:*), Bash(bash:*), AskUserQuestion

---

# /wise-simplify-auto — simplify recently-modified code and commit

- Before executing, follow [model fallback](../../references/workflow-host-control.md#model-fallback)

- for unavailable models or delegation routes, including in autonomous procedures.

+ This skill has no model preference: the simplify pass runs on the current

+ model, and the [model fallback](../../references/workflow-host-control.md#model-fallback)

+ picker only applies to an explicit `--on` selection.

At every skill start, identify your main/child role and the current client

and GUI/TUI question tools, then read and follow the

[question lifecycle](../../references/workflow-host-control.md#keep-asynchronous-questions-open).

Keep asynchronous prompts open until answered; this rule does not authorize

questions in autonomous or otherwise prompt-free procedures.

## Why this skill exists

- The simplify pass (the `code-simplifier` agent) is the plugin's

- lightweight per-commit cleanup; it is wired into the commit routine

+ The simplify pass is the plugin's lightweight per-commit cleanup: the

+ `code-simplifier` agent on Claude Code when the optional plugin is

+ installed, otherwise the same instructions

+ (`references/simplify-instructions.md`) applied inline on the current

+ model, so Codex, Cursor, Gemini and Grok children run it too. It is

+ wired into the commit routine

(`/wise-commit` runs it before staging) and into the implement phase

(per task). This skill exposes it as a **standalone, decision-free

building block**: run the pass, then commit the result — no prompts. The

heavier multi-agent code-review branch gate is the other tier

(the `code-review` workflow: `/wise-workflow-run code-review`).

(Close cousin of `/wise-commit`, which also simplifies-then-commits via

its `SIMPLIFY=yes` default; this skill makes the simplify step the

explicit headline and is the autonomous building block workflows reuse.)

## Invocation

```

/wise-simplify-auto

/wise:wise-simplify-auto # canonical namespaced form

```

No positionals, no flags. If the argument string contains anything other

than optional whitespace, stop with:

```

Unknown argument(s): <the extra tokens>

Usage: /wise-simplify-auto

```

## Run on another harness (`--on`)

If `$ARGUMENTS` contains `--on <harness>[:<model>[:<effort>]]` (or

`--on ask` / a bare `--on`), do

NOT run the procedure below in this conversation. Strip the `--on`

tokens (everything left is `SKILL_ARGS`), then read

`${CLAUDE_PLUGIN_ROOT}/references/dispatch.md` and follow it with:

- `SKILL_MD` = `${CLAUDE_PLUGIN_ROOT}/skills/wise-simplify-auto/SKILL.md`

- `SKILL_ARGS` = the remaining tokens

`--on ask` (or a bare `--on`) picks harness, model and effort through

- one composite `AskUserQuestion` before any child spawns. Unavailable execution

- models additionally require the shared model-fallback picker; routine work

- remains decision-free.

+ one composite `AskUserQuestion` before any child spawns; routine work

+ remains decision-free. The child runs the pass inline on its own model.

The reference probes the harness login, validates model and effort

against the engine catalog, and runs the procedure as a headless child

via `engine.sh dispatch --relay`. Follow its run, handle any required gates

in the main harness, and relay the final result as specified by the shared

dispatch reference. This does not add routine prompts to autonomous paths.

Without `--on`, this section does not apply.

## Procedure

### 1. Simplify recently-modified code

Run the simplify pass per

- `${CLAUDE_PLUGIN_ROOT}/references/simplify-pass.md` — dispatch the

- `code-simplifier` agent (a `Task` subagent) over the working tree's

- recently-modified code. Surface its summary verbatim. On a simplify

- failure, follow that reference's failure policy and stop with

- `SIMPLIFY: failed reason="<one-line>"`. An unavailable model or named agent

- uses the reference's GUI/TUI model-fallback gate before any replacement runs.

+ `${CLAUDE_PLUGIN_ROOT}/references/simplify-pass.md` over the working

+ tree's recently-modified code. The reference picks the route: the

+ `code-simplifier` agent (a `Task` subagent) when this Claude Code session

+ lists it, otherwise the cleanup inline on the current model per

+ `${CLAUDE_PLUGIN_ROOT}/references/simplify-instructions.md`. A missing

+ agent is not an error and opens no picker. Surface the summary and the

+ route verbatim. On a simplify failure (the pass ran and broke the tree),

+ follow that reference's failure policy and stop with

+ `SIMPLIFY: failed reason="<one-line>"`.

### 2. Commit the result

Follow `${CLAUDE_PLUGIN_ROOT}/skills/wise-commit/commit-routine.md` with

`SIMPLIFY=no PUSH=no` (the pass already ran in §1 — `SIMPLIFY=no` avoids

a redundant second pass; `PUSH=no` because this skill never pushes). The

routine stages, drafts a Conventional-Commits subject, and commits.

### 3. Final line

Relay the routine's final `COMMIT:` line verbatim:

```

COMMIT: ok subject="<subject>" pushed=no

COMMIT: skip reason="nothing to commit"

COMMIT: failed reason="<verbatim error>"

```

## Guardrails

- - No routine mid-run questions. The `--on ask` picker and required model-fallback

- gate are explicit exceptions, rendered only by the main harness.

- - One simplify pass — never re-dispatch the agent to iterate-to-clean.

+ - No routine mid-run questions. The `--on ask` picker is the one

+ exception, rendered only by the main harness.

+ - One simplify pass — never re-run it to iterate-to-clean.

+ - Never fail or stop because the `code-simplifier` agent is absent; run

+ the pass inline instead.

- Never `git push` — use `/wise-commit-push` for that.

- All of `commit-routine.md`'s guardrails apply (no `--amend` /

`--no-verify` / `--force`, no AI-attribution trailer, no retry on

failure).

- Never invoke another wise action skill.

- Profile-insensitive by design: the session token-budget profile (`/wise-profile`) does not change this skill — one cheap behavior-preserving pass; scaling it would risk the behaviour-preserved guarantee.