How do I install the Precheck skill?

Run npx skillmds@latest add krzysztofdudek/precheck in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Precheck safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Precheck?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Precheck free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Precheck?

krzysztofdudek (@krzysztofdudek) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Precheck by krzysztofdudek — Agent Skill

Name: Precheck
Author: krzysztofdudek

Precheck

Use when user has drafted file content that is NOT on disk yet and wants a reviewer verdict before writing — explicit phrases like "would this pass review?", "check before I save", "will this be accepted?" with content not yet persisted. Costs 1 LLM call per rule. If the user references an existing file ("does this file pass", "check src/foo.ts") use autoreview:review instead — precheck requires unsaved draft content (pasted, or pointed at via `--content-file`). For plain "I'm about to edit file X", use autoreview:context (free) — don't precheck every edit. Skip when no `.autoreview/` exists — use autoreview:setup first.

File contents

AutoReview Pre-check

Use only when user has a draft in hand and wants a verdict BEFORE writing to disk. For "list rules that apply to this path", use autoreview:context instead — it's free.

Cross-platform. Snippets below use bash-style env-var syntax (${CLAUDE_PLUGIN_ROOT}). Translate to your shell on PowerShell ($env:CLAUDE_PLUGIN_ROOT) or cmd (%CLAUDE_PLUGIN_ROOT%). The plugin requires Node ≥22 — that's the only assumed binary.

When NOT to use

User said "edit this file". Just run autoreview:context to see rules, then write. Precheck is overkill for routine edits.
User is about to commit an already-written file. Use autoreview:review instead.
User already has the content on disk. Use autoreview:review --files <path>.

Steps

Save the draft to a temp file. Resolve the platform tmp dir cross-platform:
```
node -e "console.log(require('os').tmpdir())"
```
Then write the draft into that dir under a unique name (e.g. <tmpdir>/ar-draft-<timestamp>.ts). Any writable path is fine — /tmp/... on POSIX, %TEMP%\... on Windows; the Node call returns the right one for whichever you're on.
Pick a rule that you suspect is the strictest/most relevant for the target path. Use autoreview:context first to list applicable rules.
Run the reviewer test:
```
node ${CLAUDE_PLUGIN_ROOT}/scripts/bin/reviewer-test.mjs \
  --rule <rule-id> --file <target-path> --content-file <tmpdir>/ar-draft-<timestamp>.ts
```
Where <target-path> is the logical destination (e.g. src/api/users.ts) and --content-file is where the draft actually lives. The reviewer automatically resolves the rule's tier from config.yaml — no provider flags needed. To get file:line reasoning, temporarily set tiers.<name>.mode: thinking in .autoreview/config.yaml before running, then revert.
Parse the === RESULT === JSON. If satisfied: false, revise the draft and re-run. If satisfied: true, commit the write to <target-path>.

For multiple rules, run once per rule. Each call is ~1 LLM invocation at the rule's configured tier.

---name: precheckdescription: Use when user has drafted file content that is NOT on disk yet and wants a reviewer verdict before writing — explicit phrases like "would this pass review?", "check before I save", "will this be accepted?" with content not yet persisted. Costs 1 LLM call per rule. If the user references an existing file ("does this file pass", "check src/foo.ts") use autoreview:review instead — precheck requires unsaved draft content (pasted, or pointed at via `--content-file`). For plain "I'm about to edit file X", use autoreview:context (free) — don't precheck every edit. Skip when no `.autoreview/` exists — use autoreview:setup first.---# AutoReview Pre-checkUse only when user has a draft in hand and wants a verdict BEFORE writing to disk. For "list rules that apply to this path", use `autoreview:context` instead — it's free.> **Cross-platform.** Snippets below use bash-style env-var syntax (`${CLAUDE_PLUGIN_ROOT}`). Translate to your shell on PowerShell (`$env:CLAUDE_PLUGIN_ROOT`) or cmd (`%CLAUDE_PLUGIN_ROOT%`). The plugin requires Node ≥22 — that's the only assumed binary.## When NOT to use- User said "edit this file". Just run `autoreview:context` to see rules, then write. Precheck is overkill for routine edits.- User is about to commit an already-written file. Use `autoreview:review` instead.- User already has the content on disk. Use `autoreview:review --files <path>`.## Steps1. **Save the draft to a temp file.** Resolve the platform tmp dir cross-platform: ``` node -e "console.log(require('os').tmpdir())" ``` Then write the draft into that dir under a unique name (e.g. `<tmpdir>/ar-draft-<timestamp>.ts`). Any writable path is fine — `/tmp/...` on POSIX, `%TEMP%\...` on Windows; the Node call returns the right one for whichever you're on.2. **Pick a rule** that you suspect is the strictest/most relevant for the target path. Use `autoreview:context` first to list applicable rules.3. **Run the reviewer test:** ``` node ${CLAUDE_PLUGIN_ROOT}/scripts/bin/reviewer-test.mjs \ --rule <rule-id> --file <target-path> --content-file <tmpdir>/ar-draft-<timestamp>.ts ``` Where `<target-path>` is the logical destination (e.g. `src/api/users.ts`) and `--content-file` is where the draft actually lives. The reviewer automatically resolves the rule's tier from `config.yaml` — no provider flags needed. To get file:line reasoning, temporarily set `tiers.<name>.mode: thinking` in `.autoreview/config.yaml` before running, then revert.4. **Parse the `=== RESULT ===` JSON.** If `satisfied: false`, revise the draft and re-run. If `satisfied: true`, commit the write to `<target-path>`.For multiple rules, run once per rule. Each call is ~1 LLM invocation at the rule's configured tier.

Frequently asked questions