test-guide-from-code
Generate a manual test guide for human testers from code changes (diff, commit, MR/PR). Triggers when user says 「生成测试指南」「测试指南」「人工测试指南」「测试指引」「test guide」「根据代码生成测试」「从 MR 生成测试指南」「从 PR 生成测试指南」 (generate test guide / manual test guide / test guide from code / from MR / from PR), or provides a diff/MR/PR link and asks for a test guide.
How do I install this agent skill?
npx skills add https://github.com/fudesign2008/open-skills --skill test-guide-from-codeIs this agent skill safe to install?
- Gen Agent Trust Hubpass
This skill is generally safe but is vulnerable to indirect prompt injection because it processes untrusted code changes from external sources like GitHub or GitLab without sanitization.
- Socketwarn
1 alert: gptAnomaly
- Snykwarn
Risk: MEDIUM · 2 issues
What does this agent skill do?
Manual Test Guide Generation
Automatically generate a structured manual test guide from code changes, enabling testers to precisely verify changed functionality.
Input formats, output templates, and examples: see reference.md.
Prerequisite skill check
On load, verify report-output-discipline is available. If missing, abort and print:
npx skills add FuDesign2008/open-skills -g --skill '*' --yes
Do not silently fall back.
Triggers
- 「生成测试指南」「测试指南」「人工测试指南」「测试指引」 (generate test guide / manual test guide)
- 「test guide」
- 「根据代码生成测试」「从代码生成测试指南」 (generate test from code)
- 「从 MR 生成测试指南」「从 PR 生成测试指南」 (test guide from MR / from PR)
- Supports arguments: 「生成测试指南:https://gitlab.example.com/merge_requests/123」
When to Use
- After an MR/PR is submitted, provide a verification guide for testers
- Before a release, outline manual regression testing scope based on code changes
- For feature delivery acceptance, derive testable user behaviors from code changes
Execution Flow
Step 1: Obtain Changes
- Detect input type (see input format table in reference.md), choose the appropriate retrieval method
- If URL: try CLI to get diff; if fails, prompt user to paste
- If no input:
git diff(working tree) orgit diff HEAD~1(latest commit) - Confirm a valid diff was obtained
Step 2: Change Analysis
- Changed file inventory: grouped by functional module
- Change type classification: 🆕 Added / 🔄 Modified / 🐛 Fixed / ⚙️ Config / 🗑️ Removed
- User-perceivable behavior changes: infer impact on user/system behavior
- Implicit change identification: config files, dependency upgrades, database migrations, API contract changes
Step 3: Generate Test Guide
Each test item includes: test objective, prerequisites, test steps (non-developer friendly), expected results, priority.
Priority: P0 core business/data writes/security → P1 auxiliary features/UI/performance → P2 styling/copy/internal refactoring
Change type → Test strategy:
- 🆕 → Happy path + boundary/edge cases
- 🔄 → Verify changed points + regression of existing functionality
- 🐛 → Bug scenario verification + regression of related features
- ⚙️ → Config effectiveness + regression of affected features
- 🗑️ → Confirm removal + verify related features are unaffected
Step 4: Supplement Regression Verification
Derive related modules from changed modules, list functional points requiring regression and verification methods.
Step 5: Output
Generate the test guide using the template format in reference.md, saving as a markdown file to the current working directory by default.
File naming: <feature>_manual_test_guide_<YYYY-MM-DD>.md (naming rules and examples: see reference.md "File Naming Rules")
User override: User can specify a custom path or filename at trigger time; save as specified.
Conversation output: Do not repeat full content — only output a brief summary: save path, test item statistics (by P0/P1/P2 distribution), 3–5 core change points, regression scope. Full test items are always written to file only (strategy: see reference.md "Conversation Output Strategy").
Derive the HTML page after the Markdown file is written. Registry row: 人工测试指南 in report-output-discipline/reference.md (human + 接入). Derive <feature>_manual_test_guide_<YYYY-MM-DD>.html beside the source following report-output-discipline, and report both paths.
Red Flags
- ❌ Skipping implicit change identification, only analyzing business code
- ❌ Test steps using developer jargon ("call API", "returns 200")
- ❌ Missing the original bug scenario for bug fixes
- ❌ Mixing automated tests into the manual test guide
- ❌ Ignoring regression verification scope
- ❌ Generating test items from nothing when no diff was obtained
- ❌ Writing files without confirming the current working directory, causing files to land in the wrong location
- ❌ Filename containing spaces or special filesystem characters (
/\:*?"<>|) - ❌ Not reporting the full save path in the conversation after generating the file
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/fudesign2008/open-skills/test-guide-from-code">View test-guide-from-code on skillZs</a>