1---2name: full-test3description: "Complete testing pipeline — full test suite, test everything, automated and manual tests."4version: "2.0.0"5category: test6instructions: |7 You are an autonomous testing agent. Do NOT ask the user questions.8 Run the full pipeline below without pausing between phases.910 TARGET:11 $ARGUMENTS1213 If arguments are provided, focus testing on those features/flows.14 If no arguments are provided, test the entire application.1516 ============================================================17 PHASE 1: AUTOMATED E2E TESTS (/e2e)18 ============================================================1920 Follow the instructions defined in the `/e2e` skill exactly.21 Run all 9 phases: Stack Discovery > Environment Setup > Backend API Tests >22 Frontend UI Tests > Integration Tests > Test Execution > Self-Healing Fix Loop >23 Full Regression > Coverage Report.2425 Record the final test results and coverage report. Note which areas have26 strong automated coverage and which areas are harder to test automatically27 (complex user flows, visual regressions, edge cases requiring human judgment).2829 Do NOT stop here. Continue immediately to Phase 2.3031 ============================================================32 PHASE 2: MANUAL TEST PLAN (/manual-test-plan)33 ============================================================3435 Follow the instructions defined in the `/manual-test-plan` skill exactly.3637 IMPORTANT: When generating the manual test plan, factor in the automated38 test coverage from Phase 1:3940 - Do NOT duplicate scenarios that are already well-covered by automated tests.41 - FOCUS manual test scenarios on:42 1. Areas where automated tests found bugs (verify fixes manually)43 2. Flows that are hard to automate (multi-step UX, visual layout, real device behavior)44 3. Edge cases the automated suite couldn't cover (network conditions, permissions, etc.)45 4. Exploratory testing suggestions for areas with low automated coverage46 - Reference the automated test results: "Automated tests cover X; manually verify Y."4748 Do NOT stop here. Continue immediately to Phase 3.4950 ============================================================51 PHASE 3 (OPTIONAL): PERFORMANCE & LOAD TESTING52 ============================================================5354 Run this phase if the project has API endpoints, database queries, or55 user-facing pages that could degrade under load. Skip if the project is56 a CLI tool, library, or has no server component.5758 Steps:59 1. Identify critical endpoints and database-heavy operations from Phase 1.60 2. Write load test scenarios using the project's existing tooling (k6, Artillery,61 Locust, ab, autocannon, or simple scripted loops if no tool is installed).62 3. Run baseline benchmarks: measure p50, p95, p99 latency and throughput for63 each critical endpoint under normal load (10 concurrent users).64 4. Run stress tests: ramp to 50, then 100 concurrent users. Record where65 latency degrades or errors spike.66 5. Profile slow queries or endpoints. Check for N+1 queries, missing indexes,67 unbounded result sets, and missing pagination.68 6. Record results: response times, throughput limits, and bottleneck locations.6970 Do NOT stop here. Continue immediately to Phase 4.7172 ============================================================73 PHASE 4 (OPTIONAL): SECURITY TESTING (OWASP)74 ============================================================7576 Run this phase if the project has a web interface, API endpoints, or77 handles user input/authentication. Skip if the project is a pure library78 with no I/O surface.7980 Check each applicable OWASP Top 10 category:81 1. **Injection** — Test for SQL injection, NoSQL injection, command injection,82 and template injection on all user inputs and query parameters.83 2. **Broken Authentication** — Verify session management, password policies,84 token expiration, and brute-force protections.85 3. **Sensitive Data Exposure** — Check for secrets in source, unencrypted86 PII, missing HTTPS enforcement, overly verbose error messages.87 4. **Broken Access Control** — Test for IDOR, privilege escalation, missing88 authorization checks on endpoints, direct object references.89 5. **Security Misconfiguration** — Check default credentials, open CORS,90 debug mode in production configs, unnecessary open ports.91 6. **XSS** — Test for reflected, stored, and DOM-based XSS on all inputs92 that render in HTML or client-side templates.93 7. **CSRF** — Verify anti-CSRF tokens on state-changing requests.94 8. **Dependency Vulnerabilities** — Run `npm audit`, `pip audit`, `cargo audit`,95 or equivalent. Flag critical and high severity issues.96 9. **Rate Limiting** — Verify rate limits exist on auth endpoints, API routes,97 and form submissions.98 10. **Logging & Monitoring** — Check that auth failures, access denials, and99 input validation failures are logged without leaking sensitive data.100101 Record all findings with severity (Critical / High / Medium / Low / Info).102103 ============================================================104 OUTPUT105 ============================================================106107 When all phases are complete, print a summary:108109 ---110 ## Full Test Pass Complete111112113 PARALLEL EXECUTION: Use the Agent tool to run both phases concurrently.114 - Agent A (E2E Tests): "Run /e2e skill instructions on this project. Auto-detect the stack, generate and run exhaustive integration tests. Apply self-healing for failures. Return: test results, coverage summary, issues found."115 - Agent B (Manual Test Plan): "Run /manual-test-plan skill instructions on this project. Generate a comprehensive manual test plan from the codebase and any specs. Return: the complete test plan document."116 - Wait for both agents to complete.117 - Cross-reference: Remove manual test steps that are fully covered by passing automated tests from Agent A.118 - Merge into final output: automated test results + complementary manual test plan.119120121 **Automated E2E Results:**122 - Tests run: [N]123 - Passed: [N] | Failed: [N] | Fixed: [N]124 - Quality verdict: [ROCK SOLID / STABLE / FRAGILE / BROKEN]125126 **Manual Test Plan:**127 - Manual scenarios generated: [N]128 - Focus areas: [list areas needing manual verification]129130 **Performance & Load Testing:** [SKIPPED or results]131 - Baseline p95 latency: [Xms]132 - Stress test limit: [N concurrent users before degradation]133 - Bottlenecks found: [list or "none"]134135 **Security Testing (OWASP):** [SKIPPED or results]136 - Critical: [N] | High: [N] | Medium: [N] | Low: [N]137 - Key findings: [list top issues]138139 **Coverage assessment:**140 - Strong automated coverage: [list areas]141 - Needs manual verification: [list areas]142 - Gaps (no coverage): [list areas, if any]143144 **Next steps:**145 - Execute the manual test plan before merging146 - Fix any Critical/High security findings before merging147 - Address performance bottlenecks if p95 > 500ms148 - Run `/polish` for UX audit + QA verification + domain analysis149 - Run `/qa` for additional functional verification150 ---151152 STRICT RULES:153154 - Do NOT skip Phase 1 and only generate a manual test plan.155 - Do NOT duplicate automated test coverage in the manual plan.156 - Phase 2 must reference Phase 1 results to produce a complementary (not redundant) plan.157 - Phases 3 and 4 are optional — skip them only when the criteria above say to skip.158 - All rules from `/e2e` and `/manual-test-plan` apply to their respective phases.159platforms:160 - CLAUDE_CODE161metadata:162 source: https://skillmd.com/skills/tinh2/full-test163---164165166============================================================167SELF-EVOLUTION TELEMETRY168============================================================169170After producing output, record execution metadata for the /evolve pipeline.171172Check if a project memory directory exists:173- Look for the project path in `~/.claude/projects/`174- If found, append to `skill-telemetry.md` in that memory directory175176Entry format:177```178### /full-test — {{YYYY-MM-DD}}179- Outcome: {{SUCCESS | PARTIAL | FAILED}}180- Self-healed: {{yes — what was healed | no}}181- Iterations used: {{N}} / {{N max}}182- Bottleneck: {{phase that struggled or "none"}}183- Suggestion: {{one-line improvement idea for /evolve, or "none"}}184```185186Only log if the memory directory exists. Skip silently if not found.187Keep entries concise — /evolve will parse these for skill improvement signals.