templates/skills/testing/api-manual/SKILL.md
Manually verify API behavior with reproducible requests and response checks; not contract version compatibility governance
npx skillsauth add ozerohax/assistagents testing-api-manualInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
<input_requirements> <required>Base environment URL</required> <required>Auth scheme and access</required> <required>Endpoints and contracts list (params, bodies, responses)</required> <optional>Spec (OpenAPI/Swagger) and API version</optional> <optional>Test data and initial state</optional> <optional>Rate limit and timeout constraints</optional> </input_requirements>
<preparation> <steps> <step>Verify environment availability and basic health</step> <step>Prepare tokens/keys and store them in variables</step> <step>Prepare a minimal set of reusable curl templates</step> <step>Set up request-id/correlation-id variables (if used)</step> </steps> </preparation><execution_rules> <rule importance="critical">Every request must be reproducible</rule> <rule importance="critical">Verify status code and response contract</rule> <rule importance="high">Cover positive and negative scenarios</rule> <rule importance="high">Record headers and parameters that affect behavior</rule> <rule importance="high">Verify response schema and field types</rule> <rule importance="high">Verify pagination, sorting, and filtering</rule> <rule importance="high">Verify idempotency where applicable</rule> <rule importance="medium">Note dependencies between requests (chain)</rule> </execution_rules>
<coverage> <focus> <item>CRUD scenarios</item> <item>Input validation</item> <item>Authorization and access control</item> <item>Errors and edge cases</item> <item>Pagination/filtering/sorting</item> <item>Rate limiting and error codes</item> </focus> </coverage><quality_rules> <rule importance="critical">Expected outcome is stated unambiguously</rule> <rule importance="high">No duplicate scenarios with different wording</rule> <rule importance="high">Server and client errors are distinguished and verified separately</rule> <rule importance="high">Record request-id/correlation-id when available</rule> <rule importance="medium">If result recording is needed, use a single consistent format</rule> </quality_rules>
<do_not> <item importance="critical">Do not run destructive requests in production</item> <item importance="high">Do not use real user data</item> <item importance="high">Do not mutate state unless the scenario requires it</item> <item importance="high">Do not leak tokens/keys into shell history or logs</item> </do_not>
<example_templates> <template>curl -X GET "$BASE_URL/resource" -H "Authorization: Bearer $TOKEN"</template> <template>curl -X POST "$BASE_URL/resource" -H "Content-Type: application/json" -d '{"key":"value"}'</template> <template>curl -X GET "$BASE_URL/resource?page=1&limit=20" -H "Authorization: Bearer $TOKEN"</template> </example_templates>
testing
Visual regression (baseline diffs, UI stabilization)
testing
Design detailed reproducible test cases (Given/When/Then, one check per case); not smoke/regression checklists
development
Mobile testing (native/hybrid/web, gestures, offline, interruptions)
testing
Load testing (k6/JMeter scenarios, benchmarks)