@glubean/cli 0.2.5 → 0.3.0

package/templates/AI-INSTRUCTIONS.md

@@ -1,160 +0,0 @@
-# Glubean Test Project
-
-API test automation project using `@glubean/sdk`. Runtime is **Node.js**. Packages are from **npm**.
-
-## Project Structure
-
-```
-tests/           # Permanent test files — run in CI and Cloud
-explore/         # Exploratory tests — quick iteration in your editor
-data/            # Test data files (JSON, CSV, YAML)
-context/         # API specs and reference docs for AI and tooling
-.env             # Public variables (BASE_URL)
-.env.secrets     # Credentials — gitignored
-package.json     # Runtime config, dependencies, glubean settings
-```
-
-## Setup
-
-If `glubean run` fails with module-not-found errors, run:
-
-```bash
-npm install
-```
-
-## BEFORE Writing or Modifying Any Test (MANDATORY)
-
-You MUST complete these steps before writing any test code. Do NOT skip this — not even for "simple" tests.
-
-1. Search `context/` for the API you are testing. Check OpenAPI specs (`context/*.json`) for exact paths, methods,
-   request/response schemas. **Do NOT guess endpoint paths or response shapes.**
-2. Check `.glubean/docs/` exists. If missing, download:
-   ```bash
-   mkdir -p .glubean/docs
-   curl -sL https://raw.githubusercontent.com/glubean/cheatsheet/main/sdk-reference.md -o .glubean/docs/sdk-reference.md
-   curl -sL https://raw.githubusercontent.com/glubean/cheatsheet/main/patterns.md -o .glubean/docs/patterns.md
-   curl -sL https://raw.githubusercontent.com/glubean/cheatsheet/main/cli-reference.md -o .glubean/docs/cli-reference.md
-   ```
-3. Read `.glubean/docs/sdk-reference.md` — all exports, signatures, options.
-4. Read `.glubean/docs/patterns.md` — runnable examples for every test type.
-5. THEN write the test using only patterns from those docs and API details from step 1.
-
-**If you write test code without completing steps 1-4, you are violating this rule.** Do NOT guess — the endpoint paths,
-response schemas, and SDK API are all documented.
-
-**If `context/` has no spec or docs for the API under test:** Unless it is a well-known public API (e.g. GitHub,
-Stripe), STOP and ask the user for the endpoint path, method, and response schema. Do NOT hallucinate paths or response
-shapes.
-
-**If the OpenAPI spec exists but lacks response schemas:** Warn the user that response structure will be inferred at
-runtime, which lowers first-pass accuracy. Suggest adding response examples to `context/` to enable one-shot test
-generation.
-
-## Quick Checklist
-
-Before submitting any test, verify:
-
-- [ ] Checked `context/` for endpoint spec + response schema
-- [ ] Response schema available? If not, warned user about reduced accuracy
-- [ ] Test ID is kebab-case and unique across the project
-- [ ] Tags are set (e.g. `["api"]`, `["smoke"]`, `["e2e"]`)
-- [ ] Creates resources → has `.teardown()` cleanup
-- [ ] No hardcoded secrets or test data in the test file
-
-## Coverage Expectations
-
-Do not stop at a single happy-path test unless the user explicitly asks for only one case.
-
-- For authentication/identity endpoints, include:
-  - one authenticated success case
-  - one unauthenticated or invalid-credential case
-- For protected resources, include authorization failure cases (`401` or `403`) when applicable.
-- For create/update endpoints, include at least one invalid-input case when the contract documents validation rules.
-- If the OpenAPI spec lacks enough detail for a negative case, say so explicitly and choose the safest verifiable case.
-
-## Definition of Done for API Tests
-
-Before considering an API test complete, verify whether the endpoint has:
-
-- a success path
-- an auth boundary
-- a validation boundary
-- a not-found or forbidden boundary
-
-Cover all applicable boundaries unless the user asked for a narrower scope.
-
-## Import Convention
-
-Always use the npm package name:
-
-```typescript
-// Correct
-import { test } from "@glubean/sdk";
-```
-
-## Conventions
-
-- **One test export per behavior.** Each `export const` is a test case.
-- **Tags:** `["smoke"]` for health checks, `["api"]` for API tests, `["e2e"]` for browser tests.
-- **IDs:** kebab-case, unique across the project. Used for history tracking.
-- **Data separation:** keep test data in `data/`, never hardcode payloads in test files.
-- **Secrets:** use `secrets.require("KEY")` or configure's secret binding. Never hardcode credentials.
-- **Cleanup:** use `.teardown()` for any resource created during the test.
-- **Imports:** use `.ts` extensions for local files.
-
-## Running Tests
-
-```bash
-glubean run                          # Run all tests in tests/
-glubean run tests/                   # Run a specific directory
-glubean run --filter smoke           # Run by tag
-glubean run --explore                # Run explore/ tests
-glubean run --upload                 # Run and upload results to Cloud
-```
-
-For all CLI options, run `glubean --help` or `glubean run --help`.
-
-## MCP Setup (enables AI closed-loop)
-
-Without MCP, the AI can only write tests. With MCP, the AI can: **write → run → read failures → fix → rerun → pass** —
-without human intervention.
-
-### Claude Code
-
-```bash
-claude mcp add glubean -- npx @glubean/mcp
-```
-
-### Cursor
-
-Add to `~/.cursor/mcp.json`:
-
-```json
-{
-  "mcpServers": {
-    "glubean": {
-      "command": "npx",
-      "args": ["@glubean/mcp"]
-    }
-  }
-}
-```
-
-### Available MCP Tools
-
-- `glubean_run_local_file` — run a test file locally, returns structured results (assertions, logs, traces)
-- `glubean_discover_tests` — scan a file and return test export metadata (id, name, tags)
-- `glubean_list_test_files` — list all test files in the project
-- `glubean_diagnose_config` — check project config for common issues (.env, package.json, dirs)
-- `glubean_get_last_run_summary` — get summary of the most recent local run
-- `glubean_get_local_events` — get filtered events (assertions, logs, traces) from the last run; useful for debugging
-  failures
-
-## Type Definitions
-
-For full API details beyond the cheatsheet, go to the import source:
-
-```typescript
-import { configure, fromDir, test } from "@glubean/sdk";
-// Go-to-definition on these for exact signatures and options
-```

package/templates/ci-config/ci.yaml

@@ -1,13 +0,0 @@
-# CI run config — used by: npm run test:ci
-# Optimized for CI pipelines: fail fast, structured output, no pretty-print
-run:
-  testDir: ./tests
-  failFast: true
-  pretty: false
-  # Emit full HTTP trace in results (useful for debugging CI failures)
-  emitFullTrace: false
-  # Stop after N failures (uncomment to enable)
-  # failAfter: 5
-
-redaction:
-  replacementFormat: simple

package/templates/ci-config/default.yaml

@@ -1,9 +0,0 @@
-# Default run config — used by: npm test
-run:
-  testDir: ./tests
-  exploreDir: ./explore
-  verbose: false
-  pretty: true
-
-redaction:
-  replacementFormat: simple

package/templates/ci-config/explore.yaml

@@ -1,5 +0,0 @@
-# Explore run config — used by: npm run explore
-run:
-  exploreDir: ./explore
-  verbose: false
-  pretty: true

package/templates/ci-config/staging.yaml

@@ -1,9 +0,0 @@
-# Staging run config — used by: npm run test:staging
-run:
-  testDir: ./tests
-  envFile: .env.staging
-  verbose: false
-  pretty: true
-
-redaction:
-  replacementFormat: simple