Claude Code & MCP Integration

Use TINAA MSP tools directly from Claude Code and Claude Desktop via the Model Context Protocol.

Claude Code and MCP Integration

TINAA MSP exposes a full Model Context Protocol (MCP) server that allows Claude Code and Claude Desktop to interact with your platform directly. Instead of switching between your IDE and the TINAA dashboard, you can check quality scores, run tests, and analyse metrics through a natural conversation with Claude.

What Is MCP?

The Model Context Protocol (MCP) is an open protocol developed by Anthropic that defines a standard way for AI assistants to call external tools and read contextual resources. When a Claude-powered tool (like Claude Code) is configured with an MCP server, it can call that server’s functions and incorporate the results directly into its responses.

TINAA’s MCP server is built with FastMCP and communicates over stdio (standard input/output), which is the default transport supported by Claude Desktop and Claude Code.

Starting the MCP Server

Start the TINAA MCP server as a standalone process:

1
2
3
4
5

# From the TINAA MSP directory
python -m tinaa.mcp_server

# Or using the installed script entry point
tinaa-mcp

The server starts in stdio mode and is ready to accept MCP client connections. No network port is opened — communication happens entirely over stdin/stdout, which Claude Desktop and Claude Code manage automatically.

Configuring Claude Desktop

Add TINAA to your Claude Desktop configuration file.

macOS / Linux: ~/.config/claude-desktop/claude_desktop_config.json

Windows: %APPDATA%\claude-desktop\claude_desktop_config.json

1
2
3
4
5
6
7
8
9
10
11
12
13
14

{
  "mcpServers": {
    "tinaa-msp": {
      "command": "python",
      "args": ["-m", "tinaa.mcp_server"],
      "cwd": "/path/to/tinaa-playwright-msp",
      "env": {
        "DATABASE_URL": "postgresql+asyncpg://tinaa:tinaa@localhost:5432/tinaa",
        "REDIS_URL": "redis://localhost:6379/0",
        "TINAA_API_KEY": "your-api-key"
      }
    }
  }
}

If TINAA is running in Docker and your Python environment is inside the container, use:

1
2
3
4
5
6
7
8
9
10
11

{
  "mcpServers": {
    "tinaa-msp": {
      "command": "docker",
      "args": [
        "exec", "-i", "tinaa-playwright-msp-tinaa-1",
        "python", "-m", "tinaa.mcp_server"
      ]
    }
  }
}

Restart Claude Desktop after editing the configuration file. You will see the TINAA tools available in the tool picker.

Configuring Claude Code

If you use Claude Code (the CLI tool), add TINAA to your project’s .claude/settings.json:

1
2
3
4
5
6
7
8
9

{
  "mcpServers": {
    "tinaa-msp": {
      "command": "python",
      "args": ["-m", "tinaa.mcp_server"],
      "cwd": "/path/to/tinaa-playwright-msp"
    }
  }
}

All 14 MCP Tools

register_product

Register a new application with TINAA for continuous quality monitoring.

Parameters:

Parameter Type Required Description
name string Yes Human-readable product name
repo_url string Yes Git repository URL
environments object Yes Map of environment name to base URL
description string No Optional description

Example:

1
2
3
4
5
6
7
8
9

register_product(
  name="Checkout Service",
  repo_url="https://github.com/myorg/checkout-service",
  environments={
    "production": "https://checkout.myapp.com",
    "staging": "https://checkout-staging.myapp.com"
  },
  description="Cart, payment, and order confirmation"
)

list_products

List all registered products, optionally filtered by status.

Parameters:

Parameter Type Required Description
status string No Filter by active, paused, or archived

Example:

1

list_products(status="active")

Returns: List of product summaries including name, slug, quality score, and environment count.

get_product_detail

Get comprehensive details for a single product including environments, endpoints, quality score, and recent test runs.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID (e.g. prod-a1b2c3d4) or slug (e.g. checkout-service)

Example:

1

get_product_detail(product_id_or_slug="checkout-service")

get_quality_score

Get the current quality score for a product with a full breakdown by component.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
environment string No Scope the score to a specific environment

Example:

1
2
3
4

get_quality_score(
  product_id_or_slug="checkout-service",
  environment="production"
)

Returns: Composite score (0–100), grade, component breakdown, trend, and recommendations.

get_quality_report

Generate a comprehensive quality report with trend analysis and top issues.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
environment string No Scope to a specific environment

Example:

1
2
3
4

get_quality_report(
  product_id_or_slug="checkout-service",
  environment="production"
)

Returns: Quality score, 7-day and 30-day trends, test/performance/security/accessibility summaries, top issues, and recommendations.

run_playbook

Execute a specific test playbook against an environment.

Parameters:

Parameter Type Required Description
playbook_id_or_name string Yes Playbook ID or name (e.g. checkout-regression)
environment string No Target environment (defaults to staging)
target_url string No Override the base URL for this run

Example:

1
2
3
4

run_playbook(
  playbook_id_or_name="checkout-regression",
  environment="staging"
)

Returns: Run ID, status, pass/fail counts, duration, and quality score impact.

run_suite

Run all playbooks for a product/environment combination.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
environment string No Target environment (defaults to staging)
suite_type string No One of smoke, regression, accessibility, performance, security (defaults to regression)

Example:

1
2
3
4
5

run_suite(
  product_id_or_slug="checkout-service",
  environment="staging",
  suite_type="smoke"
)

Returns: Suite run ID, number of playbooks executed, pass/fail counts, duration, and updated quality score.

get_test_results

Get recent test run results for a product.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
limit integer No Maximum results to return (default: 10)
status string No Filter by passed, failed, error, or running

Example:

1
2
3
4
5

get_test_results(
  product_id_or_slug="checkout-service",
  status="failed",
  limit=5
)

suggest_tests

Get test suggestions based on codebase analysis and changed files.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
changed_files list[string] No List of changed file paths from the current diff

Example:

1
2
3
4

suggest_tests(
  product_id_or_slug="checkout-service",
  changed_files=["src/checkout.py", "src/cart.py", "src/payment.py"]
)

Returns: Prioritised list of playbook suggestions with reasoning and affected journeys.

create_playbook

Create a new test playbook for a product.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
name string Yes Human-readable playbook name
steps list[dict] Yes Ordered list of action dicts
assertions dict No Global assertion rules
performance_gates dict No Performance thresholds

Example:

1
2
3
4
5
6
7
8
9
10

create_playbook(
  product_id_or_slug="checkout-service",
  name="homepage-smoke",
  steps=[
    {"action": "navigate", "url": "https://staging.myapp.com"},
    {"action": "assert_visible", "selector": "h1"},
    {"action": "screenshot", "name": "homepage"}
  ],
  performance_gates={"lcp_ms": 2500, "cls": 0.1}
)

get_metrics

Query APM metrics for a product or specific endpoint.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
endpoint_path string No Filter to a specific endpoint path
metric_type string No One of response_time, lcp, cls, availability, error_rate
hours integer No Lookback window in hours (default: 24)

Example:

1
2
3
4
5

get_metrics(
  product_id_or_slug="checkout-service",
  metric_type="lcp",
  hours=6
)

Returns: Time-series data points, baseline P50/P95, current average, and trend direction.

configure_alerts

Configure alert rules for a product.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
rules list[dict] Yes List of alert rule configurations

Example:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

configure_alerts(
  product_id_or_slug="checkout-service",
  rules=[
    {
      "type": "quality_score_drop",
      "threshold": 10,
      "channels": ["slack:#alerts"]
    },
    {
      "type": "endpoint_down",
      "threshold": 3,
      "channels": ["pagerduty"]
    }
  ]
)

get_deployments

Get recent deployments for a product with their quality impact.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug
environment string No Filter to a specific environment
limit integer No Maximum deployments to return (default: 10)

Example:

1
2
3
4
5

get_deployments(
  product_id_or_slug="checkout-service",
  environment="production",
  limit=5
)

explore_codebase

Trigger codebase exploration to discover routes, endpoints, and user journeys.

Parameters:

Parameter Type Required Description
product_id_or_slug string Yes Product ID or slug

Example:

1

explore_codebase(product_id_or_slug="checkout-service")

Returns: Discovered routes, API endpoints, forms, and inferred user journeys.

Example Workflow: “Check My App’s Quality”

The following shows how a typical Claude conversation might unfold when a developer asks about their application’s quality.

Developer asks Claude:

“How is the checkout service doing quality-wise? Are there any regressions?”

Claude calls TINAA tools:

1

get_quality_score(product_id_or_slug="checkout-service", environment="production")

1

get_quality_report(product_id_or_slug="checkout-service")

Claude responds:

The Checkout Service has a quality score of 87/100 (Grade A) in production. Here is the breakdown:

  • Test Health: 91/100 — 94% pass rate, strong coverage
  • Performance: 84/100 — P95 response time is 420ms (budget: 500ms); LCP is 1.8s (good)
  • Security: 88/100 — All headers present, TLS grade A
  • Accessibility: 82/100 — 4 violations found

The score is improving (+2 points over the last 7 days).

Top issues to address:

  1. 3 images on /checkout are missing alt text (accessibility, medium severity)
  2. /api/search P95 response time is 520ms, exceeding the 500ms budget

Example Workflow: “Run Smoke Tests Before Deploy”

Developer asks Claude:

“I’m about to deploy to production. Can you run the smoke tests on staging first?”

Claude calls TINAA tools:

1
2
3
4
5

run_suite(
  product_id_or_slug="checkout-service",
  environment="staging",
  suite_type="smoke"
)

Claude responds:

I ran the smoke suite on staging. Here are the results:

  • Suite: smoke
  • Duration: 28.4 seconds
  • Playbooks executed: 8
  • Passed: 7
  • Failed: 1
  • Quality score: 85

1 failure detected: The checkout-regression playbook failed — assert_url expected a match for /order-confirmation/[0-9]+ but got /cart?error=payment_declined.

I would recommend investigating the payment flow on staging before deploying to production. Would you like me to look at the full test run details?

Next Steps