MCP Referenceinvarium_prepare_blueprint

invarium_prepare_blueprint

Validate and preview an agent blueprint locally before uploading to Invarium. Runs schema validation, confidence scoring, and a full Agent Readiness Audit — nothing is sent to the backend.

When to Use

Call invarium_prepare_blueprint as the first step when registering or updating an agent. It validates the blueprint YAML, computes a confidence level, and runs a local Agent Readiness Audit (ARS) that scores the blueprint on security, reliability, system design, and tool quality.

Use this to iterate on your blueprint before committing to an upload. Fix findings, re-run invarium_prepare_blueprint to see the updated score, and only call invarium_upload_blueprint once you are satisfied.

See Create and Upload a Blueprint for the end-to-end workflow.

Parameters

NameTypeRequiredDefaultDescription
blueprintstringYesYAML string of the agent blueprint. Must include agent_name (or pass it separately), at least one tool, and valid schema fields.
agent_namestringNonullAgent name override. If omitted, the name is extracted from the agent_name or name field in the blueprint YAML.
invarium_prepare_blueprint

Validate and preview an agent blueprint BEFORE uploading to Invarium. Runs schema validation, confidence scoring, and a full static audit locally -- nothing is sent to the Invarium backend. Use this to review the blueprint with the user before calling invarium_upload_blueprint.

Parameters

NameTypeRequiredDescription
blueprintstringrequiredYAML string of the agent blueprint.
agent_namestring | nulldefault: nullOptional agent name override. Falls back to blueprint field.

Returns

Validation result with blueprint preview, confidence level, Agent Readiness Score, per-category breakdown, and actionable findings.

Example

"Validate the blueprint for my customer-support-agent before uploading"

Response

On success, the tool returns a formatted text block with five sections:

Blueprint saved to: .invarium/customer_support_agent_blueprint.yaml

Blueprint preview for "customer-support-agent"

  Agent: customer-support-agent
  Description: Handles customer inquiries via chat
  Framework: langchain
  Tools (3): search_tickets, create_ticket, escalate_to_human
  Workflow chains: 2
  Confidence: high

Agent Readiness Score: 82/100 (Green)

  Category breakdown with per-category scores and recoverable points.

  Fix these to reach 100:

  SECURITY
    Tool 'search_tickets' missing error_handling
    No rate limiting defined

  RELIABILITY
    No fallback tools defined

Review the blueprint file and the score above. To improve your
score, fix the findings above and re-run prepare_blueprint.
When ready, confirm and I'll upload to Invarium.

Blueprint Preview

FieldDescription
AgentThe resolved agent name.
DescriptionAgent description from the blueprint.
FrameworkThe framework the agent uses (e.g., langchain, crewai, custom).
ToolsNumber of tools and their names.
Workflow chainsNumber of workflow chains defined.
ConfidenceHow well the blueprint describes the agent: high, medium, or low.
GraphNode and edge counts, if a graph is included in the blueprint.

Agent Readiness Score (ARS)

The ARS is a 0—100 score computed from four weighted categories:

Scores across four categories: Security, Reliability, System Design, and Tool Quality — weighted by production risk.

CategoryWhat It Measures
SecurityGuardrails, PII handling, input validation, authorization checks.
ReliabilityError handling, fallbacks, retry policies, timeout configuration.
System DesignWorkflow completeness, constraint coverage, system prompt quality.
Tool QualityParameter definitions, descriptions, return types, side effect declarations.

Each finding deducts points from its category. The “Fix these to reach 100” section shows which findings to address.

See Agent Readiness Audit for a full explanation of scoring and categories.

Saved Blueprint File

The tool saves the blueprint to .invarium/<agent_name>_blueprint.yaml in your project directory. You can review this file before proceeding to upload.

Examples

Basic Usage

Validate a blueprint and review the audit results:

"Validate this blueprint for my customer-support-agent before uploading"

With Agent Name Override

Pass the agent name separately instead of including it in the blueprint YAML:

"Prepare the blueprint and use 'travel-booking-agent' as the agent name"

Iterative Improvement

After reviewing findings, fix the blueprint and re-run to see the updated score:

// First run: ARS = 62/100
"Validate the blueprint for my agent"

// Fix findings: add error_handling, guardrails, descriptions
// Second run: ARS = 91/100
"Re-validate the improved blueprint"

// Satisfied -- proceed to upload
"Upload the blueprint to Invarium"

Error Responses

ErrorCauseFix
Blueprint is not valid YAML: ...The blueprint string is not parseable YAML.Check for syntax errors. Regenerate the blueprint if needed.
Missing agent_name. Provide it as a parameter or include it in the blueprint.No agent name found in the blueprint YAML or the agent_name parameter.Add agent_name to the blueprint YAML or pass it as the agent_name parameter.
Blueprint validation failed: ...One or more schema validation errors.Fix the listed validation errors and try again. See Blueprint Schema.

See Error Codes for the full error reference.

Was this page helpful?