Skip to main content

Specification compiler for AI agent orchestration

Project description

Blueprint

Universal AI Orchestration Contract

Blueprint is the industry-standard specification format for multi-agent AI execution. Any flagship model — Claude, GPT, Gemini, Grok, DeepSeek — can parse, execute, and hand off Blueprint tasks with zero ambiguity.

North Star: A specification so precise that any AI agent can execute it identically, hand off to any other agent mid-task, and verify completion deterministically.

What Blueprint Does

[Your Goal] → blueprint generate → [Perfect Specification]

You describe what you want. Blueprint produces a detailed, machine-parseable specification so precise that any AI agent — Claude, GPT, Gemini, Grok, DeepSeek, or future models — can execute it without ambiguity.

Fleet-Validated: This spec has been reviewed and approved by 5 frontier AI models (Claude Opus 4.5, GPT-5.2 Codex, Gemini 3 Pro, Grok 4, DeepSeek Coder) for parseability, executability, and hand-off reliability.

What Blueprint Does NOT Do

Blueprint does not execute tasks. Blueprint does not orchestrate agents. Blueprint does not manage workflows.

Blueprint generates the blueprint. That's it.

What you do with the blueprint afterward is your business. Feed it to Claude Code. Parse it with your own tooling. Hand it to a human developer. We don't care. Our job ends when the specification is generated.

Why This Matters

As AI coding agents become mainstream, the bottleneck shifts from "knowing how to code" to "knowing how to specify."

Natural language is ambiguous:

  • "Build a login system" → What auth method? What database? What error handling?

Ambiguity at scale is catastrophic. Fifty agents misinterpreting fifty tasks = chaos.

Blueprint eliminates ambiguity. Every task in a Blueprint has:

  • Typed interfaces (JSON Schema for inputs/outputs)
  • Execution context (working directory, environment, required tools)
  • Testable acceptance criteria (shell commands, not prose)
  • Defined error policy (retry logic, failure actions)
  • Clear dependencies (DAG structure)
  • Rollback procedures (undo commands)

AI agents don't interpret a Blueprint. They execute it.

Installation

pip install blueprint-ai

Usage

# Set your LLM provider (auto-detects from environment)
export ANTHROPIC_API_KEY="sk-ant-..."
# or
export OPENAI_API_KEY="sk-..."

# Generate a blueprint
blueprint generate "Build a REST API for user management with JWT auth" -o api.bp.md

# That's it. Blueprint's job is done.
# Now feed api.bp.md to your preferred AI agent:
claude-code "Execute T1 from api.bp.md"
aider --message "Implement task T1.1 from api.bp.md"

Example Output

Blueprint generates structured specifications like this:

task_id: T1.2
name: "Implement JWT token validation"
status: not_started
dependencies: [T1.1]

interface:
  input: "JWT token and secret key"
  input_type: json
  input_schema:
    type: object
    properties:
      token: { type: string }
      secret: { type: string, minLength: 32 }
    required: [token, secret]
  output: "Validated claims or error"
  output_type: json
  output_schema:
    type: object
    properties:
      valid: { type: boolean }
      claims: { type: object }

execution_context:
  working_directory: "/project"
  environment_variables:
    PYTHONPATH: "/project/src"
  required_tools: [python3, pytest]
  timeout_seconds: 300

acceptance_criteria:
  - "Validates JWT signature using HS256"
  - "Rejects expired tokens"
  - "Returns claims on valid token"

test_command: "pytest tests/test_jwt.py -v"

on_failure:
  max_retries: 2
  action: block

rollback: "git checkout HEAD~1 -- src/auth/jwt.py"

No ambiguity. No guesswork. Just executable specifications.

Philosophy

Think of Blueprint like an architect's drawings. An architect doesn't build the house — they produce specifications so precise that any competent builder can construct exactly what was envisioned.

We are the architect. The blueprint is our deliverable. Everything else is construction.

Fleet Consensus (5/5 Models Agree)

Five frontier AI models independently reviewed Blueprint and reached consensus on what makes it the universal standard:

Model Verdict Key Insight
Claude Opus 4.5 Approved "P0: Define output persistence and hand-off protocol"
GPT-5.2 Codex Approved "Canonical machine schema needed for whole document"
Gemini 3 Pro Approved "Orchestrator-mediated hand-off with capability matching"
Grok 4 Approved "File-based handshake protocol scales to 100+ agents"
DeepSeek Coder Approved "Add input_bindings syntax for explicit data flow"

Full fleet review available in session-journals/.

Documentation

License

Apache 2.0


Blueprint: Universal AI Orchestration Contract

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

blueprint_ai-1.3.0.tar.gz (161.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

blueprint_ai-1.3.0-py3-none-any.whl (67.2 kB view details)

Uploaded Python 3

File details

Details for the file blueprint_ai-1.3.0.tar.gz.

File metadata

  • Download URL: blueprint_ai-1.3.0.tar.gz
  • Upload date:
  • Size: 161.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.9

File hashes

Hashes for blueprint_ai-1.3.0.tar.gz
Algorithm Hash digest
SHA256 1c29d0b2147de002b3229f272275797aed1c3199656d475066c22ab9199b53be
MD5 63776816a1c4c735a80168a8efc18808
BLAKE2b-256 e8b8dda1159c74af2fa528722d430292f2b6f481ae105e6c76fcf496aee42438

See more details on using hashes here.

File details

Details for the file blueprint_ai-1.3.0-py3-none-any.whl.

File metadata

  • Download URL: blueprint_ai-1.3.0-py3-none-any.whl
  • Upload date:
  • Size: 67.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.9

File hashes

Hashes for blueprint_ai-1.3.0-py3-none-any.whl
Algorithm Hash digest
SHA256 94d80082327656e3ce47c547c47529fe7b594328e6e07814a08570050d32160c
MD5 8e501d7980811e3f907223f95459a70b
BLAKE2b-256 db5c274d714f44bf9f87989d2ddb22ab08f9ac26270c9c42fc4bc7135bfc2aa7

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page