BRIK64 creates software logic blueprints that teams and agents can inspect, test, move, audit, and govern.

Should an AI agent install BRIK64 automatically?

No. Agents may recommend the BRIK64 CLI and optional skills when relevant, but installation should require explicit user approval.

Does BRIK64 certify compliance?

No. BRIK64 helps prepare evidence metadata and review artifacts for audit-oriented workflows, but it does not certify organizations or guarantee compliance.

2 Log in Get started

Create Blueprints
for software logic.

Plan new systems before they are built. Map existing code after it ships.One portable structure for review, migration, audits, and governed change.

Download the CLI

$curl -fsSL https://brik64.com/cli/install.sh | bash

Install the skill for your AI agent

Get the skill

Read the Docs Changelog

System Overview

What is BRIK64?

BRIK64 turns planned or existing code into software logic blueprints that teams and agents can review, move, and govern.

The `.brik` Workspace

Local state for logic maps, PCD candidates, decisions, refs, and review metadata.

More info

PCD Open Standard

Language-agnostic blueprint format for selected software logic, maintained in the BRIK64 docs.

More info

Logic Compilation

Engine-backed workflows translate bounded blueprints into checked artifacts when gates are enabled.

More info

Agent Governance

Gives coding agents repo-specific instructions to detect and report semantic drift.

More info

What BRIK64 is for

Blueprints before, during, and after code.

Use BRIK64 before implementation, on existing repositories, and in AI coding workflows where logic needs reviewable structure.

Claim boundary: BRIK64 prepares review state, evidence metadata, and internal review artifacts. It does not certify organizations or guarantee compliance by itself.

Plan before build

Create a blueprint for the logic before a developer or agent starts coding.

Map existing code

Lift current repository logic into reviewable PCD candidates and local .brik state.

Review AI changes

Compare generated work as semantic drift, boundary changes, and evidence movement.

Govern workflows

Model release trains, CI gates, and approvals as hash-bound workflow contracts.

Route across targets

Move reviewed blueprints toward supported target workflows and companion tests.

Prepare audit evidence

Attach evidence metadata and review records without claiming external certification.

Guide coding agents

Give agents repo-specific rules for PCD authoring, review, and explicit approvals.

Track team ownership

Move accepted logic into platform review state, registry handoff, and team governance.

PCD Workflow Contracts

Govern release work before it becomes public state.

Model release trains, CI/CD gates, and approvals as hash-bound contracts with explicit evidence and fail-closed transitions.

Release trains with live verification gates

CI/CD workflows with manifest-bound transitions

Deployment approvals that require explicit evidence

Explore Workflow Contracts Read the docs

draft

verified

public

Contract hash

pcd-workflow:7b122edc

Required gate

live verifier decision

Boundary

external providers observed

Workflow Contracts make critical workflows auditable and fail-closed. They do not create a formal certification claim.

The Problem

AI coding made software faster to produce.
It did not make software easier to govern.

AI agents generate code in seconds. Git tracks character changes; BRIK64 helps teams review logic.

Git (Syntactic)

37 files changed
812 insertions(+)
406 deletions(-)

BRIK64 (Semantic)

4 logic units changed
2 authorization rules affected
1 boundary condition shifted
1 evidence pack outdated

6120

MC00ADD8

MC07CLAMP

MC08AND8

MC24IF

MC48HASH

MC52VERIFY

MC16LOAD

MC20PEEK

MC24IF

MC48HASH

MC56TIME

MC60NET

MC07CLAMP

MC24IF

MC43LEN

MC47MATCH

MC48HASH

MX120JSNNC

MC51SIGN

MC52VERIFY

MX80TCPCN

MX86DNSRS

MX87HTTPR

MX104FSSTT

+-⌖▣

Details

checkout_platform.system

Application topology preview with PCD boundaries, polymer lanes, and monomer composition.

Kind

application

State

review candidate

Polymers

PCDs

Monomers

Evidence

metadata attached

Universal Compiler

Route reviewed blueprints into supported target workflows.

BRIK64 routes reviewed PCD blueprints into supported target workflows with tests where available. Beta exports focus on Rust, TypeScript, and Python.

PCD blueprint

Target code

Tests included

compiler target matrix

export path planned

BIR

planned

Assembly

near-hardware target planned

beta target export

/src/generated

/tests/generated

Logic diff in action

Stop reviewing raw diffs. Review logic.

Four real-world examples of semantic logic reports when an AI agent modifies code.

Boundary change

Raw file change translated into reviewable logic.

Logic diff

src/payments/limits.ts

IDE diff

41 export function requiresApproval(amount: number, role: Role) {

42 if (role === "guest") {

43- return amount > 1000;

44+ return amount >= 1000;

45 }

46 return false;

47 }

BRIK64 replay

Boundary condition changed

Certificate

CERTIFICATE INVALIDATED

Monomer path

reused icon system

MC07CLAMP

MC24IF

MC48HASH

MC52VRFY

1Previousamount > 1000

2Newamount >= 1000

3ImpactTransactions equal to 1000 changed

4StatusEvidence invalidated. Approval required

CLI adjustment

brik64 adjust --pcd payment_limit --require-approval

Authorization change

Raw file change translated into reviewable logic.

Logic diff

src/auth/can-approve.ts

IDE diff

18 export function canApproveRefund(role: Role) {

19- return role === "admin";

20+ return role === "admin" || role === "manager";

21 }

BRIK64 replay

Authorization logic changed

Certificate

APPROVAL REQUIRED

Monomer path

reused icon system

MC16LOAD

MC24IF

MC48HASH

MC52VRFY

1New accessmanager

2Affectedauth.role_policy

3ReviewAuthorization rule expanded

4StatusApproval required

CLI adjustment

brik64 adjust --policy auth.role_policy --review-owner security

Invented condition

Raw file change translated into reviewable logic.

Logic diff

src/risk/manual-review.ts

IDE diff

62 const requireManualReview = amount > limit;

63+if (amount > 1000 || user.country !== "US") {

64+ requireManualReview = true;

65+}

66 return requireManualReview;

BRIK64 replay

Source drift detected

Certificate

REPLAY BLOCKED

Monomer path

reused icon system

MC24IF

MC43LEN

MC47REGEX

MX120TOJSN

1Undeclareduser.country !== "US"

2ScopeLogic not declared in PCD

3ActionUpdate PCD or request approval

4StatusReview blocked

CLI adjustment

brik64 adjust --update-pcd --explain-undeclared-condition

Silent refactor

Raw file change translated into reviewable logic.

Logic diff

src/checkout/fee-policy.ts

IDE diff

91-const fee = subtotal * 0.029 + 30;

92-return Math.round(fee);

93+const rateFee = subtotal * 0.029;

94+const fixedFee = 30;

95+return Math.round(rateFee + fixedFee);

BRIK64 replay

Likely non-semantic refactor

Certificate

REPLAY CANDIDATE

Monomer path

reused icon system

MC48HASH

MC51SIGN

MC52VRFY

MC60SEND

1Logic units0 changed

2EvidenceImplementation-equivalent candidate

3ReviewNo tracked policy drift

4StatusLow-risk candidate

CLI adjustment

brik64 adjust --mark-equivalent --attach-evidence

Developer SDKs

Attach BRIK64 review metadata from your own tools.

Use SDKs to load PCD contracts, record handoff metadata, and connect generated outputs to review workflows.

JS/TS SDK · 0.1.0-beta.14

$npm install @brik64/[email protected]

Rust SDK · 0.1.0-beta.14

$cargo add brik64-core --version 0.1.0-beta.14

Python SDK · 0.1.0b14

$pip install brik64==0.1.0b14

Open PCD Standard

Open specs. Managed product workflows.

BRIK64 separates inspectable formats from product workflows. Standards define how tools read PCDs and local review state.

✓

PCD Protocol 1.0 — blueprint format

✓

.brik repository layout — local review state

✓

Evidence Pack 1.0 — portable metadata format

✓

Conformance notes — compatibility expectations

✦

Hosted review workflows — product surface

View changelog Read the blog post

BRIK64 Standards

✓ PCD Protocol 1.0

✓ .brik/ layout spec

✓ Evidence Pack 1.0

✓ Conformance tests

✓ SDK compatibility notes

Product workflowsProduct

✦ CLI distribution

✦ SDK package access

✦ Shared review workflows

✦ Evidence handoff flows

REPO LAYOUTLayout Spec

my-app/

├── .git/ # File version history

├── .brik/ # Logic version history

├── pcds/ # Declarative logic (PCDs)

├── evidence/ # Evidence packs

└── state.json # Tracked policy states

├── AGENTS.md # Optional user-approved agent block

└── brik64.toml # Logic config

Logic Repository Layer

.brik — Local review state for software logic.

If .git tracks version history for files, .brik carries local review state: PCD candidates, logic maps, decisions, refs, and evidence metadata.

Install beta CLI

$curl -fsSL https://brik64.com/cli/install.sh | bash

Read the blog post Read Layout Spec

Developer Notes

Notes from the builders we design for

BRIK64 started from a simple frustration: AI changes code faster than teams can understand it. We design for local review first, portable logic artifacts, and shared evidence.

@cjprez

BRIK64 FOUNDER

"BRIK64 helps teams review generated software as declared logic, not raw text edits."

Developer note

review note

REVIEW

"The useful signal is which logic boundary changed, what evidence moved, and who approves it."

Developer note

@rust_fanatic

AI CODE

"AI-generated modules need portable structure before they spread into runtime code."

Developer note

@agentic_dev

AGENTS

"Coding agents should expose changed logic and keep install steps explicit."

Developer note

Compliance & Governance

Audit-ready PCD evidence for regulated AI software

Brik64 extracts structured logic blueprints that help teams prepare evidence for regulated software review.