Skip to content

Release 0.12.0

Choose a tag to compare

View all tags
@cdcavell cdcavell released this 28 Aug 16:19
· 63 commits to main since this release
v0.12.0
6454e4e
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

ASI Backbone Learning v0.12.0 is an advanced architecture release focused on what happens when governed execution extends beyond a single request, service, or policy boundary.

This release expands the Learning curriculum into cross-system authority, federated governance, distributed acknowledgment, human-readable decision explanation, adaptive risk and context drift, and durable decision evidence. It also substantially improves discovery of the repository's growing executable sample suite and publishes two new practitioner-oriented technical articles.

The recurring architectural question throughout this release is:

What must remain true when authority, context, policy, and execution are separated across systems or across time?

Highlights

Advanced Distributed Governance

Six substantial additions expand the Advanced learning area.

Cross-System Capability Exchange and Delegated Authority

Explores how narrowly scoped execution authority can cross a system boundary without becoming an ambient credential.

Topics include:

  • delegated authority
  • audience and resource binding
  • least privilege
  • trust relationships
  • replay resistance
  • downstream validation
  • authority attenuation
  • execution ownership

The material reinforces that transferring a representation of authority does not transfer unrestricted control.

Federated Governance and Independent Authority Coordination

Examines decisions involving multiple independently governed authorities.

Rather than assuming that every policy can be collapsed into one central rules engine, the material explores how independent authorities can coordinate while preserving their individual responsibility and decision boundaries.

Topics include:

  • authority-set resolution
  • independent policy ownership
  • coordination semantics
  • partial availability
  • conflicting outcomes
  • degraded operation
  • evidence across authorities

Distributed Acknowledgment and Continuation Workflows

Extends acknowledgment beyond the simple synchronous:

Prompt
  ↓
User accepts
  ↓
Execute

The new material examines acknowledgment as a continuation boundary that may span processes or time.

It covers:

  • acknowledgment records
  • continuation state
  • expiry
  • policy re-evaluation
  • context changes
  • replay
  • stale acknowledgment
  • execution-time validation

Acknowledgment remains distinct from standing authorization and from final execution authority.

Decision Explainability for Human Operators

Introduces a deliberate separation between internal governance evidence and the explanation presented to a human operator.

A governance system may possess detailed internal diagnostics that should not automatically be disclosed verbatim.

The new material explores:

  • explanation projection
  • operator-safe reason information
  • internal versus public detail
  • structured explanation models
  • decision provenance
  • avoiding accidental disclosure of sensitive internal state

The goal is explainability without weakening the underlying trust boundary.

Adaptive Risk, Context Freshness, and Drift

A decision that was valid when issued may no longer be safe when execution occurs.

This material examines governance as a time-sensitive process in which:

Valid earlier
    ≠
Automatically valid now

Topics include:

  • authoritative-context freshness
  • policy drift
  • risk changes
  • resource changes
  • freshness windows
  • execution-time re-evaluation
  • revocation
  • stale authority
  • fail-closed behavior when required facts can no longer be established

Durable Decision Ledgers and Cryptographic Audit Chains

Introduces a deeper treatment of durable governance evidence.

The material explores:

  • canonical record encoding
  • deterministic fingerprints
  • chained decision records
  • sequence validation
  • checkpoints
  • resumed verification
  • broken-link detection
  • verification confidence
  • evidence continuity

The documentation deliberately distinguishes tamper-evident structures from stronger claims such as absolute immutability or guaranteed non-repudiation.

New Executable Advanced Samples

Version 0.12.0 adds runnable companion projects for the advanced material:

  • cross-system-capability-exchange
  • federated-governance-coordination
  • distributed-acknowledgment-continuation
  • decision-explainability
  • adaptive-risk-context
  • durable-decision-ledger-audit-chain

Each sample remains intentionally small enough to expose the architectural boundary being taught.

The learning pattern continues to be:

Concept
   ↓
Runnable Sample
   ↓
Observable Boundary
   ↓
Invariant Test

The examples are teaching systems, not production distributed-governance infrastructure.

New Hands-On Lab

Compare Competing Policy Architectures

Adds a new lab centered on architectural comparison rather than implementation of a predetermined pattern.

Learners evaluate competing approaches based on concerns such as:

  • policy ownership
  • enforcement location
  • independent authority
  • policy distribution
  • availability
  • freshness
  • failure behavior
  • operational complexity
  • evidence requirements

The lab reinforces an established Learning principle:

Use the smallest architecture that preserves the boundaries you actually need.

New Practitioner Articles

When ASP.NET Core Authorization Is Not Enough

This new standalone article provides a decision guide for determining when ordinary ASP.NET Core authorization is sufficient and when a broader governed-execution workflow may be warranted.

It distinguishes ordinary authorization from concerns such as:

  • acknowledgment
  • delayed continuation
  • policy re-evaluation
  • scoped execution authority
  • policy provenance
  • execution-time context
  • host-owned consequential side effects

ASP.NET Core authorization remains the preferred solution when it already models the required trust boundary.

Why an AI Tool Call Is a Proposal, Not Authority

This article expands one of the repository's central AI integration principles:

The model may propose. The host retains execution authority.

A syntactically valid tool call is not treated as permission to perform a side effect.

The host remains responsible for:

  • tool registration and allowlisting
  • schema and semantic validation
  • authoritative resource identity
  • caller context
  • policy evaluation
  • credential custody
  • concurrency checks
  • admission and rate controls
  • retry and idempotency behavior
  • final execution
  • safe disclosure of failures

The core testing invariant remains:

Rejected or denied proposal
        ↓
Protected executor invocation count = 0

The article also distinguishes this from concurrency cases where an executor may legitimately be invoked but a conditional write refuses the final mutation.

Improved Executable Sample Discovery

The Executable Samples documentation has been substantially expanded to reflect the actual size of the sample suite.

The published experience now includes:

  • a complete sample catalog
  • architectural grouping of samples
  • dedicated docs/samples/toc.yml navigation
  • direct sample README links
  • run and test commands
  • links to related tutorials, labs, articles, and architecture material
  • left-hand navigation for individual samples
  • simplified breadcrumb behavior on the samples landing page

Samples are grouped broadly into areas such as:

  • Foundational
  • Governance and Policy Architecture
  • Security and Trust Architecture
  • ASP.NET Core Architecture

This makes the executable material easier to browse as a learning catalog rather than requiring readers to discover projects directly from the repository tree.

Roadmap and Navigation Updates

Version 0.12.0 refreshes:

  • ROADMAP.md
  • Advanced navigation
  • Articles navigation
  • Labs navigation
  • Executable Samples navigation
  • article backlog status
  • related cross-references throughout the Learning material

These changes reflect the growing maturity of the repository and its movement from foundational governed-execution concepts toward distributed and time-sensitive governance problems.

Publication and Archival Metadata

Release metadata has been synchronized for the new educational snapshot:

  • CITATION.cff → 0.12.0
  • .zenodo.json → 0.12.0
  • release date → 2026-08-28

This keeps GitHub release identity, citation metadata, and archival metadata aligned.

What v0.12.0 Represents

Earlier releases established the core governed-execution model:

Intent
   ↓
Authoritative Context
   ↓
Policy Decision
   ↓
Acknowledgment when required
   ↓
Scoped Authority
   ↓
Host-Owned Execution
   ↓
Evidence

Version 0.12.0 asks what happens when those boundaries become distributed:

Authority crosses systems
        ↓
Independent authorities coordinate
        ↓
Acknowledgment survives continuation
        ↓
Context changes over time
        ↓
Humans need understandable explanations
        ↓
Evidence must remain verifiable

This is the repository's strongest step yet into advanced governance architecture.

Release Scope

ASI Backbone Learning v0.12.0 remains an educational and architectural release.

It does not introduce:

  • a production governance engine
  • an interoperability or delegated-authorization standard
  • a security certification
  • a compliance guarantee
  • an immutable-ledger guarantee
  • an autonomous-agent runtime
  • an AI model
  • an AGI or ASI implementation

The executable projects are intentionally constrained teaching artifacts designed to make architectural boundaries observable and testable.

The repository continues to pursue its core purpose:

Teach architectural reasoning through explanations, executable examples, invariant tests, labs, comparisons, and explicit trust boundaries.

Assets 2
Loading