ACE 0.8.2 — Intelligence Catalog

Intelligence starts with an outcome

ACE 0.8.2 makes first use outcome-led. Atrium now opens by asking What intelligence do you want?, discovers every admitted onboarding profile through the governed Intelligence resource plane, and carries the selection into the existing Connect → Map → Watch → Brief → Activate journey.

What shipped

• A governed Intelligence Catalog for World, Market, and Custom Intelligence experiences.
• Domain-provided starter questions and source groups without domain vocabulary entering Core.
• Durable catalog discovery through the existing authenticated resource plane—not a second registry or state store.
• Selection that remains non-authorizing: choosing an experience cannot connect a source, start a monitor, or activate an effect.
• The unchanged Core 0.8 boundary: schema head v177, 24 resource kinds, and exactly eleven public MCP tools.

Install

pip install ace-core==0.8.2

or

uv add ace-core==0.8.2

World Intelligence 0.12.0 and Market Intelligence 0.7.0 are the paired domain releases and remain independently packaged. This patch does not claim the collaborative and multi-tenant completion planned for ACE 0.9.

See the changelog and roadmap for the detailed contract and boundaries.

ACE Core 0.8.1 — Decision-readable Intelligence

ACE 0.8.1 makes the Intelligence Operating System tangible as a product experience.

What changed

• Atrium now renders durable Intelligence Builder onboarding progress and opens the first Brief only when the governed session reaches first_briefing_ready.
• Intelligence, Opportunities, Agents, Connections, and Strategy are first-class destinations; downstream Work remains outside the command center.
• Opportunities are evidence-backed decision openings across Cases, material Shifts, and early Signals—not leads or tasks.
• Briefs and every decision-facing intelligence record become truthful visual What / Why / How / When blocks; missing materiality or event time is disclosed, and unknowns, limitations, receipts, and lineage remain inspectable.
• ACE now has its own blue-led design system, with customer branding kept as a replaceable deployment overlay.
• A local immutable resource replay powers the paired World AI Command Center demonstration through the same governed resource plane.
• Cited resource details remain warning-free across desktop and mobile.

Preserved boundaries

• schema head v177
• exactly 11 public MCP tools
• 24 domain-neutral Intelligence resource kinds
• unchanged separately packaged reference action adapter 0.4.0
• no new authority path, persistence path, autonomous connector access, or collaborative-tenancy claim

Verification

The release candidate passed the full Core, naked-kernel, security, Docker, package, clean-install, Canvas, desktop, and mobile gates. See the candidate evidence in docs/evidence/intelligence-os-v0.8.1-release-candidate-v1.md.

ACE 0.8.0 — Intelligence OS Realignment

ACE 0.8.0 — Intelligence OS Realignment

ACE 0.8.0 turns the Intelligence Builder foundation into one coherent Intelligence Operating System: connect sources, map the domain, watch for change, receive cited briefings, govern decisions and actions, observe outcomes, and improve through bounded feedback.

What is new

• One Intelligence resource plane across Sources, Connections, Observations, Entities, Signals, Shifts, Cases, Briefs, Monitors, Subscriptions, Agents, Decisions, Actions, Outcomes, Feedback, memory, and provenance.
• Atrium, rebuilt around intelligence with first-class Intelligence, Opportunities, Agents, Connections, and Strategy. Work stays downstream; Ask ACE answers from authorized intelligence and cites exact revisions.
• A complete builder stack: Connect → Map → Watch → Brief → Activate, governed composition through AC7, and authorized Agent Memory through AM4.
• Clear product boundaries: Core owns durable cognition, identity, state, authority, and outcomes; Intelligence owns domain-neutral sensing and orientation; inert Domain Packs supply vocabulary and policy; reviewed connectors/adapters own I/O and effects.
• Independent domain proof: World Intelligence demonstrates a live official-source path through reviewed action and measured feedback; Market Intelligence demonstrates a competitive-price path with an explicit no-action disposition.

Install

pip install ace-core==0.8.0

The optional reference action adapter is attached to this release as version 0.4.0 and requires ace-core>=0.8.0,<0.9.

Boundaries

ACE remains a self-hosted developer preview. Atrium is an optional repository-delivered workspace, not part of the Python wheel or a second source of truth. Domain Packs remain inert. Feedback is proposal-only and cannot apply itself. This release does not claim hosted operation, arbitrary autonomous effects, autonomous self-modification, or general real-world accuracy.

See the 0.8 release candidate evidence and public roadmap.

ACE 0.7.0 — Intelligence Builder Foundation

ACE 0.7.0 is the Intelligence Builder Foundation.

ACE is now framed for what it is becoming: an open, self-hosted Intelligence Operating System. The first product experience is the Intelligence Builder—connect sources, map the domain, watch for meaningful change, generate cited briefings, and activate approved actions.

The product journey

Connect → Map → Watch → Brief → Activate

0.7 establishes the governed foundation for that journey:

• five bounded onboarding agents for source discovery, concept mapping, monitoring, briefing, and activation
• stable generated Domain Packs with fail-closed compilation, compatibility, and conformance
• cited, editable concept models and first briefings
• exact approval and activation receipts with durable restart behavior
• Agent Composition AC1–AC7 for policy-admitted, measurable agent collaboration
• Agent Memory AM0–AM3 for authorized ingestion, reconciliation, recall, and context manifests
• independent World Intelligence and private B2B Market Intelligence consumer proof
• unchanged schema v177 and exact eleven-tool public MCP boundary

Why this matters

Reasoning is no longer the product category. It is one governed substrate alongside memory, learning, planning, state, authority, and outcomes. ACE owns the living intelligence lifecycle around models; LLMs remain optional inference providers rather than the system of record.

Install

pip install ace-core==0.7.0

Python 3.12 is required. ACE remains a developer preview: the foundation is real and reproducible, while the guided graphical Builder workspace is still a later milestone.

Verification

The release candidate passed all six GitHub gates: lint, security audit, fast tests, naked-kernel tests, Canvas typecheck/tests/build, and Docker build/health verification. Local cumulative verification also reproduced the packaged 0.7 metadata and exact eleven public tools.

Full details: CHANGELOG · Roadmap · Manifesto

ACE Core 0.6.0 — Measured Intelligence

ACE Core 0.6.0 — Measured Intelligence

ACE Core 0.6.0 adds the first durable, domain-neutral measured-impact slice. Under explicit product-owned criteria, ACE can classify a governed intelligence artifact or cognition revision as useful, harmful, or unproven, while preserving exact attribution, uncertainty, controls, provenance, and append-only replay.

Highlights

• Versioned criteria, conditions, evidence, evaluation, and governance-proposal contracts link an exact evaluand to material use, Decision, reviewed Action, observed result, Outcome, controls, and cutoff provenance.
• Missing attribution, mismatched conditions, future leakage, unavailable outcomes, duplicate evidence, and insufficient support fail closed or remain explicitly unproven.
• Evaluation and proposal receipts are appended atomically and replay exactly across restart; divergent replay and scope drift are rejected.
• Promotion, rejection, rollback, and retirement remain proposals only. They are non-effective, non-selectable, and never silently applied.
• Owner-governed Monitor and Subscription lifecycle contracts support durable create, pause, resume, revoke, and explicitly requested sensing windows without granting scheduler, source-transport, delivery, or external-action authority.
• The separately packaged reference workspace action adapter advances to distribution version 0.2.0 and requires ace-core>=0.6.0,<0.7; its capability implementation identity remains 0.1.0.

Install Core with:

pip install ace-core==0.6.0

The reference adapter wheel and source archive are attached to this GitHub Release rather than published as part of the Core distribution.

Boundary and non-claims

This release proves a bounded matched-evidence mechanism, not general causality, universal benefit, self-certification, automatic promotion, distributed exactly-once effects, or untrusted-code isolation. Issue #49 F3 remains open under its documented containment with a 2026-11-05 deadline. The World Intelligence public-artifact reproduction is a separate external release gate and will be recorded independently.

See the repository changelog and Measured Intelligence work-packet/evidence records for the exact supported boundary.

ACE Core 0.5.0 — Reasoning into Action

ACE Core 0.5.0 — Reasoning into Action

ACE can now carry an authorized decision beyond reasoning into a governed, bounded action lifecycle.

What shipped

• durable action records with explicit lifecycle states
• effect-free planning and exact human review before execution
• trusted, registered action adapters with bounded authority
• honest terminal states, verification, repair, and separate promotion
• restart-safe action recovery, attempt history, replay protection, cancellation, and resource receipts
• a separately packaged ace-reference-workspace-action 0.1.0 adapter demonstrating the public action contract
• updated public positioning and acceptance evidence for the Core → Intelligence → Domain architecture

Distribution

• Install Core from PyPI: pip install ace-core==0.5.0
• The reference action adapter wheel and source distribution are attached to this GitHub release as independent artifacts.

Supported boundary

This release supports one ACE host, one durable store, and trusted in-process adapters. It does not claim arbitrary third-party plugin safety, distributed execution, exactly-once delivery across external systems, or autonomous self-modification.

The external release proof is the World Intelligence live reality-to-action journey, reproduced using only these public artifacts.

ACE Core 0.4.4 — task-source provenance compatibility

ACE Core 0.4.4 restores governed-cognition teaching from freshly completed tasks whose nested evidence receipts contain database-native timestamps.

What changed:

• task-source evidence is normalized to JSON-native values before canonical hashing
• deterministic source provenance is preserved
• no schema migration; schema remains v176
• public APIs and stored contracts remain compatible with 0.4.3

Verification:

• focused cognition API/CLI/restart suite: 11 passed
• normal release gate: 7,378 passed
• extension-disabled release gate: 7,376 passed
• all six GitHub checks green
• isolated wheel install and cognition CLI smoke passed

ACE Core 0.4.3 — canonical cognition persistence

Governed Cognition persistence correctness

ACE Core 0.4.3 fixes exact governed-cognition replay across durable database round trips.

A real external Market Intelligence consumer exposed the defect: SurrealDB flexible-object normalization could omit nested optional values, causing content-derived proposal validation to fail closed after restart. Core now persists canonical JSON beside queryable object projections for cognition identities, revisions, heads, proposals, review receipts, selection receipts, and use receipts.

Compatibility

• canonical hash semantics are unchanged
• pre-v176 object-only records remain readable
• schema v176 is additive
• the eleven-tool MCP surface is unchanged
• Core + Intelligence and Domain Pack boundaries are unchanged

Verification

• real SurrealDB proposal → approval → restart → discovery → material use → lifecycle regression
• 7,390-test normal Core gate
• 7,376-test extension-disabled gate
• official lint, security, Canvas, Python, naked-kernel, and Docker checks
• isolated wheel/sdist build and CLI smoke test

GC1 remains active until the clean public 0.4.3 external-consumer journey, failure controls, restart, and attribution evidence are complete.

Full changelog: v0.4.2...v0.4.3

ACE Core 0.4.2 — Governed Cognition Builder

What's Changed

• Close GI2 with public cross-domain evidence by @eamirian in #61
• Add governed cognition builder CLI by @eamirian in #62
• Release ace-core 0.4.2 by @eamirian in #63

Full Changelog: v0.4.1...v0.4.2

ACE Core 0.4.1 — Governed Cognition Evidence Closeout

ACE Core 0.4.1 closes the 0.4.x governed-intelligence release line with a usable public contract surface and the evidence required to audit it.

What shipped

• Re-exports the per-statement epistemic-status and Case-bound Brief synthesis contracts from the supported ace.intelligence surface.
• Adds 23 additive public names without changing contract identities, schemas, or record behavior.
• Restores MANIFESTO.md, the P1/P2 platform evidence archive, and the roadmap narrative that explains the Core / Intelligence / Domain Pack boundary.
• Adds evidence-index integrity and release-positioning regression tests.
• Aligns uv.lock with the 0.4.1 project version.

Verification

GitHub CI is green across:

• lint and formatting
• fast Python test gate
• naked kernel with zero extensions
• security audit
• Canvas typecheck, tests, and build
• Docker build and health verification

The local full non-e2e run recorded 7,369 passed, 50 skipped, and 259 deselected. One SurrealDB teardown write conflict passed on isolated retry.

Scope

This release establishes the packaged Core + Intelligence substrate and its governed-cognition evidence. It does not promote GI2 domain neutrality by itself. That gate remains open until a tagged, packaged World Intelligence release and a clean-install two-domain reproduction both pass.

Changes

• Public contract surface: #57
• 0.4.1 release preparation: #58
• GI1/GI2 roadmap correction: #59
• Evidence and roadmap closeout: #60