From d48be6af9d9bf1a55d74f02b59b26986eb2c7b87 Mon Sep 17 00:00:00 2001 From: Edwin Amirian Date: Wed, 12 Aug 2026 14:45:01 -0700 Subject: [PATCH 1/3] docs: launch ACE 0.8 Intelligence OS realignment --- ROADMAP.md | 70 +++-- docs/README.md | 4 + ...ce-os-realignment-v0.8.0-work-packet-v1.md | 241 ++++++++++++++++++ tests/test_public_roadmap_positioning.py | 15 +- 4 files changed, 304 insertions(+), 26 deletions(-) create mode 100644 docs/design/intelligence-os-realignment-v0.8.0-work-packet-v1.md diff --git a/ROADMAP.md b/ROADMAP.md index 81a550e..bc2d2fb 100644 --- a/ROADMAP.md +++ b/ROADMAP.md @@ -179,7 +179,7 @@ Release targets describe sequence, not calendar commitments. ACE follows three r | 0.5.0 | Reasoning into Action | **Passed** | T1 and B1 pass for bounded attributable action under the explicit single-host, trusted-adapter topology; I1 authority receipts govern the execution-adapter slice | [#37](https://github.com/augmented-cognition-engine/core/issues/37) | | 0.6.0 | Measured Intelligence | **Passed** | Public Core 0.6.0 and World 0.10.0 artifacts reproduce exact useful/harmful/unproven classification, matched leakage-bounded controls, durable replay, and proposal-only governance under product-owned criteria; SI4 remains a broader future outcome | [#38](https://github.com/augmented-cognition-engine/core/issues/38) | | 0.7.0 | Intelligence Builder Foundation | **Passed** | Public Core 0.7.0 composes 0.7A–0.7E, AC1–AC7, and AM0–AM3; all release checks, trusted publication, public clean-install reproduction, and independent World/Market consumer gates passed | [#39](https://github.com/augmented-cognition-engine/core/issues/39) | -| 0.8.0 | Intelligence Workspace | **Later** | Builds on G1 and IA-R1 to expose the I1–I3, E1, B1, L1, and SI1–SI4 lifecycle through a coherent permission-aware human experience | [#40](https://github.com/augmented-cognition-engine/core/issues/40) | +| 0.8.0 | Intelligence OS Realignment | **Next** | Makes the evidence → orientation → decision/action → outcome → governed-feedback lifecycle coherent in code, public resources, Atrium, and independent World/Market demonstrations | [#40](https://github.com/augmented-cognition-engine/core/issues/40) | | 0.9.0 | Collaborative Intelligence | **Later** | Advances H1, SI3 sensitive-source governance, and the remaining T1/E2 operational guarantees across tenancy, shared authority, privacy, portability, recovery, and managed operation | [#41](https://github.com/augmented-cognition-engine/core/issues/41) | | 1.0.0 | Intelligence Operating System | **Later** | Stabilizes the complete supported loop and all milestone-critical contracts across Core, Intelligence, Domain Packs, connectors, product surfaces, governance, action, outcomes, operation, and portability | [#42](https://github.com/augmented-cognition-engine/core/issues/42) | @@ -494,25 +494,44 @@ pack material or required knowledge of ACE internals; independent packs, connect extensions pass their applicable public compatibility and conformance checks; requested source scope and authority are visible before activation; and failures remain isolated and diagnosable. -### 0.8.0 — Intelligence Workspace - -Public promise: builders and operators can inspect and govern the full intelligence lifecycle from one -coherent, permission-aware workspace. - -- Evolve the G1 and IA-R1 read foundation into navigable views of context, evidence, provenance, - decisions, corrections, cognition revisions, actions, outcomes, uncertainty, and conflict. -- Add permission-aware case files for standing investigations, as-of orientations, claim versus - action assessments, reaction timelines, competing hypotheses, unknowns, watch conditions, - revision diffs, and entitled source deep links. -- Expose approvals and other governed controls only when their underlying public authority and - runtime contracts are supported. -- Drive workspace behavior through the same public interfaces available to products and extensions. -- Preserve accessibility, stable identity, visible degraded behavior, and the durable records as - the source of truth. - -Release gate: one end-to-end journey can be understood, inspected, and governed from the workspace -without reading internal implementation details. The workspace is optional and does not become a -second source of truth. +### 0.8.0 — Intelligence OS Realignment + +Public promise: ACE presents one coherent Intelligence Operating System. Governed data enters, +evidence-grounded intelligence is derived, people and machines consume it downstream, and +attributable outcomes return through governed feedback. Reasoning is a governed component of that +lifecycle rather than the product category. Atrium is its permission-aware intelligence dashboard +and human control plane. + +The bounded delivery sequence is: + +- **0.8A — architecture and compatibility:** freeze the canonical lifecycle, ownership map, + compatibility seams, work packets, and golden acceptance journey. +- **0.8B — runtime realignment:** enforce Core, Intelligence, Domain Pack, adapter/plugin, and + application ownership in code while preserving supported contracts through declared migration. +- **0.8C — unified intelligence resources:** expose one public resource plane for sources, entity + state, Observations, Signals, Shifts, Cases, Briefs, monitors, subscriptions, decisions, actions, + outcomes, feedback, provenance, uncertainty, revisions, Context Manifests, and memory lineage. +- **0.8D — Atrium experience:** rebuild the optional dashboard and control plane around attention, + intelligence review, investigation, semantic diffs, governed controls, and Ask ACE over the same + durable resources. +- **0.8E — cross-domain proof:** run a public World AI Command Center and the parallel Market + Intelligence command center through unchanged Core + Intelligence APIs. +- **0.8F — release acceptance:** pass compatibility, migration, restart, installed-artifact, + accessibility, entitlement, degraded-state, and two-domain publication gates. + +Canonical current architecture and product surfaces replace the octopus, arm-growth, MAKE/SHIP, +partner-team, committee-first, product-manager-toolbox, and Living Product Graph system-category +narratives. Historical evidence and supported compatibility identities are not rewritten. + +The complete packet, compatibility map, landing order, stop conditions, and signature demonstration +are frozen in the +[Intelligence OS Realignment work packet](docs/design/intelligence-os-realignment-v0.8.0-work-packet-v1.md). + +Release gate: World and Market complete the same governed evidence-to-outcome lifecycle without a +domain branch in Core or Intelligence; Atrium and a machine interface consume the same public +durable intelligence; supported compatibility and migrations are tested; and accessibility, +authority, entitlement, restart, failure, and degraded behavior pass. Atrium remains optional and +does not become a second source of truth. ### 0.9.0 — Collaborative Intelligence @@ -683,7 +702,7 @@ flowchart LR | 4. Reasoning into action | Can an approved decision safely produce attributable work? | **Passed** | T1 and B1 passed in 0.5.0 for the bounded single-host, trusted-adapter topology | 0.5.0 | | 5. Measured intelligence | Can ACE prove when retained intelligence or a capability helped, hurt, or remains unproven? | **Passed for the public 0.6.0 boundary** | Product-owned matched evaluation, exact attribution, durable replay, and proposal-only governance are public; F2 and general benefit remain not ready | 0.6.0 | | 6. Intelligence Builder foundation | Can any product connect sources, specialize, operate, and retain ownership of ACE intelligence? | **Passed for the bounded public 0.7.0 contract** | 0.7A–0.7E, AC1–AC7, and AM0–AM3 compose in the public artifact; independent World/Market activation gates passed, while H1 remains later | 0.7.0 and 0.9.0 | -| 7. Human experience | Can people inspect and govern the full loop without learning ACE internals? | **Read-only and governance foundations passed** | IA-R1, G1, E1, and L1 passed; the writable workspace remains bounded by B1 and H1 | 0.8.0 | +| 7. Intelligence OS experience | Can people and machines consume and govern one coherent intelligence loop without learning ACE internals? | **0.7 foundation passed; 0.8 realignment next** | IA-R1, G1, E1, L1, the 0.7 agents, AC7, and AM3 are inputs; unified resources, Atrium, compatibility, and cross-domain proof remain | 0.8.0 | | 8. Continuous situational intelligence | Can ACE maintain a trustworthy, changing orientation over any bounded subject without making its domain ontology part of Core or Intelligence? | **GI1 substrate passed; product outcome not ready** | SI1–SI4 require GI1, independent Domain Packs, K1–K3, I1–I3, E2, L1, F2 where justified, and the workspace and collaboration slices of H1 | cross-cuts 0.4.x–0.9.0; complete by 1.0.0 | ## Immediate dispatch @@ -714,10 +733,11 @@ System identity and exact eleven-tool MCP surface. The binds the tag, artifacts, hashes, workflow, consumer gates, installation, limitations, and rollback boundary. -Exit condition: **passed.** The next release milestone is 0.8.0 Intelligence Workspace: expose the -living intelligence lifecycle as a coherent, permission-aware human experience. 0.9 collaboration -and the complete 1.0 Intelligence Operating System contract remain later; 0.7 does not silently -claim either. +Exit condition: **passed.** The next release milestone is 0.8.0 Intelligence OS Realignment: make +the living intelligence lifecycle coherent in runtime ownership, public resources, Atrium, and +World/Market product proof. The bounded A–F sequence begins with the +[0.8 work packet](docs/design/intelligence-os-realignment-v0.8.0-work-packet-v1.md). 0.9 +collaboration and the stable complete 1.0 contract remain later; 0.7 does not silently claim either. ### Completed technical prerequisite: bounded Product State outcomes diff --git a/docs/README.md b/docs/README.md index da82c89..88f8f33 100644 --- a/docs/README.md +++ b/docs/README.md @@ -71,6 +71,10 @@ support the public roadmap but do not compete with it for outcome state or dispa - [Intelligence Builder onboarding sequence](design/guided-intelligence-bootstrap-v0.7.0-work-packet-v1.md) — the cumulative 0.7A–0.7E Connect → Map → Watch → Brief → Activate contracts, authority boundaries, state machine, and full-demo acceptance. +- [Intelligence OS Realignment](design/intelligence-os-realignment-v0.8.0-work-packet-v1.md) — the + 0.8A–0.8F canonical lifecycle, ownership and compatibility map, Atrium and public-resource + sequence, World AI Command Center demonstration, Market falsifier, release gates, and stop + conditions. - [Intelligence Builder Connect work packet](design/intelligence-builder-connect-v0.7.0-work-packet-v1.md) — the bounded 0.7B Connection Agent implementation, reuse audit, failure controls, and evidence plan. diff --git a/docs/design/intelligence-os-realignment-v0.8.0-work-packet-v1.md b/docs/design/intelligence-os-realignment-v0.8.0-work-packet-v1.md new file mode 100644 index 0000000..c54f581 --- /dev/null +++ b/docs/design/intelligence-os-realignment-v0.8.0-work-packet-v1.md @@ -0,0 +1,241 @@ +# ACE 0.8.0 Intelligence OS Realignment work packet + +Status: **active architecture and compatibility packet** +Public milestone: [issue #40](https://github.com/augmented-cognition-engine/core/issues/40) +Release input: `main@11b44e84d92e0674fa433103779f4050eeca2725` (`ace-core==0.7.0`) +Target: `ace-core==0.8.0` + +## Outcome + +ACE 0.8.0 must make the product and the implementation tell the same story: + +> ACE is an open, provider-neutral Intelligence Operating System. The Intelligence Builder is its +> first guided experience. Reasoning, memory, planning, learning, graph state, authority, and +> optional models are governed substrate inside the system rather than separate product arms. + +The user-visible builder journey shipped in 0.7 remains intact: + +```text +Connect → Map → Watch → Brief → Activate +``` + +0.8 makes the complete operating loop coherent and inspectable: + +```text +authorized evidence +→ Observation +→ temporal Entity state +→ Signal / Shift +→ Case / Brief +→ Decision / authorized Action +→ observed Outcome +→ governed Feedback proposal +→ explicit review and later use +``` + +This is a realignment release, not a rename-only release and not permission for a mass rewrite. +Existing supported contracts remain available through declared compatibility adapters, +deprecations, and migrations until their removal policy permits otherwise. + +## Release truth and preserved inputs + +The 0.8 line begins only from the published 0.7.0 source and evidence closeout. A local dirty +checkout, an unmerged branch, or a later Agent Memory or Agent Composition candidate is not an +implicit release dependency. In particular, AM4 and AM6 may be evaluated as bounded inputs, but +they enter 0.8 only through an explicit compatibility review and ordinary public review authority. + +The following 0.7 capabilities are released inputs rather than work to rebuild: + +- separate Connection, Ontology, Intelligence, and Briefing Agent application services; +- exact governed Domain Activation plan admission for the Activate stage; +- the restart-safe Connect → Map → Watch → Brief → Activate acceptance path; +- inert generated Domain Pack schema, compiler, compatibility, conformance, and activation; +- governed Agent Composition through AC7; +- authorized Agent Memory and Context Manifest lineage through AM3; +- independent World and private Market activation consumers; and +- the exact eleven-tool thin MCP boundary. + +0.8 turns those capabilities into one supported experience. It does not pretend they were absent, +nor does it treat provider-free reference strategies and test fixtures as production connectors or +a finished UI. + +## Constitutional ownership + +### Core owns governed mechanics + +Core owns product-scoped identity, temporal and immutable state, graph mechanics, transactions, +provenance, authority, planning, execution admission, assurance receipts, decisions, actions, +outcomes, feedback governance, provider-neutral cognition, and restart/failure semantics. Core +must not know that an opaque reasoning request is producing a Brief or what a domain entity means. + +### Intelligence owns the evidence-to-orientation lifecycle + +Intelligence owns Observation, Entity projection, Signal, Shift, Case, Brief, Monitor, +Subscription, attention, resolution, delta detection, synthesis policy enforcement, Domain Pack +compilation, activation bindings, feedback proposals, and rebuildable projections over Core state. +Those projections are derived caches, never a second authoritative persistence engine. + +### Domain Packs own declarative meaning and policy + +Packs own ontology, aliases, source mappings, watched attributes, detector configuration, +materiality, personas, routing, analytical policy, templates, permitted action parameters, +evaluation fixtures, compatibility ranges, and limitations. Packs contain no network access, +secrets, imports, imperative control flow, or executable templates. + +### Adapters and strategy plugins own executable specialization + +Connectors acquire and translate authorized sources. Action and delivery adapters perform bounded +external effects. Strategy plugins may supply specialized algorithms that cannot be expressed +declaratively. They are separately versioned, explicitly enabled, least-authorized, and do not +become Core or Domain Pack code. + +### Applications own the experience + +Atrium, domain applications, APIs, agents, workflows, and mature downstream products render and +operate the same public intelligence resources. An application does not gain authority by +rendering a control and does not become another source of truth. + +## Compatibility map + +| Existing surface | 0.8 canonical interpretation | 0.8 treatment | +|---|---|---| +| `ace.core` | governed mechanics and control plane | preserve and strengthen inward dependency boundaries | +| `ace.intelligence` | evidence-to-orientation domain-neutral lifecycle | preserve; complete the stable resource/API facade | +| `ace.application` | orchestration across Core and Intelligence | preserve as application services; prevent transport and domain nouns from leaking inward | +| `core/engine/orchestration`, `orchestrator`, `cognition` | provider-neutral reasoning and composition substrate | retain behind Core-facing services; stop presenting it as the product category | +| `core/engine/arms` MAKE implementations | downstream artifact-production strategies | freeze as experimental compatibility implementations; migrate behind explicit adapter or strategy contracts | +| SHIP arm and gates | assurance, verification, delivery, and promotion | separate the assurance contract from delivery effects; preserve compatibility while canonical vocabulary changes | +| Living Product Graph | historical name for a bounded product-state projection | retain compatible reads; introduce canonical intelligence/entity-state views without rewriting historical evidence names | +| broad engine MCP | experimental host and compatibility surface | do not confuse with the exact eleven-tool public boundary; route new product work through supported APIs | +| Atrium / Canvas | optional permission-aware intelligence dashboard and control plane | rebuild as a public-API consumer; no direct authority or independent state | +| extension registry | explicitly enabled executable capabilities | preserve compatibility while separating declarative packs, connectors, strategy plugins, and action adapters | + +Historical evidence titles and immutable receipts are not rewritten. Canonical current documents, +onboarding, diagrams, runtime labels, and product surfaces must stop teaching the octopus, +partner-team, committee-first, product-manager-toolbox, Living Product Graph as the system category, +or MAKE/SHIP arms as the architecture. + +## Bounded delivery sequence + +### 0.8A — architecture, compatibility, and release control + +- Reconcile the public roadmap and milestone issue to Intelligence OS Realignment. +- Freeze this ownership map, lifecycle, compatibility policy, and release acceptance journey. +- Inventory old canonical vocabulary and every runtime boundary it currently names. +- Define additive migration seams before moving or renaming implementation. +- Keep the eleven-tool public MCP surface unchanged unless a separate contract decision passes. + +Exit: maintainers can place every current capability and planned change on one side of a declared +boundary, with an owner, compatibility treatment, landing order, and rollback. + +### 0.8B — runtime boundary realignment + +- Put generic planning, authorization, execution admission, assurance receipts, and outcome + governance behind Core services. +- Put Observation-to-Feedback lifecycle behavior behind Intelligence services. +- Move artifact creation and external effects behind explicit adapters or strategy plugins. +- Remove inward imports and duplicate persistence/authority implementations before removing names. +- Add compatibility tests for retained public paths and migrations. + +Exit: the dependency graph enforces the constitutional ownership without a domain-specific branch +in Core or Intelligence and without breaking the 0.7 builder journey. + +### 0.8C — unified intelligence resource plane + +- Publish one supported resource/query facade for Sources, source health, Entities, Observations, + Signals, Shifts, Cases, Briefs, Monitors, Subscriptions, Decisions, Actions, Outcomes, Feedback, + evidence lineage, uncertainty, conflicts, semantic revisions, Context Manifests, and memory use. +- Give every resource stable product-scoped identity, authorization, as-of semantics, pagination, + degraded state, revision lineage, and exact provenance. +- Serve Atrium and at least one machine interface from the same contracts. + +Exit: UI-specific state is disposable and can be rebuilt entirely from supported public resource +queries and commands. + +### 0.8D — Atrium intelligence experience + +- Build an intelligence-review inbox and a briefing-first home. +- Make Intelligence, Opportunities, Agents, Connections, and Strategy first-class navigation; + treat Work as downstream execution. +- Add source health, entity exploration, Signals/Shifts, Cases, Briefs, evidence lineage, + uncertainty/conflict, semantic revision diff, Context Manifest and memory lineage, Decisions, + Actions, Outcomes, Subscriptions, and degraded-state views. +- Make Ask ACE a governed intelligence query over the same resources and citations. +- Retain whiteboard investigation as a Case workspace, never a second source of truth. +- Meet keyboard, screen-reader, contrast, responsive, loading, empty, error, and denied-state gates. + +Exit: a user can understand, review, correct, approve, and trace the complete loop without learning +ACE package boundaries or hand-authoring a Domain Pack. + +### 0.8E — live cross-domain product proof + +- World Intelligence provides the public AI Command Center journey using changing authorized + sources: model/capability releases, token economics, cybersecurity, regulation/geopolitics, + investment, market response, and attributable executive commitments. +- Market Intelligence provides the parallel commercial intelligence command-center journey. The + public pack remains generic; HPE-specific sources, vocabulary, and policy remain a private + organization overlay/deployment. +- Both applications use unchanged Core + Intelligence APIs and materially different ontology, + sources, detectors, personas, epistemic policy, and cadence. + +Exit: each domain connects sources, maps concepts, activates monitoring, emits or suppresses a +material Shift, produces a cited Brief, records a Decision or bounded Action, admits a later +Outcome, proposes governed Feedback, and reproduces the state after restart. + +### 0.8F — release acceptance and publication + +- Run focused, full, package, schema, migration, naked-kernel, exact-eleven, restart, installed- + artifact, accessibility, entitlement, degraded-state, and cross-domain gates. +- Reproduce Atrium and a machine consumer against the same public package and durable state. +- Publish exact artifacts, hashes, workflows, limitations, compatibility, rollback, and consumer + evidence before marking the milestone passed. + +Exit: public `ace-core==0.8.0`, its GitHub release, independent World and Market evidence, and the +reconciled roadmap tell one verifiable story. + +## Golden acceptance journey + +The signature public demonstration is an AI-focused World Intelligence command center because it +uses changing public evidence and is independently reproducible: + +1. Install the public ACE and World packages. +2. Choose the AI Command Center goal and inspect the exact source capabilities requested. +3. Connect at least three materially different authorized sources. +4. Review and edit the cited AI concept model. +5. Review monitors, materiality, audiences, cadence, and source limitations. +6. Receive a cited executive Brief that exposes uncertainty, disagreement, missing evidence, and + why each item matters. +7. Approve the exact activation and receive an immutable receipt. +8. Admit later evidence and show either a material Shift or an explicit no-material-change result. +9. Ask ACE a follow-up and receive an answer grounded in the same durable evidence and citations. +10. Record a Decision or authorized Action, later Outcome, and governed Feedback proposal. +11. Restart the runtime and reproduce identities, revisions, authority, lineage, and current views. + +The Market journey must reproduce the same platform lifecycle without sharing World ontology or +fixtures. A prerecorded or fixture-only UI walkthrough is useful development evidence but cannot +close the release. + +## Non-goals and stop conditions + +0.8 does not promise multi-tenant collaboration, arbitrary hostile-code sandboxing, marketplace, +managed operation, universal source coverage, unrestricted autonomy, omniscience, or general +real-world causal accuracy. Those claims do not enter through UI language or a domain demo. + +Stop and return to review if a change: + +- embeds World, Market, HPE, competitor, model, campaign, or another domain noun in Core or + Intelligence; +- adds a second persistence, authorization, scheduler, secret, or model-routing system; +- lets a UI render, pack, model, or agent grant itself authority; +- requires hand-authored pack files for first value; +- removes a supported contract without a declared compatibility and migration path; +- expands the exact eleven-tool public MCP surface incidentally; or +- folds an unmerged candidate branch into the release without an explicit compatibility gate. + +## Landing and rollback order + +Each packet lands independently in A → B → C → D → E → F order. Domain work may develop in +parallel after A, but Core accepts only domain-neutral requirements. Every packet must preserve a +reversible compatibility seam and name its rollback before the next packet depends on it. The 0.8 +tag is created only after both external consumer gates and public installed-artifact reproduction +pass. diff --git a/tests/test_public_roadmap_positioning.py b/tests/test_public_roadmap_positioning.py index c203e64..e7c6220 100644 --- a/tests/test_public_roadmap_positioning.py +++ b/tests/test_public_roadmap_positioning.py @@ -20,11 +20,24 @@ def test_current_release_and_passed_milestone_are_not_conflated() -> None: assert "| 0.7.0 | Intelligence Builder Foundation | **Passed** |" in ROADMAP readme = (Path(__file__).resolve().parents[1] / "README.md").read_text(encoding="utf-8") assert "0.7.0 is a published developer-preview release" in " ".join(readme.split()) - assert "| 0.8.0 | Intelligence Workspace | **Later** |" in ROADMAP + assert "| 0.8.0 | Intelligence OS Realignment | **Next** |" in ROADMAP assert "| 0.9.0 | Collaborative Intelligence | **Later** |" in ROADMAP assert "| 1.0.0 | Intelligence Operating System | **Later** |" in ROADMAP +def test_080_is_a_bounded_intelligence_os_realignment_not_a_rename() -> None: + assert "### 0.8.0 — Intelligence OS Realignment" in ROADMAP + assert "**0.8A — architecture and compatibility:**" in ROADMAP + assert "**0.8B — runtime realignment:**" in ROADMAP + assert "**0.8C — unified intelligence resources:**" in ROADMAP + assert "**0.8D — Atrium experience:**" in ROADMAP + assert "**0.8E — cross-domain proof:**" in ROADMAP + assert "**0.8F — release acceptance:**" in ROADMAP + assert "World AI Command Center" in ROADMAP + assert "Atrium and a machine interface consume the same public durable intelligence" in ROADMAP_ONE_LINE + assert "does not become a second source of truth" in ROADMAP_ONE_LINE + + def test_060_public_world_journey_closes_the_bounded_measured_intelligence_release() -> None: assert "[0.6.0 GitHub Release]" in ROADMAP assert "public [`ace-core==0.6.0`]" in ROADMAP From 7994541c5e40fd3d9d35c7e45420e9efb99edf8f Mon Sep 17 00:00:00 2001 From: Edwin Amirian Date: Wed, 12 Aug 2026 14:49:19 -0700 Subject: [PATCH 2/3] docs: mark ACE 0.8 as active --- ROADMAP.md | 2 +- tests/test_public_roadmap_positioning.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/ROADMAP.md b/ROADMAP.md index bc2d2fb..84d2523 100644 --- a/ROADMAP.md +++ b/ROADMAP.md @@ -179,7 +179,7 @@ Release targets describe sequence, not calendar commitments. ACE follows three r | 0.5.0 | Reasoning into Action | **Passed** | T1 and B1 pass for bounded attributable action under the explicit single-host, trusted-adapter topology; I1 authority receipts govern the execution-adapter slice | [#37](https://github.com/augmented-cognition-engine/core/issues/37) | | 0.6.0 | Measured Intelligence | **Passed** | Public Core 0.6.0 and World 0.10.0 artifacts reproduce exact useful/harmful/unproven classification, matched leakage-bounded controls, durable replay, and proposal-only governance under product-owned criteria; SI4 remains a broader future outcome | [#38](https://github.com/augmented-cognition-engine/core/issues/38) | | 0.7.0 | Intelligence Builder Foundation | **Passed** | Public Core 0.7.0 composes 0.7A–0.7E, AC1–AC7, and AM0–AM3; all release checks, trusted publication, public clean-install reproduction, and independent World/Market consumer gates passed | [#39](https://github.com/augmented-cognition-engine/core/issues/39) | -| 0.8.0 | Intelligence OS Realignment | **Next** | Makes the evidence → orientation → decision/action → outcome → governed-feedback lifecycle coherent in code, public resources, Atrium, and independent World/Market demonstrations | [#40](https://github.com/augmented-cognition-engine/core/issues/40) | +| 0.8.0 | Intelligence OS Realignment | **Now** | Makes the evidence → orientation → decision/action → outcome → governed-feedback lifecycle coherent in code, public resources, Atrium, and independent World/Market demonstrations | [#40](https://github.com/augmented-cognition-engine/core/issues/40) | | 0.9.0 | Collaborative Intelligence | **Later** | Advances H1, SI3 sensitive-source governance, and the remaining T1/E2 operational guarantees across tenancy, shared authority, privacy, portability, recovery, and managed operation | [#41](https://github.com/augmented-cognition-engine/core/issues/41) | | 1.0.0 | Intelligence Operating System | **Later** | Stabilizes the complete supported loop and all milestone-critical contracts across Core, Intelligence, Domain Packs, connectors, product surfaces, governance, action, outcomes, operation, and portability | [#42](https://github.com/augmented-cognition-engine/core/issues/42) | diff --git a/tests/test_public_roadmap_positioning.py b/tests/test_public_roadmap_positioning.py index e7c6220..e8bbbf8 100644 --- a/tests/test_public_roadmap_positioning.py +++ b/tests/test_public_roadmap_positioning.py @@ -20,7 +20,7 @@ def test_current_release_and_passed_milestone_are_not_conflated() -> None: assert "| 0.7.0 | Intelligence Builder Foundation | **Passed** |" in ROADMAP readme = (Path(__file__).resolve().parents[1] / "README.md").read_text(encoding="utf-8") assert "0.7.0 is a published developer-preview release" in " ".join(readme.split()) - assert "| 0.8.0 | Intelligence OS Realignment | **Next** |" in ROADMAP + assert "| 0.8.0 | Intelligence OS Realignment | **Now** |" in ROADMAP assert "| 0.9.0 | Collaborative Intelligence | **Later** |" in ROADMAP assert "| 1.0.0 | Intelligence Operating System | **Later** |" in ROADMAP From 89fa20e448bd5c36256ebc9010ade2cd59708e50 Mon Sep 17 00:00:00 2001 From: Edwin Amirian Date: Wed, 12 Aug 2026 15:01:54 -0700 Subject: [PATCH 3/3] ci: rerun transient parser acquisition gate