Project Infinity 1 — AAIS

Adaptive Autonomous Intelligence Substrate (AAIS) — a law-governed Jarvis runtime with operator dashboard, brain sessions, decision ledger, plugins/workflows, and 199 governed subsystem schemas.

License: Apache 2.0 · Repo: Project-Infinity1 · Audit report: Zenodo

---

Quick start (after clone)

Windows (PowerShell):

git clone https://github.com/warheart1984-ctrl/Project-Infinity1.git
cd Project-Infinity1
.\scripts\start-infinity1.ps1

If port 8000 is busy: .\scripts\start-infinity1.ps1 -ReplaceExisting

Mac / Linux:

git clone https://github.com/warheart1984-ctrl/Project-Infinity1.git
cd Project-Infinity1
chmod +x scripts/start-infinity1.sh
./scripts/start-infinity1.sh

The script creates a .venv, runs pip install -e ".[dev]", prepares runtime data, and starts AAIS in mock mode (no API keys required).

Manual start (venv already installed): .venv\Scripts\python.exe -m aais start --data-dir ./.runtime/aais-data --preset mock --no-browser (Windows) or .venv/bin/python -m aais start ... (Unix).

---

How to join in 10 minutes

Goal: run AAIS locally in mock mode (no API keys), confirm health, then pick one Help Wanted ticket.

Fast arc smoke (optional, ~1–2 min): make civilizational-arc-smoke (or python -m pytest tests/test_inter_substrate_diplomacy_observe.py tests/test_norm_federation_observe.py tests/test_constitutional_evolution_observe.py -q)

Next: docs/community/HELP_WANTED_HUB.md · docs/operations/OPERATOR_GOLDEN_PATH.md · docs/operations/FIRST_TIME_OPERATOR_GUIDE.md

---

What's in Infinity 1 (this repo)

Verify locally (optional): python tools/governance/run_infinity1_flagship_verification.py · make flagship-chaos-stack

Deeper docs: Operator Golden Path · Operator Guide · First-Time Operator Guide · Workflow Skills · Infinity Pilot (Docker) · GitHub Packages / GHCR · Governance Dashboard Guide

Operator surfaces (reference)

Production gates: make infinity1-flagship-verification (Windows: python tools/governance/run_infinity1_flagship_verification.py)

Baseline: INFINITY_PILOT_BASELINE_CHECKLIST · Early adopters: Charter

What AAIS Is

AAIS is not a single chatbot wrapper. It is a governed cognition runtime that:

• routes operator turns through Jarvis (src/api.py, src/jarvis_operator.py)
• adapts every outward payload through AAIS-UL (structure + visibility before expansion)
• enforces Project Infi law on chat replies, forge contractors, and repo mutations
• stages work on the CISIV ladder: concept → identity → structure → implementation → verification
• exposes traces, law enforcement, and UL substrate envelopes on live API responses

Think of it in three cooperating layers:

Authoritative references:

• Strategy (vision → execution): docs/spine/STRATEGY.md
• Subsystem map: docs/runtime/AAIS_SUBSYSTEM_SPEC.md · Remaining work: docs/runtime/SUBSYSTEMS_REMAINING_MAP.md
• UL doctrine: docs/contracts/AAIS_UL_DOCTRINE.md
• Latest UL/CISIV proof: docs/proof/aais-ul/UL_CISIV_PHASES_1_5_PROOF.md

This repository is also Project Infi — constitutional engineering where claims require proof, not intent.

Terminology (Project Infinity)

Architecture is unchanged; documentation and gates use standard engineering language. Legacy repo identifiers (*_organ, *.genome.v1.json, altN-summon-wave) remain for compatibility.

Linguistic naming schema — each governed subsystem schema carries mythic and engineering names on its SSP block (170 governed genomes at v1.25.0):

Snapshots: governance/linguistic_snapshots/ · schema linguistic_snapshot.v1.json · grandfathered aliases: legacy_engineering_aliases.v1.json.

make naming-gate naming-genome-gate meta-linguistic-gate   # Codex lint + genome alignment + meta orchestration
make translate-mythic MYTHIC='V9 runtime steward'   # mythic → engineering_class helper

Codex / Cursor naming (new code): AAIS_CODEX_CURSOR_NAMING_PROTOCOL.md · Meta-linguistic governance: AAIS_META_LINGUISTIC_GOVERNANCE.md — mythic in comments, engineering in identifiers; legacy *_organ / *_fabric paths grandfathered until Wave 4 MP-X rename.

License: Apache 2.0 · Latest release: v1.26.1 — Release 30.1 OTEM Level 10 Safe Activation (tag on GitHub) · URG Cloud Platform: docs/URG_CLOUD_PLATFORM.md · tag urg-cloud-platform-v1.9 (federation; prior v1.6) · Release history: CHANGELOG.md · Onboarding: First-Time Operator Guide

---

Architecture

AAIS separates cognition authority, workflow shell, and optional ops planes. Jarvis runtime truth lives in Flask (src/api.py); the FastAPI shell (app/main.py) hosts the operator UI and bridges to Jarvis at /legacy_api.

flowchart LR
    subgraph cognition [Cognition executive]
        Flask["src/api.py Flask Jarvis"]
        Law[project_infi_law]
        UL[aais_ul]
    end
    subgraph shell [Workflow shell]
        FastAPI["app/main.py :8000"]
        UI[React app/static]
    end
    subgraph ops [Ops membrane optional]
        Platform["platform serve :8090"]
        UGR[UGR ledger]
    end
    subgraph contractors [Contractors optional]
        Forge[forge :6060]
        Evolve[evolve :6062]
    end
    UI --> FastAPI
    FastAPI -->|"/legacy_api bridge"| Flask
    Flask --> Law
    Flask --> UL
    Platform --> UGR
    Flask -.-> Forge

Authority boundaries

Request path (operator chat)

1. Operator turn enters /legacy_api/api/chat/... (Flask Jarvis).
2. Chat turn governance and AAIS-UL substrate adapt the outward payload.
3. Project Infi law admits or filters the reply.
4. Provider registry routes to local, mock/laptop, or any configured frontier adapter (OpenAI, Claude, Gemini, Nemotron, OpenRouter, and others — see Installing API keys).
5. Response includes ul_substrate, modular_preview, law_enforcement, and cisiv_stage.

Optional subsystems (Platform, Wolf-CoG-OS ISO forge, forge/evolve contractors) attach at the edges — core chat works without them.

Deep dives: docs/runtime/AAIS_SUBSYSTEM_SPEC.md, docs/operations/FULL_STACK_PILOT_INTEGRATION.md.

Using OTEM Level 10 (v1.26.1)

OTEM is activated at capability level 10 by default. The chat lane stays proposal-only; governed execution runs only after operator approval on the workflow shell.

Operator flow

1. In Jarvis (/app/jarvis), run OTEM on a task that matches a workflow template (e.g. daily brief). The reply includes a workflow handoff proposal — no tools or patches run in chat.
2. Open Workflow Approvals: http://127.0.0.1:8000/workflows/approvals (or use the Dashboard OTEM handoff link when execution_approval_queue is pending).
3. Approve or reject in the same API process that handled the OTEM turn (approve runs substrate approve() + apply()).

Configure (optional — default is already 10):

# In .env (see .env.example)
AAIS_OTEM_CAPABILITY_LEVEL=10

Check posture:

curl -s http://127.0.0.1:8000/legacy_api/api/jarvis/otem-bounded/status | python -m json.tool

Expect otem_runtime_version: v10_governed, otem_capability_level: 10, execution_via_workflow_approvals: true.

After server restart: Pending approvals may reference an in-memory substrate workflow that no longer exists. Reject the stale row and re-run the OTEM handoff, or approve before restarting. Substrate persistence phase 2 is deferred until cross-restart durability is required.

Contract: OTEM_EXECUTION_SUBSTRATE.md · Onboarding: FIRST_TIME_OPERATOR_GUIDE.md § OTEM Level 10.

Operator Workflow Skills (Infinity 1)

Governed admission of Cursor skills, Hugging Face agent skills, and workflow bundle chains into six workflow-family organs — with Brain-layer proposals, fitness scoring, sessions, and deliberation on top.

Skill libraries: 5 Cursor (lib_skill_*), 10 HF (lib_hf_*), 27 workflow bundles (lib_workflow_*). Full catalog: OPERATOR_WORKFLOW_SKILLS.md.

Contracts: BRAIN_PROPOSAL_CONTRACT · BRAIN_SESSION_CONTRACT · BRAIN_DELIBERATION_CONTRACT

Verification:

make operator-workflow-stack-gate
make infinity1-flagship-verification

Architecture: AAIS_ANATOMICAL_LAYERS.md · Capability map: AAIS_AGENT_WORKFLOW_CAPABILITY_MAP.md

Alt-3 partial-live subsystems (v0.3.0)

Three archive families from Audit Alt-3 are now partial live — governed MVPs with API routes, capability bridge actions, UL lineage, and proof packets:

Grok imagine render requires STORY_FORGE_XAI_API_KEY or XAI_API_KEY in the environment (no request-body keys). Active docs: RECIPE_MODULE, IMAGINE_GENERATOR, HUMAN_VOICE_EXTRACTION.

Verification:

make alt3-gate
# or: python .github/scripts/check-recipe-module-governance.py
#     python .github/scripts/check-imagine-generator-governance.py
#     python .github/scripts/check-human-voice-extraction-governance.py

Infinity 1 — Governance runtime and constitutional layer (v1.3.0)

Self-governing AAIS with executable Alt-4 lifecycle subsystems, one hundred thirty-eight governed subsystem schemas, Tier 5 adaptive governance, Alt-5 releases 1–2, Alt-6 adaptive lane Coherence Layer, Alt-7 operator–cognition coherence layer, Alt-8 mind-plane subsystems, Alt-9 infrastructure layer, Alt-10 memory/forensics/immune observe layer, Alt-11 authority trace/boundary/coding layer, Alt-12 OTEM/predictive/execution-depth layer, Alt-13 creative chain/constitutional closure layer, Alt-14 route choice/perception layer, Alt-15 Nova cortex lobe and voice layer, Alt-16 factory and kinetic layer, Alt-17 authority shell and protocol layer, Alt-18 Project Infi law layer, Alt-19 operator product interfaces, Release 20 operator workspace and extended interfaces, Release 21 creative runtime V9/V10 fabric, and Release 22 meta-linguistic governance fabric.

Promotion scripts: tools/governance/alt5_promote_wave2_mvp.py, alt5_promote_governed.py, barebones_promote_governed.py, alt6_promote_governed.py, alt7_promote_governed.py, alt8_promote_mvp.py, alt8_promote_governed.py, alt9_promote_mvp.py, alt9_promote_governed.py, alt10_promote_mvp.py, alt10_promote_governed.py, alt11_promote_mvp.py, alt11_promote_governed.py, alt12_promote_mvp.py, alt12_promote_governed.py, alt13_promote_mvp.py, alt13_promote_governed.py, alt14_promote_mvp.py, alt14_promote_governed.py, alt15_promote_mvp.py, alt15_promote_governed.py, alt16_promote_mvp.py, alt16_promote_governed.py, alt17_promote_mvp.py, alt17_promote_governed.py, alt18_promote_mvp.py, alt18_promote_governed.py, alt19_promote_mvp.py, alt19_promote_governed.py, alt20_promote_mvp.py, alt20_promote_governed.py, alt21_promote_mvp.py, alt21_promote_governed.py, alt22_promote_mvp.py, alt22_promote_governed.py, alt25_promote_mvp.py, alt25_promote_governed.py, alt26_promote_mvp.py, alt26_promote_governed.py, alt27_promote_mvp.py, alt27_promote_governed.py, alt28_promote_mvp.py, alt28_promote_governed.py.

Verification:

make genome-gate alt4-gate alt5-gate barebones-gate tier5-gate alt6-governed-gate alt7-governed-gate alt7-1-gate alt7-2-gate alt8-gate alt8-1-gate alt8-2-gate alt8-governed-gate alt9-gate alt9-1-gate alt9-2-gate alt9-governed-gate alt10-gate alt10-1-gate alt10-2-gate alt10-governed-gate alt11-gate alt11-1-gate alt11-2-gate alt11-governed-gate alt12-gate alt12-1-gate alt12-2-gate alt12-governed-gate alt13-gate alt13-1-gate alt13-2-gate alt13-governed-gate alt14-gate alt14-1-gate alt14-2-gate alt14-governed-gate alt15-gate alt15-1-gate alt15-2-gate alt15-governed-gate alt16-gate alt16-1-gate alt16-2-gate alt16-governed-gate alt17-gate alt17-1-gate alt17-2-gate alt17-governed-gate alt18-gate alt18-1-gate alt18-2-gate alt18-governed-gate alt19-gate alt19-1-gate alt19-2-gate alt19-governed-gate alt20-gate alt20-1-gate alt20-2-gate alt20-governed-gate alt21-gate alt21-1-gate alt21-2-gate alt21-governed-gate alt22-gate alt22-1-gate alt22-2-gate alt22-governed-gate
python -m pytest tests/test_governance_organs_alt4.py tests/test_adaptive_governance.py \
  tests/test_adaptive_lane_organ.py tests/test_alt6_governed_eligibility.py \
  tests/test_adaptive_lane_bridge.py tests/test_coherence_fabric_bridge.py \
  tests/test_alt7_governed_eligibility.py tests/test_operator_cognition_coherence_fabric.py \
  tests/test_continuity_witness_organ.py tests/test_narrative_continuity_organ.py tests/test_intent_agency_organ.py \
  tests/test_safety_envelope_organ.py tests/test_operator_profile_organ.py \
  tests/test_reflection_runtime_organ.py tests/test_memory_runtime_organ.py -q

Operator guide: AAIS_ALT4_RUNTIME_OPERATOR_GUIDE · Adaptive law: AAIS_ADAPTIVE_GOVERNANCE · Adaptive lanes: ADAPTIVE_LANE_ORGAN · Coherence Layer: OPERATOR_COGNITION_COHERENCE_FABRIC · Schema registry: governance/subsystem_genomes/README.md · Naming contract: AAIS_CODEX_CURSOR_NAMING_PROTOCOL.md

v1.0.0 shipped the initial Infinity 1 slice (Alt-4, Tier 5, Alt-5 wave 1 at MVP). v1.1.0 completes the constitutional layer (barebones wave + Alt-5 wave 2). v1.2.0 adds Alt-6 adaptive lanes at governed. v1.3.0 adds Alt-7 coherence fabric with cross-plane bridge enforcement. v1.4.0 adds Alt-8 mind-plane organs and coherence fabric v1.3. v1.5.0 adds Alt-9 infrastructure organs and coherence fabric v1.4. v1.6.0 adds Alt-10 memory/forensics/immune observe organs and coherence fabric v1.5. v1.7.0 adds Alt-11 authority trace/boundary/coding organs and coherence fabric v1.6. v1.8.0 adds Alt-12 OTEM/predictive/execution-depth organs and coherence fabric v1.7. v1.9.0 adds Alt-13 creative chain/constitutional closure organs and coherence fabric v1.8. v1.10.0 adds Alt-14 route choice/perception organs and coherence fabric v1.9. v1.11.0 adds Alt-15 Nova cortex lobe and voice organs and coherence fabric v1.10. v1.12.0 adds Alt-16 factory and kinetic organs and coherence fabric v1.11. v1.13.0 adds Alt-17 authority shell and protocol organs and coherence fabric v1.12. v1.14.0 adds Alt-18 Project Infi law organs and coherence fabric v1.13. v1.15.0 adds Alt-19 operator product shell organs and coherence fabric v1.14. v1.16.0 adds Release 20 operator workspace and extended interface subsystems and Coherence Layer v1.15. v1.17.0 adds Release 21 creative runtime V9/V10 subsystems and Coherence Layer v1.16. v1.18.0 adds Release 22 meta-linguistic governance subsystems, meta-linguistic gates (Waves 0–10), and Coherence Layer v1.17. v1.19.0 adds Release 23 predictive linguistic cycle subsystems and Coherence Layer v1.18. v1.20.0 adds Release 24 attested linguistic closed-loop subsystems, Wave 14 attestation engines, and Coherence Layer v1.19. v1.21.0 adds Release 25 governed linguistic lifecycle fabric (nine organs) and Coherence Layer v1.20. v1.22.0 adds Release 26 operational closure (three organs), Coherence Layer v1.21, Codex engineering headers, and cumulative stack gates with naming-gate at zero warnings.

Three Ideas MVP partial-live subsystems (v0.4.0)

Three repo-grounded ideas promoted from concept to partial live — runtime modules, governance gates, and proof packets:

Active docs: UL_LINEAGE_CONSOLE, TRIANGULATION, NARRATIVE_TRUST_PACK.

Verification:

make lineage-gate triangulation-gate narrative-gate
python -m pytest tests/test_ul_lineage.py tests/test_triangulation.py tests/test_narrative_trust_pack.py -q
python -m tools.ul.smoke --lineage-graph tools/ul/fixtures/lineage_multi_hop.json --no-pytest

---

How to start operations

Prerequisites

• Python 3.10+ and Git
• Optional: Node.js 18+ — only to rebuild the frontend (frontend/)
• Optional: API keys in .env for frontier models (mock preset needs none)

Install and start (automated)

Use the bootstrap script (recommended):

# Windows
.\scripts\start-infinity1.ps1

# Mac / Linux
./scripts/start-infinity1.sh

Manual install (alternative)

git clone https://github.com/warheart1984-ctrl/Project-Infinity1.git
cd Project-Infinity1
python -m venv .venv
# Windows: .venv\Scripts\activate
# Unix:    source .venv/bin/activate
python -m pip install -e ".[dev]"
cp .env.example .env
python -m aais prepare --data-dir ./.runtime/aais-data
python -m aais doctor --data-dir ./.runtime/aais-data
python -m aais start --data-dir ./.runtime/aais-data --preset mock --no-browser

Installing API keys (frontier models)

Release 28 registers every major frontier adapter in the Jarvis provider picker. Adapters stay off until you add the matching key to .env and restart AAIS.

Steps

1. Copy the template: cp .env.example .env
2. Uncomment or set only the keys for providers you use (never commit .env).
3. Restart AAIS: python -m aais start --data-dir ./.runtime/aais-data --preset default --no-browser
4. Confirm in the UI or API:
   • GET http://127.0.0.1:8000/legacy_api/api/jarvis/providers — each entry shows available: true when configured
   • Set chat/session preferred_provider to the provider id (e.g. openai, claude, nvidia, google)

Common keys (full list in .env.example and docs/providers/FRONTIER_MODEL_ADAPTERS.md):

NVIDIA Nemotron (new) — Nemotron 3 Nano is the current open agentic frontier line; optional chain-of-thought: AAIS_NVIDIA_ENABLE_THINKING=1. Self-hosted NIM: set AAIS_NVIDIA_BASE_URL to your http://host:8000/v1/chat/completions.

Mock dev (no keys) — --preset mock uses MockMultiModalAI; frontier providers still appear in the list as unavailable until keyed.

Optional chat speed (after keys or for local mock):

AAIS_COHERENCE_FABRIC_CACHE_SEC=45
AAIS_GOVERNED_PIPELINE_CACHE_SEC=45
AAIS_SLINGSHOT_CACHE_SEC=30

Set any cache to 0 to disable.

Prepare runtime data (if not using bootstrap script)

python -m aais prepare --data-dir ./.runtime/aais-data
python -m aais doctor --data-dir ./.runtime/aais-data

prepare stages the packaged UI into app/static/. A prebuilt bundle ships with the repo; use --force-build only after npm install in frontend/.

Operational Entry Point

Start AAIS

Recommended: .\scripts\start-infinity1.ps1 (Windows) or ./scripts/start-infinity1.sh (Unix)

Manual:

python -m aais start --data-dir ./.runtime/aais-data --preset mock --no-browser

Presets (src/main.py):

Developer alternative (uvicorn directly):

make run
# equivalent: uvicorn app.main:app --reload

Open operator surfaces

Verification Step

curl -fsS http://127.0.0.1:8000/health

Create a chat session and send a message:

curl -fsS -X POST http://127.0.0.1:8000/legacy_api/api/chat/sessions \
  -H "Content-Type: application/json" \
  -d "{\"system_prompt\":\"You are Jarvis.\"}"

# Use session_id from response:
curl -fsS -X POST http://127.0.0.1:8000/legacy_api/api/chat/sessions/<session_id>/message \
  -H "Content-Type: application/json" \
  -d "{\"message\":\"Summarize AAIS.\",\"response_mode\":\"operator\"}"

A healthy turn returns ul_substrate, modular_preview, law_enforcement, and cisiv_stage on the payload.

UL governance smoke:

python -m tools.ul.drift
python -m tools.ul.smoke
python -m pytest tests/test_cisiv.py tests/test_chat_turn_governance.py tests/test_forge_repo_governance.py -q

Alt-3 subsystem gate (v0.3.0+):

make alt3-gate

Three Ideas MVP gates (v0.4.0+):

make lineage-gate triangulation-gate narrative-gate

6. Optional contractor lanes

These are isolated HTTP services — start only when you need forge/evolve features:

Without them, core chat and patch-review paths still work; explicit forge routes return routing errors until the contractor is up.

Failsafe Notes

• Stop foreground runtime with Ctrl+C.
• Do not delete .runtime/aais-data during active sessions.
• Missing proof or constitutional ambiguity is a stop condition — see governance section below.

---

GitHub

Pull requests to main run governance gates (CoGOS CI, documentation baseline, UGR trust bundle, operator console, Forgekeeper, Scorpion). Significant claims in PRs must include proof posture (asserted / proven / rejected) per REPO_PROOF_LAW.md.

Do not commit: ISO images, .runtime/, wolf-cog-os/output/, or duplicate import folders (*-main/).

---

Repository Layout (operator view)

aais/              Cross-platform launcher (start | prepare | doctor)
app/               FastAPI workflow shell + packaged static UI
src/               Jarvis runtime authority (api, operator, UL, law)
frontend/          React operator UI source (build → app/static)
forge/             Isolated Forge contractor service
platform/          Multi-tenant Platform Membrane (ops ingress :8090)
wolf-cog-os/       CoGOS ISO/rootfs forge (scripts tracked; outputs local-only)
deploy/            Docker compose stacks (pilot | platform | ugr)
tools/ul/          UL drift + smoke verification (+ lineage graph smoke)
tools/narrative/   Narrative Trust Pack CLI (pack | verify | signoff)
triangulation/     Forensic Triangulation correlator (Mechanic + Scorpion + Slingshot)
tools/recipe/      Recipe Module CLI + fixtures
tools/imagine/     Imagine Generator fixtures
tools/human_voice/ Human Voice Extraction fixtures
tools/governance/  SSP completeness and genome gates
docs/              Contracts, subsystem spec, proof packets
tests/             Pytest suite
external/          Vendored third-party integrations (see each package)

---

Constitutional Governance

Behavior is constitutional, not aspirational. No fix, test, or release claim is complete without evidence.

Precedence: Law > Blueprint > Contract > Implementation > Pipeline > Tool

Governance references:

• META_ARCHITECT_LAWBOOK.md
• REPO_PROOF_LAW.md
• HUMAN_AI_CO_COLLABORATION_CHARTER.md
• docs/TRUST_BUNDLE_SPEC.md

Doctrine summary (twelve doctrines)

Templates: templates/PROOF_BUNDLE_TEMPLATE.md, templates/PROJECT_BASELINE_CHECKLIST.md

---

Contributor Oath

1. I will not present unproven claims as complete.
2. I will attach traceable evidence to significant fix/test/release claims.
3. I will preserve constitutional precedence and no-bypass governance.
4. I will track documentation/governance debt instead of hiding it.
5. I will treat missing evidence as a stop condition, not a paperwork delay.

---

Contributors

See CONTRIBUTORS.md.

• Jon Halstead — maintainer and constitutional authority
• Cursor Agent (Auto) — AI implementation collaborator (UL/CISIV Phases 1–5, governance modules, proof bundles, operational README; commits 7b4e806, b086b1e)

Human–AI collaboration follows HUMAN_AI_CO_COLLABORATION_CHARTER.md.