formal/ — canonical TLA+ spec (G12 non-bypass)

formal/ — canonical TLA+ spec (G12 non-bypass)

antahkarana 5.0.0 — Self-governing (Gate G12): machine-checked non-bypass + self-improvement adopted

antahkarana 5.0.0 — Self-governing (Gate G12): machine-checked non-bypass + self-improvement adopted

antahkarana 4.0.1 — atk CLI: plan + audit (G11.6/G11.4); companion antahkarana-cli 0.5.0

antahkarana 4.0.0 — Platform (Gate G11): multi-tenant, provably isolated + audit export

antahkarana 4.0.0 — Platform (Gate G11): multi-tenant, provably isolated + audit export

antahkarana 3.5.0 — Multi-agent mesh (Gate G10): one-tick control, ring inheritance, fleet drift

antahkarana 3.5.0 — Multi-agent mesh (Gate G10): one-tick control, ring inheritance, fleet drift

antahkarana 3.0.0 — Memory & feedback (Gate G9): the gate learns; strict dominance on 3 held-out replays

antahkarana 3.0.0 — Memory & feedback (Gate G9): the gate learns; strict dominance on 3 held-out replays

antahkarana 2.5.0 — Integration (Gate G8): BuddhiLLM proposes/ring disposes; 500-case real-7B red-team, 0 escalations

antahkarana 2.5.0 — Integration (Gate G8): BuddhiLLM proposes/ring disposes; 500-case real-7B red-team, 0 escalations

antahkarana 2.2.0 — Hardening (Gate G7): telemetry+replay, budgets/breakers, policyspec, baseline registry, atk CLI

antahkarana 2.2.0 — Hardening (Gate G7): telemetry+replay, budgets/breakers, policyspec, baseline registry, atk CLI

2.1.0 — release banner

2.1.0 — CGB performance figures

2.1.0 — CGB leaderboard results

2.1.0 — CGB model card (achievements + performance plots)

2.0: clean minimal loop-engine banner (1.1-style)

2.0: clean minimal loop-engine banner (1.1-style)

2.0: the loop engine (watch/learn/improve/orchestrate) model card

2.0: add loop-engine diagram

1.1.0 model card — four no-forgetting levers, mātrā-LAMB stability, adaptive guṇa (story + perf graphs + banner)

Upload assets/control_ring_performance.png with huggingface_hub

Upload assets/control_ring_architecture.png with huggingface_hub

Cross-link PyPI: pip install antahkarana

Cross-link PyPI: pip install antahkarana

Add architecture+performance plots, gates section, any-domain framing, author=Deepak Soni, results

Antahkarana-base: continual-learning core (card + code + results + gates G0-G5)

initial commit