Skip to content

Latest commit

 

History

History
140 lines (116 loc) · 6.28 KB

File metadata and controls

140 lines (116 loc) · 6.28 KB

ROADMAP

Current posture

The bootstrap-to-federation landing path is largely complete. Phases 0 through 6 have already been landed as source and runtime seams:

  • modular runtime bootstrap
  • service extraction and profile-aware lifecycle wrappers
  • Intel-aware and Windows-usable hardening
  • aoa-agents, aoa-routing, aoa-memo, aoa-evals, aoa-playbooks, and aoa-kag advisory/read-export landings
  • Tree-of-Sophia source-owned handoff companion landing
  • bounded promoted local-worker posture through langchain-api on 5403 backed by llama.cpp, with LangGraph as the adopted execution layer for bounded long-horizon and autonomy-focused local-worker flows
  • antifragility wave two as contract-only runtime doctrine and receipt surfaces, now including bounded wave-one chaos example families, without mutating live services or deployment behavior
  • read-only diagnostic spine groundwork through scripts/aoa-diagnose and generated/diagnostic_surface_catalog.min.json, still weaker than repair authority and live-service mutation

Current release contour

The live v0.2.2 contour is runtime-substrate hardening, not a new source of AoA or ToS meaning. Its current checked surface is:

  • source/deployed split and parity posture: README.md, CHARTER.md, BOUNDARIES.md, docs/PATHS.md, docs/DEPLOYMENT.md, scripts/aoa-sync-configs, scripts/validate_stack.py, and scripts/release_check.py
  • diagnostic spine and repair-safe handoff contracts: docs/DIAGNOSTIC_SPINE.md, generated/diagnostic_surface_catalog.min.json, schemas/diagnostic_target.schema.json, schemas/diagnostic_session.schema.json, schemas/diagnosis_companion.schema.json, schemas/reviewed_diagnosis_ref.schema.json, schemas/repair_handoff.schema.json, examples/diagnostic_target.min.example.json, examples/diagnostic_session.min.example.json, examples/diagnosis_companion.min.example.json, examples/reviewed_diagnosis_ref.min.example.json, examples/repair_handoff.min.example.json, scripts/aoa-diagnose, scripts/_aoa_diagnose.py, scripts/build_diagnostic_surface_catalog.py, scripts/validate_diagnostic_surface_catalog.py, tests/test_validate_stack_diagnostic_spine.py, and tests/test_diagnostic_spine_contracts.py
  • antifragility contract-only runtime receipts: docs/ANTIFRAGILITY_RUNTIME.md, docs/RUNTIME_CHAOS_WAVE1.md, docs/REPAIR_SAFE_CLOSEOUT.md, schemas/service_degradation_receipt_v1.json, schemas/repair_safe_closeout_receipt_v1.json, examples/service_degradation_receipt.example.json, examples/service_degradation_receipt.timeout-chaos.example.json, examples/service_degradation_receipt.honest-degradation.example.json, examples/service_degradation_receipt.retrieval-outage-honesty.example.json, examples/repair_safe_closeout_receipt.example.json, examples/repair_safe_closeout_receipt.timeout-chaos.example.json, and examples/repair_safe_closeout_receipt.retrieval-outage-honesty.example.json
  • promoted local-worker and Intel serving posture: docs/RUNTIME_WINNER_PROMOTION_LOOP.md, docs/LLAMACPP_PILOT.md, docs/MACHINE_FIT_POLICY.md, compose/tuning/llamacpp.runtime-fallback.yml, compose/tuning/llamacpp.intel-285h.cpu-safe.yml, compose/tuning/intel-text.ovms-qwen3-settings.yml, docs/model-cards/qwen3-openvino-family.md, and scripts/aoa-llamacpp-pilot
  • bounded federation, runtime-chat, and ToS graph curation seams: docs/MEMO_RUNTIME_SEAM.md, docs/EVAL_RUNTIME_SEAM.md, docs/PLAYBOOK_RUNTIME_SEAM.md, docs/KAG_RUNTIME_SEAM.md, docs/SERVICE_CATALOG.md, docs/PROFILES.md, docs/TOS_GRAPH_CURATION.md, scripts/aoa-federated-check, compose/modules/52-tos-graph.yml, compose/profiles/curation.txt, and config-templates/Services/tos-graph/app/main.py

This contour keeps the source checkout ~/src/abyss-stack separate from the deployed /srv/abyss-stack/Configs mirror. It names runtime contracts, advisory seams, and validation paths without claiming live service mutation or meaning-layer authority.

The main remaining work is live runtime-loop consumption, operational cutover choices, platform hardening, and keeping deployed/runtime truth aligned with source-authored posture.

Those landings should be read carefully:

  • landed seams in source and deployed mirrors are not the same thing as a fully green federated control-plane verdict
  • the federation profile remains opt-in rather than part of the default promoted presets
  • current operator truth still depends on parity, promoted runtime verify, and aoa-status --autonomy --json

Phase 0: structured bootstrap

  • establish repository charter and boundaries
  • create modular compose skeleton
  • define first runtime profiles
  • define env and secrets posture
  • write migration notes from abyss-stack_old

Phase 1: service extraction

  • reintroduce storage services cleanly
  • reintroduce orchestration and local inference
  • reintroduce gateway and agent API modules
  • reintroduce speech, browser, and monitoring modules

Phase 2: operational hardening

  • add smoke and health routines
  • add profile-aware lifecycle wrappers
  • add backup and restore helpers
  • add validation for compose coherence
  • reduce environment-specific assumptions where possible
  • enforce the new /srv/abyss-stack canonical runtime root
  • make the Fedora-first and Windows-usable path model explicit
  • make deployment from source checkout to runtime tree explicit and repeatable

Phase 3: hybrid growth

  • clarify local versus hybrid execution paths
  • refine Intel and OVMS posture
  • define clean bridges to sibling AoA repositories

Phase 4: mature substrate

  • keep the stack legible under growth
  • resist monolith relapse
  • let new capability arrive as modules and profiles rather than as hidden sprawl

Phase 5: live runtime consumption

  • decide which federation seams remain advisory-only and which become part of the live loop
  • introduce bounded recall, playbook, eval, and KAG consumption in explicit steps
  • preserve source-owned authority while adding runtime utility

Phase 6: platform and operations hardening

  • keep runtime cleanup repeatable and legible
  • validate reboot and cold-start behavior
  • tighten runtime-secret hygiene and stateful-data discipline
  • keep Windows and Fedora rollout paths aligned with the same operational model