Repository navigation
Add model-agnostic orchestration, local decision SDKs and bounded source memory - #37
Merged
Merged
Conversation
Tool-Used: Codex
Preserve human and BACKS authorship; development assistants receive tool credit only. Tool-Used: Codex Tool-Used: local Qwen review and runtime smoke
…cision SDKs and source memory
…fy both deadline orderings
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Hydra now keeps normal model execution, optional local decisions, memory and independent review separate. Changing a provider does not replace the skills or source map, and an unavailable independent tribunal does not block ordinary work.
Validation: full repository suite 793 passed / 7 skipped on Windows; final standard topology-fixture change 20 passed. Actual local SDK tool discovery passed without authentication or app actions. Actual offline Needle proposed the correct source lookup and reported three-CPU affinity on a 24-logical-CPU host. Real child-process embedding HTTP tests, proxy bypass, stale-key rejection, account reset persistence, bounded worker cleanup and tribunal alias/fallback regressions passed. Independent code review reproduced and closed identified runtime defects.
The upload tree was verified identical to the tested local Git tree. The subsequent public-profile and test-only cleanup passed 35 focused tests. The final source freshness and configured-auditor fixes passed 35 tests with 1 optional skip; a timestamp-preserving restore now invalidates stale index metadata after a rejected read. The concurrency integrity test now uses an explicit ten-second test budget while production retains its one-second lock limit and a separate fast-refusal regression; 29 health/queue checks passed, with independent review. Exact-head CI status is visible below; earlier-head checks are not reused. This remains a development preview: 4080/iGPU inference validation (the actual AMD 680M/4080 host currently reports an NVIDIA driver/library mismatch), OCuLink verification, connected-account workflows, complete private deployment parity and Cloudflare transaction semantics are not claimed. Public canon is generic; private material has not been copied into this PR.
Project credit remains Cuzzo / BACKS AIOS. Model services and coding assistants receive tool credit only; upstream licenses remain intact.