Repository navigation
Separate app safeguards and consolidate safe global engineering guidance - #22
Merged
Merged
Conversation
dreyna001
changed the title
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Global engineering guidance and application-only safeguards were mixed across packages and older project copies. This organizes their canonical source into
global/,ai-enabled-ato/,ai-enabled-threathunting/, andllm_notable_analysis/, with installers updated for the global layout.The shared contracts now cover report/export parity, evidence-field mapping, independent frontend deployment, and safe efficiency work. A conditional
safe-reuse-and-preparationreference owns bounded cache lifecycle, current authorization/freshness, refresh coalescing, compatible extraction and per-upload checks. Existing model, API, reliability and agentic guidance adds deterministic prechecks, bounded batching, complete outcome measurements, late cancellation/owned-upload cleanup, retry identity and explicit approvals in existing chat-led workflows. Existing benchmark, evidence-regression and real-boundary testing policies are reused. Logical-operation keys remain stable across retries; attempt trace IDs remain distinct.Each app package retains its runtime, field, reporting, approval/revision and platform boundaries with source manifests. Historical app HTTP checkpoints no longer live globally. No app code, saved data or UI baseline was changed.
WSL received only task-owned deltas, preserving unrelated local approved UI policy edits and the active source checkout. Duplicate app guidance was archived outside discovery with hashes. Windows configuration was preserved. The existing shared security skill is demonstrated in WSL; no new skill was needed for the efficiency additions.
Validation: the structural change passed 28 configuration tests, isolated installer preservation checks, a fresh 60-skill installation and native WSL discovery/isolation checks. The efficiency additions passed the scoped manifest/link/mirror checker, four modified-skill validations, independent review and active WSL installation verification. No model/provider calls or app benchmarks were run; broader app testing remains deferred.
Details:
docs/scoped-guidance-2026-11-04.mdrecords installation boundaries;docs/efficiency-guidance-coverage-2026-11-04.mdmaps all 17 requested principles to their authoritative guidance.