Closed
Conversation
Align Orleans RequestContext and EventEnvelope metadata propagation through a unified AgentContext accessor, and add runtime/tests/docs to verify context isolation and restoration across grain and event paths. Made-with: Cursor
- Use dedicated RequestContext prefix (aevatarac_) to avoid Orleans internal __ prefix filtering - Change OutgoingFilter to upsert-only to prevent clearing propagated context - Extract UpsertFromContext for reuse; Remove redundant Entries iteration - Add OrleansAgentContextCrossSiloIntegrationTests Made-with: Cursor
…ntext-adapter Made-with: Cursor
eanzhao
added a commit
that referenced
this pull request
May 8, 2026
Address review batch on PR #562 (10 inline comments). All in files I have recent ownership of and require no architectural shifts: - #16 (blocker, security): ssh_exec is now opt-in via NyxIdToolOptions. EnableSshExecTool. Hosts that haven't wired the approval middleware no longer see the tool by default. Mainnet host opts in (Lark bot needs it). - #21 (major, bug): code_execute keeps the modern /execute + {language, script} contract, but on a NyxID-proxy upstream 404 it retries the legacy /run + {language, code} contract so deployments still pinned to old chrono-sandbox-service builds keep working. - #22 (major, bug): SkillRegistry.IsFresh now exempts SkillSource != Remote from TTL — local skills are baked in at registration and don't need expiring; prior behavior dropped them from use_skill after the first 5min. - #18 (major, bug): TurnRunner.TryResolveSenderBindingAsync narrows the catch to transient infra errors (Http/Timeout/IO/JSON) and surfaces non-transient (logic, NRE, serialization) at Error level so ops can distinguish "sender unbound" from "binding store broken". - #19 (major, bug): ConversationReplyGenerator narrows the sender-route-fallback catch to transient errors via IsRetryableSenderRouteFailure. Programmer errors no longer cost an LLM round on retry. - #29 + #30 (minor): inbox runtime gives metadata enrichment its own 15s budget separate from the LLM run, surfacing errorCode=llm_reply_metadata_timeout when scope/UserConfig lookup is slow. ResolveFallbackTimeout treats ResponseTimeoutSeconds<=0 as "no timeout" rather than silently snapping back to 120s. - #12 (minor): ConversationGAgent's stream-chunk and final-stream-chunk edits run under a 10s CTS now; the failure path already uses one. A hung relay can no longer pin the actor turn forever. - #27 (minor, security): ConstantTimeEquals docstring tightened — removed the "for future callers" line and added a SCOPE comment that this helper is rebuild-admin-only and shouldn't be promoted to internal/public without replacing its length-leak with a length-padding scheme. - #23 (major, bug): CLI ornn skills slug default → ornn-api (matches the registered slug; bare "ornn" is the SPA frontend that returns HTML). Build clean (NyxId / Skills / NyxidChat / Mainnet hosts), 30 AI tests + 15 inbox runtime tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
eanzhao
added a commit
that referenced
this pull request
May 8, 2026
- #7 (major, arch): wrap MEAILLMProvider's Patch.Set("$.reasoning_content") in try/catch so a future OpenAI SDK contract break (Patch surface change, serializer rewrite, field rename) degrades to "no reasoning replay" instead of crashing the chat call. The OpenAI package is already pinned to 2.9.1 in Directory.Packages.props, and the existing AIComponentCoverageTests already pin the serialized JSON shape so any drift fails the build the moment Patch stops landing in the payload. - Test fix for #22 (skill registry TTL): the old TryGet_BeyondTtl_ReturnsFalseSoCallerCanRefetch test relied on TTL expiring SkillSource.Local entries — that's the bug #22 flagged. Updated the stale-entry tests to use SkillSource.Remote (via remoteId) which is the realistic stale scenario. Added a new TryGet_LocalSkillBeyondTtl_StillFresh test pinning the new behaviour. 538 AI tests + 897 ChannelRuntime tests + 16 Ornn tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
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
1 participant
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.
No description provided.