fix(lcm): harden Hermes compaction replay bridge - #36
Conversation
Ensure threshold-triggered compaction can summarize short oversized contexts and keep oversized LCM compress responses machine-readable for the Hermes bridge.
Add Hermes optional hook shims, richer generated-plugin diagnostics, and contract-safe LCM preflight/retrieve handling so context-engine integration is easier to verify and diagnose.
Preserve project-scoped response handle retrieval and keep compact LCM replay payloads bounded for structured content.
Use map_or_else when serializing structured replay content for bounded MCP responses.
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 59e6e46e82
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Preserve exact Hermes LCM summary source contracts by routing oversized needs-summary payloads through the retrievable truncation path.
1 participant
Summary
Test plan
cargo fmt --checkandcargo test --workspace --features test-transport.