Skip to content

Improve agent parsing, split diff UX, and chunk-level review flow#16

Open
yjr24300180030 wants to merge 1 commit intoOpenDCAI:mainfrom
yjr24300180030:codex/robust-agent-diff-fixes
Open

Improve agent parsing, split diff UX, and chunk-level review flow#16
yjr24300180030 wants to merge 1 commit intoOpenDCAI:mainfrom
yjr24300180030:codex/robust-agent-diff-fixes

Conversation

@yjr24300180030
Copy link
Copy Markdown

Summary

  • Hardened agent reply parsing for wrapped/fenced/json-like responses.
  • Improved split diff readability with hunk grouping and side-by-side alignment.
  • Added per-change accept/reject flow and diff-to-editor cursor linkage.
  • Added guard against partial suggestion replacing the whole document.
  • Restored undo/redo controls and disabled repeated send while agent is busy.

Validation

  • npm --workspace apps/frontend run build
  • node --check apps/backend/src/routes/agent.js
  • node --check apps/backend/src/services/agentService.js
  • node --check apps/backend/src/services/compileService.js

Dennis-Huangm added a commit to Dennis-Huangm/OpenPrism that referenced this pull request Mar 10, 2026
Adopt the low-risk backend core from PR OpenDCAI#16 by normalizing varied LLM reply payloads in the agent route and adding stricter read/list/iteration limits in the tool agent to reduce brittle parsing and runaway context growth.
Dennis-Huangm added a commit to Dennis-Huangm/OpenPrism that referenced this pull request Mar 10, 2026
Adopt the PR OpenDCAI#16 frontend companion by normalizing agent reply and suggestion payloads in the API client so provider-specific response shapes no longer leak through to the UI.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant