3.7 KiB
Agent Note: Keep the Code Mode result card complete
Status: implemented
English | 中文
Problem
The outer run_code tool persisted complete rendered content, but its editor presenter ignored that content and rebuilt the card body from a logs-only presentationMeta projection. A result-only run appeared correct because an empty presenter body let ACP and TUI fall back to tool/result.content. Once the program emitted a log, the presenter supplied non-empty content, that fallback stopped, and the returned value disappeared from the completed card. A spill policy's final head/tail preview was vulnerable to the same split ownership whenever captured logs made the stale projection non-empty.
Nested Code calls never owned cards, so producing metadata for the outer call solely to reconstruct one incomplete card also obscured the intended one-card boundary.
Decision
The canonical tool registry pipeline owns the final model-facing outer content. On success, the run_code output renderer renders captured logs followed by the return value or the explicit no-output marker. Runtime failures and pre-execution policy denials are normalized into error content by ToolRegistry without invoking that renderer. A post-execute block runs after successful rendering and replaces the result with error content; other post-execute policy and spill decisions may replace content before persistence.
run_code.presentResult now forwards the final result.content into one generic result card. It deliberately omits the title so the pending card retains the program text. The redundant logs-only presentationMeta projection is removed: tool/result.content is the durable, replayable, post-policy projection and the card's only result-content source.
Nested dispatch remains unchanged. Calls marked by exec.parent emit bounded tool/code-dispatch diagnostics but no tool/call or tool/result surface cards, so one outer run_code invocation still produces exactly one card.
Testing
Presenter unit coverage pins logs-only, result-only, logs-plus-result, no-output, and spilled-result content. A separate integration-shaped unit drives a real runtime failure through the canonical registry result before presenting it. The successful cases prove stale metadata cannot replace final content; the failure case guards complete forwarding without claiming it reproduced the original metadata-triggered defect.
The keyless ACP and TUI Code Mode snapshots execute one outer program that performs two nested bash calls, logs captured output, and returns CODE_ONE+CODE_TWO. Both surfaces show one completed outer card containing both lines and no nested cards.
Alternatives considered
Append the return value to logs metadata. Rejected because metadata would duplicate the renderer, need a second stable formatting contract for every JSON root, and still miss post-policy content replacement or spill previews.
Merge presenter metadata with result.content. Rejected because the rendered content already contains the logs; merging would duplicate them and require brittle deduplication.
Create one card per nested dispatch. Rejected because intermediate values are intentionally execution-local and never model-facing. Multiple cards would expose an implementation trace instead of the single Code Mode operation the model and user invoked.
Consequences
ACP and TUI now display the same complete content the model receives and replay persists, including post-policy spill previews. New run_code results no longer carry the optional logs metadata, but this requires no session-format bump: existing records remain valid because the presenter ignores that field and reads their durable rendered content.