Fix: prevent Invalid diff now finding less tool calls error by Octane0411 · Pull Request #35379 · openclaw/openclaw

Octane0411 · 2026-03-05T03:53:02Z

Summary

Describe the problem and fix in 2–5 bullets:

Problem: streamed assistant partial payloads can temporarily regress (fewer toolCall blocks than a prior chunk), which triggers downstream stream-diff validation errors like Invalid diff: now finding less tool calls!.
Why it matters: this aborts live runs on affected providers/streams and prevents tool execution from completing.
What changed: in src/agents/pi-embedded-runner/run/attempt.ts, wrapStreamFnTrimToolCallNames now tracks the previous partial snapshot and enforces monotonic tool-call count via preserveMonotonicToolCallsInPartialMessage(...); helper cloning was added to avoid reference pitfalls during patching.
What did NOT change (scope boundary): no provider API contract changes, no tool schema changes, no session/transcript persistence format changes.

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

Closes [Bug]: Frequently "getting Invalid diff: now finding less tool calls!" error on v2026.3.2 #35347

User-visible / Behavior Changes

Streaming runs no longer fail when a provider emits a transient partial snapshot with fewer tool calls than the previous snapshot; tool-call streaming remains stable.

Security Impact (required)

New permissions/capabilities? (No)
Secrets/tokens handling changed? (No)
New/changed network calls? (No)
Command/tool execution surface changed? (No)
Data access scope changed? (No)

Repro + Verification

Environment

OS: macOS (local dev env)
Runtime/container: Bun test runner
Model/provider: streaming provider path (generic; wrapper-level fix)

Steps

Stream assistant partial events where first partial has 2 tool calls and a later partial drops to 1.
Pass through wrapStreamFnTrimToolCallNames.
Observe emitted partial payload tool-call counts.

Expected

Partial tool-call count remains monotonic (does not decrease), avoiding invalid diff failures.

Actual

After fix: second partial retains 2 tool calls in wrapper output (regression prevented).

Evidence

Failing test/log before + passing after

Human Verification (required)

Code path review of wrapper event iteration and partial normalization.
Regression unit test added for shrinking tool-call partial snapshots.
Formatting passes (oxfmt) on touched files.

Compatibility / Migration

Backward compatible? (Yes)
Config/env changes? (No)
Migration needed? (No)

AI Assisted

Codex

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1c49545385

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-03-05T03:56:56Z

src/agents/pi-embedded-runner/run/attempt.ts

+        const codePoint = Number.parseInt(entity.slice(2, -1), 10);
+        return Number.isNaN(codePoint) ? entity : String.fromCodePoint(codePoint);


Validate codepoint range before decoding numeric HTML entities

decodeHtmlEntities decodes any numeric entity with String.fromCodePoint, but entities like &#9999999999; or &#x110000; match this branch and throw RangeError because the value is outside Unicode’s valid range. Since xAI runs enable this decoding path (wrapStreamFnTrimToolCallNames(..., isXaiProvider(...))), a single malformed entity in tool arguments can crash stream processing instead of being treated as plain text. Add an upper/lower bound check (and/or safe fallback) before calling String.fromCodePoint.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-03-05T03:56:56Z

src/agents/pi-embedded-runner/run/attempt.ts

+  const missingCalls = previousToolCalls.slice(currentToolCalls.length).map((block) => cloneUnknown(block));
+  if (missingCalls.length > 0) {
+    currentContent.push(...missingCalls);


Reconstruct dropped tool calls by ID instead of positional slice

The monotonic repair logic assumes only trailing tool calls are missing (previousToolCalls.slice(currentToolCalls.length)), which is incorrect when an earlier call drops while a later one remains. For example, transitioning from [callA, callB] to [callB] appends another clone of callB, yielding duplicated tool calls/IDs in the live partial message; this wrapper runs on the execution stream, so duplicated calls can lead to duplicate tool execution or mismatched tool_result pairing. Compare by call ID/content and append only genuinely missing calls.

Useful? React with 👍 / 👎.

greptile-apps · 2026-03-05T03:59:26Z

Greptile Summary

This PR addresses the "Invalid diff: now finding less tool calls!" streaming error by introducing preserveMonotonicToolCallsInPartialMessage, which detects when a streaming partial message has fewer tool-call blocks than the previous partial and re-injects the "missing" blocks. The PR also adds HTML-entity decoding for xAI providers to handle encoded tool-call arguments.

The core fix is directionally sound for handling the specific issue where streaming partials regress to fewer tool calls. The implementation extracts tool blocks from both current and previous partials, identifies the difference, and appends missing blocks back to maintain monotonicity in the tool call count during streaming.

Confidence Score: 4/5

Safe to merge. The core fix properly handles the streaming regression scenario with appropriate guards and validation.
The PR successfully addresses the reported issue ([Bug]: Frequently "getting Invalid diff: now finding less tool calls!" error on v2026.3.2 #35347) by implementing monotonic tool-call preservation during streaming. The solution uses sound logic to detect regressions (fewer tool calls in new partial vs. previous partial) and re-injects missing blocks. The implementation includes proper type checking, early exits, and uses cloning to avoid unintended reference mutations. While the solution is functional and solves the immediate problem, the confidence is 4 rather than 5 because there are minor areas that could be more defensive (e.g., verifying prefix ordering or capturing state before mutation), though these are edge-case scenarios that are unlikely to occur in normal streaming operation."
No files require special attention. The changes are localized to attempt.ts with clear purpose and appropriate integration points.

_{Last reviewed commit: 1c49545}

Fix: prevent Invalid diff now finding less tool calls error

1c49545

openclaw-barnacle bot added agents Agent runtime and tooling size: S labels Mar 5, 2026

chatgpt-codex-connector bot reviewed Mar 5, 2026

View reviewed changes

voidstar-x mentioned this pull request Mar 5, 2026

Intermittent "23 validation errors: Field required" on tool calls with missing function field #35667

Closed

thewilloftheshadow closed this Mar 6, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Fix: prevent Invalid diff now finding less tool calls error#35379

Fix: prevent Invalid diff now finding less tool calls error#35379
Octane0411 wants to merge 1 commit intoopenclaw:mainfrom
Octane0411:pr-35347-tool-calls-diff

Octane0411 commented Mar 5, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 5, 2026

Uh oh!

chatgpt-codex-connector bot Mar 5, 2026

Uh oh!

greptile-apps bot commented Mar 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		const codePoint = Number.parseInt(entity.slice(2, -1), 10);
		return Number.isNaN(codePoint) ? entity : String.fromCodePoint(codePoint);

Uh oh!

Conversation

Octane0411 commented Mar 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

User-visible / Behavior Changes

Security Impact (required)

Repro + Verification

Environment

Steps

Expected

Actual

Evidence

Human Verification (required)

Compatibility / Migration

AI Assisted

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Mar 5, 2026

Greptile Summary

Confidence Score: 4/5

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Octane0411 commented Mar 5, 2026 •

edited

Loading