fix(assistant): execute provider text tool calls

[omarluq]

Summary

• execute XML/text fallback tool calls from providers that do not emit native tool-call JSON
• preserve write/edit/bash content when parsing fallback tool calls, including JSON container payloads
• reject empty writes to avoid accidental zero-byte files
• add Anthropic native tool-name compatibility in the stacked branch

Validation

• mise exec -- go test ./...
• mise exec -- task build
• mise exec -- task ci

[coderabbitai]

Caution

Review failed

Failed to post review comments

📝 Walkthrough

Summary by CodeRabbit

• New Features

  • Text-based tool-call fallback to handle tool requests embedded in assistant text.

• Improvements

  • Better provider-aware tool name mapping for Anthropic/OAuth scenarios.
  • Tool payloads now signal eager input streaming for faster interactions.

• Tests

  • Expanded test coverage for tool parsing, mapping, execution, and error cases.

Walkthrough

OAuth-aware Anthropic tool naming maps provider/local names per request, adds eager input streaming flags, and propagates CompletionRequest through the tool loop. Separately, text-fallback parsing now iteratively extracts and normalizes XML-like <tool_use> blocks into canonical tool-call arguments. Tests updated and expanded.

Changes

Anthropic OAuth Tool Naming and Text Fallback Parsing

Layer / File(s)	Summary
Tool name and JSON key constants internal/assistant/client.go	Adds jsonInputKey and extends tool-name constants (including find/ls and capitalized Anthropic variants).
Anthropic OAuth-aware naming & payload wiring internal/assistant/anthropic.go, internal/assistant/anthropic_tools_test.go	Propagates *CompletionRequest to appendAnthropicToolConversation, makes anthropicTools(request) request-dependent, adds eager_input_streaming: true, introduces anthropicProviderToolName/anthropicLocalToolName mapping (OAuth-aware, includes List alias), and updates tool-call/message translation and tests.
Text-fallback parsing and argument normalization internal/assistant/text_tool_calls.go	Implements iterative extraction of <tool_use> blocks, manual tag matching, HTML-unescape, JSON-embedded payload normalization (stringify non-strings), canonical tool-name normalization, per-tool aliasing to standardized argument keys, and JSON-encoded arguments with {} fallback.
Text-fallback unit tests internal/assistant/text_tool_calls_test.go	Adds table-driven tests covering multiline write content preservation, multiple markup variants, name/argument normalization, JSON payload parsing, and textToolResultPrompt behavior for errors and empty output.
Integration test workspace and shared constants internal/assistant/provider_tool_calls_test.go, internal/assistant/tool_loop_test.go	Introduces testToolWorkspace() with README.md fixture, switches tests to per-test CWD, marshals OpenAI tool arguments as JSON in mocks, and centralizes testCallID and jsonReadToolName usage in tests.

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Possibly related PRs

• omarluq/librecode#32: Updates Anthropic tool-use parsing and conversation wiring, overlapping with OAuth-aware tool name mapping in this PR.

"🐰
With OAuth names mapped clean,
XML fields parsed between,
Tool calls now find the right name,
Multiline writes kept the same,
Tests hop through the green!"

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: executing provider text tool calls as a fix to the assistant module.
Description check	✅ Passed	The description clearly relates to the changeset, outlining the execution of XML/text fallback tool calls, content preservation, and Anthropic compatibility additions.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/provider-tool-calls

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.6% Duplication on New Code

See analysis details on SonarQube Cloud

[codecov-commenter]

Codecov Report

❌ Patch coverage is 77.41935% with 35 lines in your changes missing coverage. Please review.
✅ Project coverage is 60.12%. Comparing base (0a4eae4) to head (2f28833).

Files with missing lines	Patch %	Lines
internal/assistant/anthropic.go	67.27%	17 Missing and 1 partial ⚠️
internal/assistant/text_tool_calls.go	83.00%	13 Missing and 4 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #34      +/-   ##
==========================================
+ Coverage   59.97%   60.12%   +0.15%     
==========================================
  Files         168      168              
  Lines       16532    16667     +135     
==========================================
+ Hits         9915    10021     +106     
- Misses       5611     5636      +25     
- Partials     1006     1010       +4     

Flag	Coverage Δ
unittests	60.12% <77.41%> (+0.15%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[omarluq]

@coderabbitai review

[coderabbitai]

✅ Actions performed

Review triggered.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.