Support updated upstream submodule APIs

[senamakel]

Update the tinyinference and tinytools submodule pointers to their upstream main heads (and the wiki to its upstream default branch), then adapt TinyAgents to the expanded message and tool-content APIs. Preserve media tool output, filter host-only custom messages before provider serialization, and update tests/examples for new required fields and variants.\n\nValidation:\n- cargo fmt --check\n- git diff --check\n- cargo build --workspace --all-targets\n- cargo test --workspace

Summary by CodeRabbit

• New Features

  • Tool dispatching now selects native or XML-based formats based on model capabilities and configuration.
  • Provider-specific content blocks are preserved and displayed during streaming.
  • Hosted subagent interactions support configurable delegation with clearer provider error reporting.

• Bug Fixes

  • Host-only records are excluded from provider requests and tool-result processing.
  • Provider-extension content is retained correctly when completing streamed responses.

• Tests

  • Expanded coverage for tool dialect recovery, hosted subagents, provider extensions, media rendering, and serialization contracts.

[tinysweeper]

Tiny Sweeper review

⚠️ Review failed for 7fd0d95c32c2. the review of #179 did not finish within 900s

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Advanced

Run ID: c51c65fb-c1d6-47e1-ae3e-265cfd8aab47

📥 Commits

Reviewing files that changed from the base of the PR and between 86d9dbf and 7fd0d95.

📒 Files selected for processing (6)

• .github/workflows/ci.yml
• crates/tinyagents-harness/src/agent_loop/dialect.rs
• crates/tinyagents-harness/src/agent_loop/dialect/test.rs
• crates/tinyagents-harness/src/providers/claude_code/mod.rs
• crates/tinyagents-harness/src/summarization/test.rs
• crates/tinyagents-integration-tests/tests/dependency_boundary.rs

🚧 Files skipped from review as they are similar to previous changes (2)

• crates/tinyagents-harness/src/agent_loop/dialect.rs
• crates/tinyagents-harness/src/providers/claude_code/mod.rs

Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review.

---

📝 Walkthrough

Walkthrough

The changes make tool dialect selection model-aware, filter custom Claude Code records before coalescing, retain provider-extension stream blocks, update rendering and integration tests, adjust CI coverage setup, and advance the tinyinference reference.

Changes

Harness behavior updates

Layer / File(s)	Summary
Native tool-calling dialect resolution crates/tinyagents-harness/src/agent_loop/dialect.rs, crates/tinyagents-harness/src/agent_loop/run_loop.rs, crates/tinyagents-harness/src/agent_loop/test.rs, crates/tinyagents-integration-tests/tests/e2e_tool_dialects.rs	RunDialect::resolve now receives the model capability. ToolDispatcher::Auto selects XML when native tool calling is disabled. Related tests configure model profiles and run policies.
Claude Code message filtering crates/tinyagents-harness/src/providers/claude_code/mod.rs, crates/tinyagents-harness/src/providers/claude_code/mod_tests.rs, crates/tinyagents-integration-tests/tests/dependency_boundary.rs	Custom messages are removed before tool-result coalescing. Tests cover the filtering behavior. The dependency-boundary baseline tracks the moved references.
Provider-extension stream reduction crates/tinyagents-harness/src/stream/frame.rs, crates/tinyagents-harness/src/stream/frame/test.rs, crates/tinyagents-harness/src/summarization/render.rs, crates/tinyagents-harness/src/summarization/test.rs	OpenBlock represents provider extensions. Frame reduction and rendering handle the new variant. Tests verify completed content and summary rendering. Media placeholder outputs remain unchanged.

Integration and infrastructure alignment

Layer / File(s)	Summary
Integration contract updates crates/tinyagents-integration-tests/tests/e2e_registry_observability_contracts.rs, crates/tinyagents-orchestration/tests/hosted_subagents.rs	The component-kind contract expects 12 variants. Hosted subagent tests configure delegation and tool access and expect the updated provider error text.
Build and vendor updates .github/workflows/ci.yml, vendor/tinyinference	The coverage job cleans build artifacts and disables debug information. The tinyinference subproject pointer advances to commit aeaecda26677161713b3a993872813e2a3594c21.

Priority: ⬇️ Low

Estimated code review effort: 3 (Moderate) | ~25 minutes

Change: Feature

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the API compatibility updates and upstream submodule changes. It does not cover every test and behavior update, but it clearly identifies the primary change.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 20 functions across 14 files. (1 skipped: …
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.

---

_{A rabbit checks the dialect gate
XML waits when tools cannot operate
Custom crumbs stay out of Claude’s tray
Provider blocks keep their shape in play
Tests hop through contracts, clean and bright
CI sheds artifacts before the night}

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

[tinysweeper]

Requesting changes: 2 lane(s) blocking, worst finding is high.

Fix or reply to the findings below and push. The next review clears this automatically once they are gone — you should not need to dismiss anything by hand.

             $0.1848 · 2,857,288 in / 57,594 out · 138,771 cached (5%) · ladder/vectors, gpt-5.6-luna, deepseek/deepseek-v4-flash · 829 embedded
critique:    $0.0941 · 1,480,299 in / 24,041 out · 72,955 cached (5%)  · gpt-5.6-luna, deepseek/deepseek-v4-flash
security:    $0.0834 · 1,313,443 in / 17,124 out · 46,616 cached (4%)  · gpt-5.6-luna
tests:       $0.0034 · 28,913 in    / 4,796 out  · 1,280 cached (4%)   · deepseek/deepseek-v4-flash
description: $0.0026 · 16,495 in    / 6,294 out  · 0 cached (0%)       · deepseek/deepseek-v4-flash

[tinysweeper]

The previously-blocking findings are resolved. Clearing the changes request.

             $0.0400 · 522,406 in / 39,856 out · 41,473 cached (8%)  · ladder/vectors, gpt-5.6-luna, deepseek/deepseek-v4-flash · 991 embedded
critique:    $0.0189 · 266,007 in / 11,174 out · 12,433 cached (5%)  · gpt-5.6-luna, deepseek/deepseek-v4-flash
security:    $0.0145 · 206,948 in / 7,067 out  · 5,488 cached (3%)   · gpt-5.6-luna
tests:       $0.0038 · 35,594 in  / 10,096 out · 17,152 cached (48%) · deepseek/deepseek-v4-flash
description: $0.0016 · 7,246 in   / 5,481 out  · 0 cached (0%)       · deepseek/deepseek-v4-flash

[tinysweeper]

Requesting changes: 1 lane(s) blocking, worst finding is high.

Fix or reply to the findings below and push. The next review clears this automatically once they are gone — you should not need to dismiss anything by hand.

             $0.0058 · 70,909 in / 13,525 out · 29,762 cached (42%) · ladder/vectors, gpt-5.6-luna, deepseek/deepseek-v4-flash · 1,002 embedded
critique:    $0.0017 · 26,997 in / 845 out    · 4,132 cached (15%)  · gpt-5.6-luna
security:    $0.0008 · 12,670 in / 531 out    · 1,822 cached (14%)  · gpt-5.6-luna
tests:       $0.0014 · 17,302 in / 6,035 out  · 17,152 cached (99%) · deepseek/deepseek-v4-flash
description: $0.0010 · 7,279 in  / 2,155 out  · 0 cached (0%)       · deepseek/deepseek-v4-flash