chore: release main

[github-actions]

Release preparation

This automated PR prepares the next Rho releases. The sections below list each package version, its generated release notes, and the version files that will be updated.

Merging this PR creates draft GitHub releases, validates and publishes the selected crates, publishes the GitHub releases, and builds the release artifacts.

rho-coding-agent: 1.36.0

1.36.0 (2026-08-11)

Features

• prompt: tell the agent which model runs it, its subagents, and the advisor (#860) (d18c377)

Bug Fixes

• prompt: wait for catalog names before model labels (#863) (71fa544)

Dependencies

• The following workspace dependencies were updated
  • dependencies
    • rho-providers bumped from 0.21.0 to 1.0.0

rho-providers: 1.0.0

1.0.0 (2026-08-11)

⚠ BREAKING CHANGES

• xai: ModelEvent::WebSearch and RunEvent::WebSearch now carry a name field so hosts can distinguish web_search from x_search.

Features

• add advisor mode with a selectable advisor model (#752) (13c1ebb)
• agents: add Claude Code subagent runtime (#541) (c1385ec)
• agents: allow pinning auth profiles on rho agents (#781) (3e1f691)
• agents: move background-run contract into tool descriptions (#405) (b75d0fa)
• auth: add active auth mode switcher (#609) (a2b0f68)
• auth: add configurable credential storage (#478) (e778eda)
• auth: add OpenRouter OAuth login (#472) (42af8e7)
• cli: add structured run output (#467) (c4088bb)
• config: mid-session edit tool, advisor, and auto preference (#840) (423d026)
• openai: add fast mode (#610) (8c5cd6d)
• prompt: tell the agent which model runs it, its subagents, and the advisor (#860) (d18c377)
• providers: add Meta Model API and collapse provider registration (#755) (b41ef92)
• providers: add native Google Gemini support (#430) (34ef307)
• providers: add Ollama Cloud API provider (#597) (f6a62dd)
• providers: add Ollama support (#466) (3a5a6d2)
• providers: add Poolside API platform (#483) (4684de7)
• providers: add Qwen Token Plan OpenAI-compatible provider (#738) (6aa6df2)
• providers: add xAI server-side context compaction (#542) (2d43f13)
• providers: use OpenAI server-side compaction for codex and api-key (#514) (b18eadd)
• providers: use provider-reported costs (#455) (27a8c27)
• questionnaire: support choice descriptions (#510) (066899c)
• readmes for extracted library crates (#388) (92c234d)
• sdk: execute independent tool calls concurrently (#459) (0bb5a83)
• subagents: add parent-child plain-text messaging for Rho runtime (#852) (dd25d8e)
• subagents: route background questionnaires to parent (#539) (e0cab31)
• tools: replace edit_file with codex-style apply_patch (#653) (eef1555)
• tui: include subagent costs in status and info (#548) (9517f00)
• tui: open or copy subagent attach from the activity rail (#552) (95e4ca6)
• tui: show model output token rate (#623) (a5aa688)
• tui: stream apply_patch diff cards (#657) (e2c932e)
• tui: unify tool transcript cards as Call + Children (#586) (ce52cdd)
• web: prefer hosted search with backup provider config (#649) (2e136e9)
• xai: support hosted x_search tool (#647) (cd0c897)

Bug Fixes

• agents: yield while background work completes (#396) (d54e9f3)
• auth: prefer credential-backed auth and fix Ollama Cloud reasoning (#619) (1a57f6f)
• auth: restore ollama device test key dir on unwind and isolate temp dir (#621) (d2a345d)
• auth: stop waiting for Ollama device callback (#616) (54288d2)
• ci: sync released tool dependency versions (#391) (fc78948)
• exclude reasoning tokens from throughput (#819) (d261b5b)
• kimi: use provider-native K3 reasoning (#402) (5453cdc)
• models: reduce GPT-5.6 Codex context window (#470) (2cf9cd6)
• openai: align Codex Responses wire contracts (#644) (76cf855)
• openai: handle terminal Codex websocket events (#421) (c7fb4cd)
• openai: retry empty websocket responses (#476) (04f3844)
• openai: route gpt-5.6 Codex models through standard Responses (#651) (219b9f5)
• poolside: publish final stream usage snapshot (#516) (d51ebab)
• prompt: wait for catalog names before model labels (#863) (71fa544)
• providers: disable parallel tools on codex responses lite (#583) (84ca3f5)
• providers: enrich empty SSE content diagnostic (#684) (79e4d48)
• providers: keep anthropic tool schemas typed after composition strip (#753) (207e74c)
• providers: retry codex server_is_overloaded as unavailable (#641) (9bb2c12), closes #639
• providers: retry transient Gemini finish reasons instead of failing permanently (#449) (041ea9d)
• providers: source Meta Muse Spark reasoning from models.dev (#758) (a11eea0)
• providers: surface rate-limit reset time and /limits pointer (#733) (b9371fc)
• release: align agent tools dependencies (#408) (ba07a4f)
• release: align dependent tool versions (#426) (7b9ea52)
• release: validate internal crate versions (#395) (8f90461)
• sdk: recover failed 1.17.1 release packaging (#587) (224189e)
• sdk: recover failed 1.32.0 release packaging (#792) (a782145)
• sdk: retry retryable provider failures instead of failing the run (#401) (b2867da)
• skills: enforce manual skill invocation (#453) (4f6f043)
• tools: allow file paths outside workspace (#537) (8a3cc24)
• tools: scrub provider credential env vars from child processes (#502) (6d66913)
• tui: improve agent tool displays (#413) (062edd0)
• tui: lead approval prompts with the command (#846) (413063c)
• tui: render narrow mermaid flowcharts and explain fallbacks (#565) (0bf7ad7)
• tui: reuse tool stream previews and allow codex parallel tools (#566) (fa0074a)
• tui: show codex fast mode and report tier fallback (#663) (177043f)
• tui: show hosted x_search tool cards (#662) (4381667)
• tui: sort slash commands and provider pickers alphabetically (#498) (0e2c16c)
• tui: wait for delegated goal work (#457) (fc6087d)
• usage: normalize cache write token accounting (#511) (4e15982)

---

This PR was generated with Release Please. See documentation.

[github-actions]

🤖 Created releases:

• rho-coding-agent-v1.36.0
• rho-providers-v1.0.0

🌻