v0.4.0

• fix: text selection bug (#1664)
• fix: modal pastes (#1677)
• fix: get session api (#1684)
• Merge agent and mode into one (#1689) The concept of mode has been deprecated, there is now only the agent field in the config. An agent can be cycled through as your primary agent with or you can spawn a subagent by @ mentioning it. if you include a description of when to use it, the primary agent will try to automatically use it Full docs here: https://opencode.ai/docs/agents/
• looser todo tool schema
• renable todo tool

v0.3.133

• fix: interface conversion panic (#1655)
• fix(docs): small_model is not used for summarization (#1360)
• add api to get session

v0.3.132

• fix: prevent title regeneration on auto compact (#1628)
• fix: markdown table renders (#1623)
• docs(plugins): fix typo (#1621)
• feat: better assistant message visual
• permissions disallow support (#1627)
• fix: handle undefined agent in task tool (#1642)
• fix: highlight after text wrap (#1640)
• fix: run command should use specified model from cli args if provided (#1648)
• add snapshot field in config to disable snapshots

v0.3.131

• fix(plugins): improve session idle event (#1615)
• fix: server.root is not a function (#1614)
• validate file part bounds to prevent panic (#1612)
• fix: allow disabling the default general agent (#1616)
• hint back to llm when tool does not exist

v0.3.130

• re-export shell $ for plugin

v0.3.129

• added vesper theme (#1602)
• fix: default value for models with no cost object (#1601)
• fix @opencode-ai/plugin exports

v0.3.128

• hack to return tool call errors back to model

v0.3.127

• Add typecheck workflow
• fix: patch marked-shiki (#1569)
• Add workflow_dispatch trigger to typecheck workflow
• fix docs
• fix double help printing (#1580)

v0.3.126

• fix bash tool getting stuck on interactive commands
• increase models.dev polling interval to hourly
• tweak: filter out duplicate instructions (#1567)
• fix bash tool not showing stderr

v0.3.122

• more efficient snapshots in parallel toolcalls