chore(deps): bump the production group across 1 directory with 9 updates

[dependabot]

Bumps the production group with 9 updates in the / directory:

Package	From	To
@composio/core	0.10.0	0.13.1
@earendil-works/pi-agent-core	0.79.9	0.80.3
@earendil-works/pi-ai	0.79.9	0.80.3
@earendil-works/pi-coding-agent	0.79.9	0.80.3
@earendil-works/pi-tui	0.79.9	0.80.3
@slack/web-api	7.17.0	7.18.0
convex	1.41.0	1.42.1
playwright-core	1.60.0	1.61.1
tar	7.5.16	7.5.19

Updates @composio/core from 0.10.0 to 0.13.1

Release notes

Sourced from @​composio/core's releases.

@​composio/core@​0.13.1

Patch Changes

• 605a726: Add Tool Router session deletion APIs.

@​composio/core@​0.13.0

Minor Changes

• d17a268: Add the first-class composio.sessions.create() API while keeping composio.create() as an alias, expose the experimental shared-connection ACL patch helper as connectedAccounts.updateAcl() while keeping experimental.updateAcl() as an alias, and include SDK docs/source in the published package.

  MCP is now opt-in. Sessions return native tools by default; the hosted MCP endpoint is only surfaced on the type when you create the session with { mcp: true }. The default create() / use() now return SessionWithoutMcp (the runtime object is unchanged — session.mcp still exists at runtime — but it is no longer in the type).

  Migration: read session.mcp only after creating with { mcp: true }.

• d17a268: Surface the resolved workbench config on Tool Router sessions.

  • Session.workbench is now populated from the API response (on create/retrieve/attach/update). It exposes the resolved workbench config, e.g. session.workbench?.enable (defaults to true server-side).

  This lets callers create a session with the remote workbench disabled (workbench: { enable: false }) and detect that state — the foundation for running code in a sandbox you own via the experimental @composio/experimental/workbench helpers.

Patch Changes

• d17a268: Prefer sandbox for session code-execution configuration while continuing to accept the existing workbench alias.
• d17a268: Add triggers.parse() to parse and optionally verify incoming webhook requests.
• d17a268: Add triggers.setWebhookSubscription() to create or update the project webhook subscription from the TypeScript SDK.

@​composio/core@​0.12.0

Minor Changes

• a0bef5d: Bump @composio/client to 0.1.0-alpha.74.

• dfd7a08: Add per-request cancellation to public SDK methods via a new ComposioRequestOptions ({ signal?: AbortSignal }) trailing argument, plus a typed ComposioRequestCancelledError for detecting caller-initiated aborts.

  Without this, a slow tools.get or tools.execute had no way to be cancelled — a 100s search would block the calling agent indefinitely. The new shape:

  try {
    const tools = await composio.tools.get(
      'user_1',
      { search: 'send email', limit: 50 },
      { signal: AbortSignal.timeout(5_000) }
    );
  } catch (err) {
    if (err instanceof ComposioRequestCancelledError) {
      return;
    }
    throw err;
  }

  The signal is forwarded to the underlying @composio/client fetch. Any abort error (APIUserAbortError, AbortError, or DOMException(name='AbortError')) coming back is normalized to ComposioRequestCancelledError so callers can instanceof-detect cancellation without unwrapping nested causes. Catch-and-wrap paths in tools.execute / tools.getRawComposioToolBySlug / toolkits.get re-throw the cancellation error rather than remapping it to ComposioToolExecutionError / ComposioToolNotFoundError / ComposioToolkitFetchError.

... (truncated)

Changelog

Sourced from @​composio/core's changelog.

0.13.1

Patch Changes

• 605a726: Add Tool Router session deletion APIs.

0.13.0

Minor Changes

• d17a268: Add the first-class composio.sessions.create() API while keeping composio.create() as an alias, expose the experimental shared-connection ACL patch helper as connectedAccounts.updateAcl() while keeping experimental.updateAcl() as an alias, and include SDK docs/source in the published package.

  MCP is now opt-in. Sessions return native tools by default; the hosted MCP endpoint is only surfaced on the type when you create the session with { mcp: true }. The default create() / use() now return SessionWithoutMcp (the runtime object is unchanged — session.mcp still exists at runtime — but it is no longer in the type).

  Migration: read session.mcp only after creating with { mcp: true }.

• d17a268: Surface the resolved workbench config on Tool Router sessions.

  • Session.workbench is now populated from the API response (on create/retrieve/attach/update). It exposes the resolved workbench config, e.g. session.workbench?.enable (defaults to true server-side).

  This lets callers create a session with the remote workbench disabled (workbench: { enable: false }) and detect that state — the foundation for running code in a sandbox you own via the experimental @composio/experimental/workbench helpers.

Patch Changes

• d17a268: Prefer sandbox for session code-execution configuration while continuing to accept the existing workbench alias.
• d17a268: Add triggers.parse() to parse and optionally verify incoming webhook requests.
• d17a268: Add triggers.setWebhookSubscription() to create or update the project webhook subscription from the TypeScript SDK.

0.12.0

Minor Changes

• a0bef5d: Bump @composio/client to 0.1.0-alpha.74.

• dfd7a08: Add per-request cancellation to public SDK methods via a new ComposioRequestOptions ({ signal?: AbortSignal }) trailing argument, plus a typed ComposioRequestCancelledError for detecting caller-initiated aborts.

  Without this, a slow tools.get or tools.execute had no way to be cancelled — a 100s search would block the calling agent indefinitely. The new shape:

  try {
    const tools = await composio.tools.get(
      'user_1',
      { search: 'send email', limit: 50 },
      { signal: AbortSignal.timeout(5_000) }
    );
  } catch (err) {
    if (err instanceof ComposioRequestCancelledError) {
      return;
    }
    throw err;
  }

... (truncated)

Commits

• 156a1eb Release: update version (#3686)
• 956f9be chore(agents): normalize repo guidance skills (#3666)
• 605a726 Add tool router session deletion (#3679)
• 20bf1b6 chore(ts): quiet tsdown build warnings (#3665)
• d102ab0 Release: update version (#3676)
• d17a268 docs: sessions-first rewrite — new guides, examples & components (+ core 0.13...
• 9764ce3 Release: update version (#3616)
• f6b5505 fix(core): preserve SDK reference types
• 507318d fix(anthropic): harden tool property-key sanitizer (#3650)
• dfd7a08 feat(core): add AbortSignal cancellation to public SDK methods + ComposioRequ...
• Additional commits viewable in compare view

Maintainer changes

This version was pushed to npm by GitHub Actions, a new releaser for @​composio/core since your current version.

Updates @earendil-works/pi-agent-core from 0.79.9 to 0.80.3

Release notes

Sourced from @​earendil-works/pi-agent-core's releases.

v0.80.3

New Features

• Anthropic Claude Sonnet 5 support - Claude Sonnet 5 is available through inherited Anthropic-compatible and Bedrock provider catalogs with adaptive thinking enabled. See Providers and Model Options.
• Configurable output spacing - outputPad controls horizontal padding for user messages, assistant messages, and thinking blocks. See Settings.
• External editor configuration - externalEditor lets Ctrl+G use a configured editor before $VISUAL/$EDITOR fallbacks. See Settings and Keybindings.
• Richer RPC session tree access - RPC clients can inspect session entries and tree snapshots with get_entries and get_tree. See get_entries and get_tree.
• Extension session metadata updates - Extensions can observe session name changes through session_info_changed. See session_info_changed.
• Modern Azure Foundry endpoint support - Azure OpenAI Responses provider setup supports current Microsoft Foundry endpoint URLs. See Azure OpenAI.

Added

• Added inherited Anthropic Claude Sonnet 5 model support.
• Added get_entries and get_tree RPC commands for reading session entries and tree snapshots over RPC (#6078 by @​geraschenko).
• Added a package ./rpc-entry export for launching Pi directly in RPC mode.
• Added session-name change events for extensions (#6175 by @​xl0).
• Added inherited Azure OpenAI Responses support for modern Microsoft Foundry endpoint URLs (#6004 by @​gukoff).
• Added inherited Usage.reasoning token counts for providers that report reasoning/thinking token usage (#6057).
• Added an externalEditor settings.json override for Ctrl+G external editor commands, with default fallbacks to Notepad on Windows and nano elsewhere (#6122).
• Added an outputPad setting for user message, assistant message, and thinking horizontal padding (#6168).

Changed

• Changed the default OpenAI model to gpt-5.5.
• Changed inherited OpenAI Codex Responses SSE response-header waits to use the configured HTTP timeout instead of the previous fixed 20 second timeout, reducing false timeouts on slow connections (#4945).

Fixed

• Fixed inherited Claude Sonnet 5 metadata to use adaptive thinking payloads for Anthropic-compatible and Bedrock requests.
• Fixed inherited generated Xiaomi MiMo model pricing to match current pay-as-you-go pricing from models.dev (#6138).
• Fixed inherited provider HTTP errors to include response bodies instead of opaque SDK messages (#5832 by @​stephanmck).
• Fixed inherited streamSimple() max-token caps so providers that count input and output against one context window do not reject long requests (#5595).
• Fixed inherited OpenAI Responses streams to preserve reasoning replay state when output items finish out of order (#6009).
• Fixed inherited Z.AI preserved thinking requests to send thinking.clear_thinking: false when thinking is enabled, allowing replayed reasoning_content to participate in provider caching (#6083).
• Fixed pre-prompt compaction to stop after compaction instead of continuing immediately (#6074 by @​yzhg1983).
• Fixed resource notifications to stay before messages when resuming sessions (#6048 by @​haoqixu).
• Fixed startup benchmark timing output to print after TUI shutdown, preserve extension timings, and drain terminal-query replies before stopping benchmark mode (#6030 by @​xl0, #6063 by @​xl0).
• Fixed extension tool changes to apply before the next provider request in the same agent run without dropping before_agent_start system-prompt overrides (#6162).
• Fixed a crash when undici emits an internal client error while terminating a mid-stream HTTP response (#6133).
• Fixed the compaction event regression test to cover status indicator cleanup and keep CI passing.
• Fixed interactive status indicators so ending work, retry, compaction, or branch-summary indicators no longer shrink the TUI when clear-on-shrink is enabled (#6026).
• Fixed --session and SessionManager.open() to reject non-empty invalid session files without overwriting them (#6002).
• Fixed user-message transcript rendering to keep visible backslashes in Markdown escape sequences such as \" (#6105).
• Fixed assistant messages stopped by output length to show a visible incomplete-response error (#4290).
• Fixed --no-session --session-id so ephemeral CLI runs can use deterministic session IDs for provider cache affinity (#6070).
• Fixed disk BMP image files to be detected, converted to PNG, and attached through read and CLI @file inputs (#6047).
• Fixed auto-retry for provider stream errors that explicitly tell callers to retry the request (#6019).

v0.80.2

Changed

... (truncated)

Changelog

Sourced from @​earendil-works/pi-agent-core's changelog.

[0.80.3] - 2026-06-30

Added

• Added prepareNextTurnWithContext for Agent users that need the next-turn loop context.

Fixed

• Fixed oversized harness shell execution timeouts to fail with a clear validation error instead of being clamped to an immediate timeout (#6181).
• Fixed Agent.prepareNextTurn to keep receiving the run abort signal instead of the next-turn context.

[0.80.2] - 2026-06-23

Changed

• Renamed the public harness shell execution options type from ExecutionEnvExecOptions to ShellExecOptions.

[0.80.1] - 2026-06-23

[0.80.0] - 2026-06-23

Breaking Changes

• AgentHarnessOptions.models is required and is the only auth path: the harness streams turns, compaction, and branch summarization through the provided Models instance (models.streamSimple()/completeSimple()), resolving auth through the providers. AgentHarnessOptions.getApiKeyAndHeaders is removed — apps that resolved keys per request now express that as provider auth (ApiKeyAuth/OAuthAuth) on the providers in the Models collection. Build one with createModels() + provider factories (or builtinModels() from @earendil-works/pi-ai/providers/all); tests use fauxProvider().
• compact(), generateSummary(), and generateBranchSummary() take a Models parameter and no longer accept explicit apiKey/headers.
• StreamFn is defined structurally ((model, context, options?) => AssistantMessageEventStream | Promise<...>); Models.streamSimple satisfies it.
• Removed the @earendil-works/pi-agent-core/base selective-provider entrypoint; use the root package with an explicit Models instance instead.

Fixed

• Fixed harness session names to normalize newline characters before storing labels (#5999 by @​haoqixu).
• Fixed harness compaction estimates to ignore malformed all-zero assistant usage after truncated responses (#5526 by @​dmmulroy).

[0.79.10] - 2026-06-22

Commits

• a23abe4 Release v0.80.3
• fd6659d fix(coding-agent): preserve run prompt during tool refresh
• e547bb9 fix(coding-agent): refresh session state before next turn
• c29bbc0 docs(agent): update models phase 9 plan
• 8277bd6 Add [Unreleased] section for next cycle
• 0201806 Release v0.80.2
• 9096d5f docs: update changelog entries
• 49fbe68 fix(ai): align api key credentials with auth json
• b377623 Type name change
• e000743 Add [Unreleased] section for next cycle
• Additional commits viewable in compare view

Updates @earendil-works/pi-ai from 0.79.9 to 0.80.3

Release notes

Sourced from @​earendil-works/pi-ai's releases.

v0.80.3

New Features

• Anthropic Claude Sonnet 5 support - Claude Sonnet 5 is available through inherited Anthropic-compatible and Bedrock provider catalogs with adaptive thinking enabled. See Providers and Model Options.
• Configurable output spacing - outputPad controls horizontal padding for user messages, assistant messages, and thinking blocks. See Settings.
• External editor configuration - externalEditor lets Ctrl+G use a configured editor before $VISUAL/$EDITOR fallbacks. See Settings and Keybindings.
• Richer RPC session tree access - RPC clients can inspect session entries and tree snapshots with get_entries and get_tree. See get_entries and get_tree.
• Extension session metadata updates - Extensions can observe session name changes through session_info_changed. See session_info_changed.
• Modern Azure Foundry endpoint support - Azure OpenAI Responses provider setup supports current Microsoft Foundry endpoint URLs. See Azure OpenAI.

Added

• Added inherited Anthropic Claude Sonnet 5 model support.
• Added get_entries and get_tree RPC commands for reading session entries and tree snapshots over RPC (#6078 by @​geraschenko).
• Added a package ./rpc-entry export for launching Pi directly in RPC mode.
• Added session-name change events for extensions (#6175 by @​xl0).
• Added inherited Azure OpenAI Responses support for modern Microsoft Foundry endpoint URLs (#6004 by @​gukoff).
• Added inherited Usage.reasoning token counts for providers that report reasoning/thinking token usage (#6057).
• Added an externalEditor settings.json override for Ctrl+G external editor commands, with default fallbacks to Notepad on Windows and nano elsewhere (#6122).
• Added an outputPad setting for user message, assistant message, and thinking horizontal padding (#6168).

Changed

• Changed the default OpenAI model to gpt-5.5.
• Changed inherited OpenAI Codex Responses SSE response-header waits to use the configured HTTP timeout instead of the previous fixed 20 second timeout, reducing false timeouts on slow connections (#4945).

Fixed

• Fixed inherited Claude Sonnet 5 metadata to use adaptive thinking payloads for Anthropic-compatible and Bedrock requests.
• Fixed inherited generated Xiaomi MiMo model pricing to match current pay-as-you-go pricing from models.dev (#6138).
• Fixed inherited provider HTTP errors to include response bodies instead of opaque SDK messages (#5832 by @​stephanmck).
• Fixed inherited streamSimple() max-token caps so providers that count input and output against one context window do not reject long requests (#5595).
• Fixed inherited OpenAI Responses streams to preserve reasoning replay state when output items finish out of order (#6009).
• Fixed inherited Z.AI preserved thinking requests to send thinking.clear_thinking: false when thinking is enabled, allowing replayed reasoning_content to participate in provider caching (#6083).
• Fixed pre-prompt compaction to stop after compaction instead of continuing immediately (#6074 by @​yzhg1983).
• Fixed resource notifications to stay before messages when resuming sessions (#6048 by @​haoqixu).
• Fixed startup benchmark timing output to print after TUI shutdown, preserve extension timings, and drain terminal-query replies before stopping benchmark mode (#6030 by @​xl0, #6063 by @​xl0).
• Fixed extension tool changes to apply before the next provider request in the same agent run without dropping before_agent_start system-prompt overrides (#6162).
• Fixed a crash when undici emits an internal client error while terminating a mid-stream HTTP response (#6133).
• Fixed the compaction event regression test to cover status indicator cleanup and keep CI passing.
• Fixed interactive status indicators so ending work, retry, compaction, or branch-summary indicators no longer shrink the TUI when clear-on-shrink is enabled (#6026).
• Fixed --session and SessionManager.open() to reject non-empty invalid session files without overwriting them (#6002).
• Fixed user-message transcript rendering to keep visible backslashes in Markdown escape sequences such as \" (#6105).
• Fixed assistant messages stopped by output length to show a visible incomplete-response error (#4290).
• Fixed --no-session --session-id so ephemeral CLI runs can use deterministic session IDs for provider cache affinity (#6070).
• Fixed disk BMP image files to be detected, converted to PNG, and attached through read and CLI @file inputs (#6047).
• Fixed auto-retry for provider stream errors that explicitly tell callers to retry the request (#6019).

v0.80.2

Changed

... (truncated)

Changelog

Sourced from @​earendil-works/pi-ai's changelog.

[0.80.3] - 2026-06-30

Added

• Added Anthropic Claude Sonnet 5 model metadata for Anthropic-compatible, Bedrock, OpenRouter, and Vercel AI Gateway providers.
• Added Azure OpenAI Responses support for modern Microsoft Foundry endpoint URLs (#6004 by @​gukoff).
• Added an optional reasoning field to Usage reporting reasoning/thinking token counts as a subset of output. Populated for Anthropic (output_tokens_details.thinking_tokens), OpenAI Responses/Codex/Azure (output_tokens_details.reasoning_tokens), OpenAI Completions (completion_tokens_details.reasoning_tokens), and Google Generative AI / Vertex (thoughtsTokenCount). Bedrock Converse and Mistral are not populated because those APIs do not return a reasoning token breakdown (#6057).

Changed

• Changed OpenAI Codex Responses SSE response-header waits to use the configured HTTP timeout instead of the previous fixed 20 second timeout, reducing false timeouts on slow connections (#4945).

Fixed

• Fixed Claude Sonnet 5 metadata to use adaptive thinking payloads for Anthropic-compatible and Bedrock requests.
• Fixed generated Xiaomi MiMo model pricing to match current pay-as-you-go pricing from models.dev (#6138).
• Fixed provider HTTP errors to include response bodies instead of opaque SDK messages (#5832 by @​stephanmck).
• Fixed streamSimple() to send a context-aware max-token cap so providers that count input and output against one context window do not reject long requests (#5595).
• Fixed OpenAI Responses streams to preserve reasoning replay state when output items finish out of order (#6009).
• Fixed retry classification for provider errors that explicitly tell callers to retry the request (#6019).
• Fixed Z.AI preserved thinking requests to send thinking.clear_thinking: false when thinking is enabled, allowing replayed reasoning_content to participate in provider caching (#6083).

[0.80.2] - 2026-06-23

Changed

• Changed ApiKeyCredential to use the auth.json-compatible discriminator type: "api_key" and provider-scoped env values instead of type: "api-key" and metadata.

Fixed

• Fixed Anthropic-compatible custom models to use explicit compatibility metadata instead of provider-name heuristics for session-affinity headers and unsupported tool-field omissions.
• Fixed request-scoped apiKey and env values to participate in provider auth resolution, so providers such as Cloudflare can derive request-specific base URLs from explicit call options (#6021).
• Restored temporary legacy per-API stream aliases such as streamSimpleOpenAICompletions on the compat entrypoint (#6016, #6017).
• Restored runtime detectCompat fallback in openai-completions for models without explicit compat metadata (#6020).

[0.80.1] - 2026-06-23

Fixed

• Fixed a regression in Amazon Bedrock scoped AWS_PROFILE endpoint resolution for built-in inference profile endpoints.
• Fixed Fireworks Anthropic-compatible requests to apply session-affinity and unsupported tool-field defaults for custom Fireworks models.
• Fixed Together MiniMax M2.7 metadata to avoid unsupported Together reasoning toggles.

[0.80.0] - 2026-06-23

Breaking Changes

• The root entrypoint (@earendil-works/pi-ai) is now core-only and side-effect free. The old global API moved to the temporary @earendil-works/pi-ai/compat entrypoint, a strict superset of the root: switching a file's import path is the only migration step. Moved symbols include stream/complete/streamSimple/completeSimple, getModel/getModels/getProviders (now deprecated aliases of getBuiltinModel/getBuiltinModels/getBuiltinProviders from @earendil-works/pi-ai/providers/all), registerApiProvider/unregisterApiProviders/resetApiProviders/getApiProvider, getEnvApiKey/findEnvKeys, setBedrockProviderModule, the per-API lazy stream wrappers (anthropicMessagesApi, ...), and the image-generation API.
• Renamed the Provider type to ProviderId. Provider now names the runtime provider interface (id, name, auth, model listing, stream behavior).
• API implementation modules moved from src/providers/ to @earendil-works/pi-ai/api/*, renamed by API id (anthropic -> api/anthropic-messages, google -> api/google-generative-ai, mistral -> api/mistral-conversations, amazon-bedrock -> api/bedrock-converse-stream), each exporting exactly stream and streamSimple. The old per-impl export names (streamAnthropic, streamSimpleAnthropic, ...) and legacy raw API subpaths (./anthropic, ./google, ./openai-completions, ...) are gone; import raw API implementations through @earendil-works/pi-ai/api/*.

... (truncated)

Commits

• a23abe4 Release v0.80.3
• f98a154 docs: audit changelog entries
• 5c1a297 fix(ai): update generated model catalogue
• 3d6acb3 fix(ai): regenerate model catalog
• 6fbeba5 Merge pull request #5832 from stephanmck/fix/provider-error-body-passthrough-...
• b91bdd5 fix(ai): preserve Z.AI thinking content
• 5411373 fix(ai): use HTTP timeout for Codex SSE headers
• 09f1059 fix(ai): clamp streamSimple max tokens
• f78b163 fix(ai): revert minimax max token clamp
• 62fad94 fix(ai): surface provider HTTP error body instead of opaque SDK message
• Additional commits viewable in compare view

Updates @earendil-works/pi-coding-agent from 0.79.9 to 0.80.3

Release notes

Sourced from @​earendil-works/pi-coding-agent's releases.

v0.80.3

New Features

• Anthropic Claude Sonnet 5 support - Claude Sonnet 5 is available through inherited Anthropic-compatible and Bedrock provider catalogs with adaptive thinking enabled. See Providers and Model Options.
• Configurable output spacing - outputPad controls horizontal padding for user messages, assistant messages, and thinking blocks. See Settings.
• External editor configuration - externalEditor lets Ctrl+G use a configured editor before $VISUAL/$EDITOR fallbacks. See Settings and Keybindings.
• Richer RPC session tree access - RPC clients can inspect session entries and tree snapshots with get_entries and get_tree. See get_entries and get_tree.
• Extension session metadata updates - Extensions can observe session name changes through session_info_changed. See session_info_changed.
• Modern Azure Foundry endpoint support - Azure OpenAI Responses provider setup supports current Microsoft Foundry endpoint URLs. See Azure OpenAI.

Added

• Added inherited Anthropic Claude Sonnet 5 model support.
• Added get_entries and get_tree RPC commands for reading session entries and tree snapshots over RPC (#6078 by @​geraschenko).
• Added a package ./rpc-entry export for launching Pi directly in RPC mode.
• Added session-name change events for extensions (#6175 by @​xl0).
• Added inherited Azure OpenAI Responses support for modern Microsoft Foundry endpoint URLs (#6004 by @​gukoff).
• Added inherited Usage.reasoning token counts for providers that report reasoning/thinking token usage (#6057).
• Added an externalEditor settings.json override for Ctrl+G external editor commands, with default fallbacks to Notepad on Windows and nano elsewhere (#6122).
• Added an outputPad setting for user message, assistant message, and thinking horizontal padding (#6168).

Changed

• Changed the default OpenAI model to gpt-5.5.
• Changed inherited OpenAI Codex Responses SSE response-header waits to use the configured HTTP timeout instead of the previous fixed 20 second timeout, reducing false timeouts on slow connections (#4945).

Fixed

• Fixed inherited Claude Sonnet 5 metadata to use adaptive thinking payloads for Anthropic-compatible and Bedrock requests.
• Fixed inherited generated Xiaomi MiMo model pricing to match current pay-as-you-go pricing from models.dev (#6138).
• Fixed inherited provider HTTP errors to include response bodies instead of opaque SDK messages (#5832 by @​stephanmck).
• Fixed inherited streamSimple() max-token caps so providers that count input and output against one context window do not reject long requests (#5595).
• Fixed inherited OpenAI Responses streams to preserve reasoning replay state when output items finish out of order (#6009).
• Fixed inherited Z.AI preserved thinking requests to send thinking.clear_thinking: false when thinking is enabled, allowing replayed reasoning_content to participate in provider caching (#6083).
• Fixed pre-prompt compaction to stop after compaction instead of continuing immediately (#6074 by @​yzhg1983).
• Fixed resource notifications to stay before messages when resuming sessions (#6048 by @​haoqixu).
• Fixed startup benchmark timing output to print after TUI shutdown, preserve extension timings, and drain terminal-query replies before stopping benchmark mode (#6030 by @​xl0, #6063 by @​xl0).
• Fixed extension tool changes to apply before the next provider request in the same agent run without dropping before_agent_start system-prompt overrides (#6162).
• Fixed a crash when undici emits an internal client error while terminating a mid-stream HTTP response (#6133).
• Fixed the compaction event regression test to cover status indicator cleanup and keep CI passing.
• Fixed interactive status indicators so ending work, retry, compaction, or branch-summary indicators no longer shrink the TUI when clear-on-shrink is enabled (#6026).
• Fixed --session and SessionManager.open() to reject non-empty invalid session files without overwriting them (#6002).
• Fixed user-message transcript rendering to keep visible backslashes in Markdown escape sequences such as \" (#6105).
• Fixed assistant messages stopped by output length to show a visible incomplete-response error (#4290).
• Fixed --no-session --session-id so ephemeral CLI runs can use deterministic session IDs for provider cache affinity (#6070).
• Fixed disk BMP image files to be detected, converted to PNG, and attached through read and CLI @file inputs (#6047).
• Fixed auto-retry for provider stream errors that explicitly tell callers to retry the request (#6019).

v0.80.2

Changed

... (truncated)

Changelog

Sourced from @​earendil-works/pi-coding-agent's changelog.

[0.80.3] - 2026-06-30

New Features

• Anthropic Claude Sonnet 5 support - Claude Sonnet 5 is available through inherited Anthropic-compatible and Bedrock provider catalogs with adaptive thinking enabled. See Providers and Model Options.
• Configurable output spacing - outputPad controls horizontal padding for user messages, assistant messages, and thinking blocks. See Settings.
• External editor configuration - externalEditor lets Ctrl+G use a configured editor before $VISUAL/$EDITOR fallbacks. See Settings and Keybindings.
• Richer RPC session tree access - RPC clients can inspect session entries and tree snapshots with get_entries and get_tree. See get_entries and get_tree.
• Extension session metadata updates - Extensions can observe session name changes through session_info_changed. See session_info_changed.
• Modern Azure Foundry endpoint support - Azure OpenAI Responses provider setup supports current Microsoft Foundry endpoint URLs. See Azure OpenAI.

Added

• Added inherited Anthropic Claude Sonnet 5 model support.
• Added get_entries and get_tree RPC commands for reading session entries and tree snapshots over RPC (#6078 by @​geraschenko).
• Added a package ./rpc-entry export for launching Pi directly in RPC mode.
• Added session-name change events for extensions (#6175 by @​xl0).
• Added inherited Azure OpenAI Responses support for modern Microsoft Foundry endpoint URLs (#6004 by @​gukoff).
• Added inherited Usage.reasoning token counts for providers that report reasoning/thinking token usage (#6057).
• Added an externalEditor settings.json override for Ctrl+G external editor commands, with default fallbacks to Notepad on Windows and nano elsewhere (#6122).
• Added an outputPad setting for user message, assistant message, and thinking horizontal padding (#6168).

Changed

• Changed the default OpenAI model to gpt-5.5.
• Changed inherited OpenAI Codex Responses SSE response-header waits to use the configured HTTP timeout instead of the previous fixed 20 second timeout, reducing false timeouts on slow connections (#4945).

Fixed

• Fixed inherited Claude Sonnet 5 metadata to use adaptive thinking payloads for Anthropic-compatible and Bedrock requests.
• Fixed inherited generated Xiaomi MiMo model pricing to match current pay-as-you-go pricing from models.dev (#6138).
• Fixed inherited provider HTTP errors to include response bodies instead of opaque SDK messages (#5832 by @​stephanmck).
• Fixed inherited streamSimple() max-token caps so providers that count input and output against one context window do not reject long requests (#5595).
• Fixed inherited OpenAI Responses streams to preserve reasoning replay state when output items finish out of order (#6009).
• Fixed inherited Z.AI preserved thinking requests to send thinking.clear_thinking: false when thinking is enabled, allowing replayed reasoning_content to participate in provider caching (#6083).
• Fixed pre-prompt compaction to stop after compaction instead of continuing immediately (#6074 by @​yzhg1983).
• Fixed resource notifications to stay before messages when resuming sessions (#6048 by @​haoqixu).
• Fixed startup benchmark timing output to print after TUI shutdown, preserve extension timings, and drain terminal-query replies before stopping benchmark mode (#6030 by @​xl0, #6063 by @​xl0).
• Fixed extension tool changes to apply before the next provider request in the same agent run without dropping before_agent_start system-prompt overrides (#6162).
• Fixed a crash when undici emits an internal client error while terminating a mid-stream HTTP response (#6133).
• Fixed the compaction event regression test to cover status indicator cleanup and keep CI passing.
• Fixed interactive status indicators so ending work, retry, compaction, or branch-summary indicators no longer shrink the TUI when clear-on-shrink is enabled (#6026).
• Fixed --session and SessionManager.open() to reject non-empty invalid session files without overwriting them (#6002).
• Fixed user-message transcript rendering to keep visible backslashes in Markdown escape sequences such as \" (#6105).
• Fixed assistant messages stopped by output length to show a visible incomplete-response error (#4290).
• Fixed --no-session --session-id so ephemeral CLI runs can use deterministic session IDs for provider cache affinity (#6070).
• Fixed disk BMP image files to be detected, converted to PNG, and attached through read and CLI @file inputs (#6047).
• Fixed auto-retry for provider stream errors that explicitly tell callers to retry the request (#6019).

[0.80.2] - 2026-06-23

... (truncated)

Commits

• a23abe4 Release v0.80.3
• f98a154 docs: audit changelog entries
• fd6659d fix(coding-agent): preserve run prompt during tool refresh
• e547bb9 fix(coding-agent): refresh session state before next turn
• 9be55bc fix(coding-agent): apply output padding to user messages
• 6564d94 feat(coding-agent): add configurable assistant output padding
• 2117b61 fix(coding-agent): handle undici mid-stream client errors
• 726a9c5 fix(coding-agent): emit session name changes to extensions
• 927e980 fix(coding-agent): fix compaction event regression test
• 5d49927 fix(coding-agent): stabilize interactive status indicators
• Additional commits viewable in compare view

Updates @earendil-works/pi-tui from 0.79.9 to 0.80.3

Release notes

Sourced from @​earendil-works/pi-tui's releases.

v0.80.3

New Features

• Anthropic Claude Sonnet 5 support - Claude Sonnet 5 is available through inherited Anthropic-compatible and Bedrock provider catalogs with adaptive thinking enabled. See Providers and Model Options.
• Configurable output spacing - outputPad controls horizontal padding for user messages, assistant messages, and thinking blocks. See Settings.
• External editor configuration - externalEditor lets Ctrl+G use a configured editor before $VISUAL/$EDITOR fallbacks. See Settings and Keybindings.
• Richer RPC session tree access - RPC clients can inspect session entries and tree snapshots with get_entries and get_tree. See get_entries and get_tree.
• Extension session metadata updates - Extensions can observe session name changes through session_info_changed. See session_info_changed.
• Modern Azure Foundry endpoint support - Azure OpenAI Responses provider setup supports current Microsoft Foundry endpoint URLs. See Azure OpenAI.

Added

• Added inherited Anthropic Claude Sonnet 5 model support.
• Added get_entries and get_tree RPC commands for reading session entries and tree s...

  Description has been truncated