build(deps): bump the python-deps group across 1 directory with 2 updates

[dependabot]

Bumps the python-deps group with 2 updates in the /plugins/bakery/skills/bakery/evals directory: inspect-ai and anthropic.

Updates inspect-ai from 0.3.265 to 0.3.268

Changelog

Sourced from inspect-ai's changelog.

0.3.268 (22 September 2026)

• Agent Bridge: Bridged Anthropic requests no longer fail under anthropic SDK 1.8.0 or later, including requests with iterator, file-path or pydantic inputs.

0.3.267 (22 September 2026)

• Anthropic: Computer use now works on Claude Opus 5.5, Fable 5/5.1 and Mythos 5/5.1 via Anthropic's computer_toolset_20260801 toolset; pass -M computer_toolset=true to use the toolset on Claude Opus 4.8, Sonnet 5 and Opus 5.
• Computer Use: the key action now honors repeat.
• Checkpointing: A pre-existing entry planted at the in-sandbox work area (/root/.cache/inspect), or a parent /root/.cache that other users could modify, now fails checkpoint setup and restore with a clear error instead of being reused.
• Agent Bridge: Approval policies now match bridged host tools that a sandboxed agent calls through a single MCP dispatcher function (Antigravity's call_mcp_tool) by the tool's own name, and approvers see and modify the tool's own arguments.
• Sample and Task Sources: a new sample_abandoned() hook reports a sample cancelled before anything was logged (cancelled while queued, or before its retry_on_error re-run), so a source waiting on it no longer stalls.
• Faster inspect CLI startup and import inspect_ai; InputRequest and request_input now annotate schema by name only, so typing.get_type_hints and pydantic schema generation for InputRequest are unsupported.
• Bugfix: Approval policies in run-config files now load correctly instead of failing evaluation startup with an AttributeError.
• Added read_eval_log_sample_summaries_async and three other async log readers to the public inspect_ai.log exports.
• Meta: New meta provider for Muse Spark models on the Meta Model API, which streams by default, preserves model reasoning across turns, and reports policy-blocked prompts as content filter stops.
• OpenAI: Support for GPT-6 Sol (gpt-6-sol) and GPT-6 Luna (gpt-6-luna), including reasoning_effort="none", which these models accept (GPT-6 Astra does not).
• OpenAI: Responses API requests no longer fail validation when a compatible service omits model from its response.
• Agent Bridge: An unexpected exception from a host tool called through sandbox_agent_bridge(bridged_tools=...) now fails the sample as a native tool exception does, and malformed arguments are reported to the model as a parsing error.
• Grok: Support for Grok 4.7.
• Bugfix: bash_session no longer sends the literal string "None" to the shell when type_submit is called without input.
• Sandbox tools: A sandbox user can no longer make the tools read another file in place of a large (chunked) tool response.
• Sandbox tools: The text editor's directory view no longer interprets shell syntax in paths and runs find only from /usr/sbin:/usr/bin:/sbin:/bin, not the image PATH.
• Anthropic: Fixed model construction failing with a default_headers TypeError when ANTHROPIC_AUTH_TOKEN is set and the caller supplies its own default headers.
• Bugfix: Cancelling a sample while its sandbox files are being copied or its setup script is running no longer skips the sandbox provider's sample_cleanup(), which could leak sandboxes on certain providers.
• Grok: Calls to a client-side code_execution() tool (native execution disabled) are now executed instead of being silently dropped when xAI reports them as its built-in tool.
• Anthropic: Support for Claude Opus 5.5 (claude-opus-5-5): thinking can't be disabled, forced tool choice degrades to auto, and computer use is not yet supported on the Claude API and Vertex.

0.3.266 (19 September 2026)

• Bugfix: inspect score --scorer pkg/name now resolves @scanner functions from installed packages (e.g. inspect_petri/audit_judge) instead of failing with LookupError; unknown names now report the "scorer couldn't be loaded" guidance rather than a raw traceback.
• Control channel: the inspect ctl task score interim-metrics payload now reports each entry's originating scorer under scorer and the score's name under name (previously scorer held the score name). This lets consumers disambiguate scores from dict-valued scorers — where several scorers can share a score name — and reconstruct the EvalScore needed to resolve a task's headline metric. Breaking for clients that read the old scorer field as the score name.

Commits

• f9837f6 update changelog for release
• 52ebb8e Fix Anthropic agent bridge rejecting every request with anthropic SDK 1.8.0 (...
• fa2f7f4 update changelog for release
• 0f2474a Anthropic: computer use on Claude Opus 5.5, Fable 5/5.1 and Mythos 5/5.1 via ...
• 6945c44 Support for Claude Opus 5.5 (#5512)
• d10f6be Make PR validation show evidence beyond CI (#5520)
• a8b2ebb Design: record bridged host tool executions as ToolEvents (#5427)
• 31b1679 Support for OpenAI GPT-6 Sol and GPT-6 Luna (#5516)
• 02e86bd Lead PR compatibility with actions and summarize validation (#5518)
• 7704741 Improve PR descriptions and agent writing guidance (#5515)
• Additional commits viewable in compare view

Updates anthropic from 1.6.0 to 1.8.0

Release notes

Sourced from anthropic's releases.

v1.8.0

1.8.0 (2026-09-22)

Full Changelog: v1.7.0...v1.8.0

Features

• api: add support for claude-opus-5-5, inline tool definitions and MCP tool-list pinning (beta) (b5cc700)

Bug Fixes

• api: share one evaluated_permission enum across Managed Agents events (f4f51c8)
• streaming: avoid a Python 3.13 crash at exit when a stream is left open (af029bd)
• tools: add_tools() takes effect straight away (#874) (06d0a6b)
• tools: leave reply-only params off the tool runner's compaction request (#871) (4687310)

Chores

• docs: add descriptions to the Dreams API reference (5574c63)
• docs: add descriptions to the User Profiles API reference (5574c63)
• docs: add memory store descriptions to the Managed Agents API reference (5574c63)
• docs: improve descriptions in the Dreams API reference (1f5e477)
• docs: simplify the session thread agent type description (a45b7ec)
• docs: update diagnostics field descriptions on beta messages (367750c)
• internal: fix tests (70c8cbd)
• internal: remove mypy (be01ed6)
• tests: update fixtures (77f6df8)

Documentation

• add path and header parameter descriptions (f429a30)

Refactors

• client: remove the request param transform in favour of the JSON encoder (539422c)

v1.7.0

1.7.0 (2026-09-18)

Full Changelog: v1.6.0...v1.7.0

Features

• api: add group with display_name to rate limits, deprecate group_type (28f0a83)
• tools: add compact_before_next_turn() to the tool runner (#641) (8b23fb3)

... (truncated)

Changelog

Sourced from anthropic's changelog.

1.8.0 (2026-09-22)

Full Changelog: v1.7.0...v1.8.0

Features

• api: add support for claude-opus-5-5, inline tool definitions and MCP tool-list pinning (beta) (b5cc700)

Bug Fixes

• api: share one evaluated_permission enum across Managed Agents events (f4f51c8)
• streaming: avoid a Python 3.13 crash at exit when a stream is left open (af029bd)
• tools: add_tools() takes effect straight away (#874) (06d0a6b)
• tools: leave reply-only params off the tool runner's compaction request (#871) (4687310)

Chores

• docs: add descriptions to the Dreams API reference (5574c63)
• docs: add descriptions to the User Profiles API reference (5574c63)
• docs: add memory store descriptions to the Managed Agents API reference (5574c63)
• docs: improve descriptions in the Dreams API reference (1f5e477)
• docs: simplify the session thread agent type description (a45b7ec)
• docs: update diagnostics field descriptions on beta messages (367750c)
• internal: fix tests (70c8cbd)
• internal: remove mypy (be01ed6)
• tests: update fixtures (77f6df8)

Documentation

• add path and header parameter descriptions (f429a30)

Refactors

• client: remove the request param transform in favour of the JSON encoder (539422c)

1.7.0 (2026-09-18)

Full Changelog: v1.6.0...v1.7.0

Features

• api: add group with display_name to rate limits, deprecate group_type (28f0a83)
• tools: add compact_before_next_turn() to the tool runner (#641) (8b23fb3)

Bug Fixes

... (truncated)

Commits

• 4421d56 Merge pull request #1946 from anthropics/release-please--branches--main--chan...
• d0247c2 release: 1.8.0
• be01ed6 chore(internal): remove mypy
• 06d0a6b fix(tools): add_tools() takes effect straight away (#874)
• af029bd fix(streaming): avoid a Python 3.13 crash at exit when a stream is left open
• 4687310 fix(tools): leave reply-only params off the tool runner's compaction request ...
• b5cc700 feat(api): add support for claude-opus-5-5, inline tool definitions and MCP t...
• 70c8cbd chore(internal): fix tests
• 539422c refactor(client): remove the request param transform in favour of the JSON en...
• 10a4c07 codegen metadata
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore <dependency name> major version will close this group update PR and stop Dependabot creating any more for the specific dependency's major version (unless you unignore this specific dependency's major version or upgrade to it yourself)
• @dependabot ignore <dependency name> minor version will close this group update PR and stop Dependabot creating any more for the specific dependency's minor version (unless you unignore this specific dependency's minor version or upgrade to it yourself)
• @dependabot ignore <dependency name> will close this group update PR and stop Dependabot creating any more for the specific dependency (unless you unignore this specific dependency or upgrade to it yourself)
• @dependabot unignore <dependency name> will remove all of the ignore conditions of the specified dependency
• @dependabot unignore <dependency name> <ignore condition> will remove the ignore condition of the specified dependency and ignore conditions