## What changed
Add optional `request_permissions` and `auto_review_suffix` fields to catalog approval messages. Use them to customize the permission tool heading and introduce automatic review guidance while preserving the bundled instruction bodies and existing policy gates. Accept `request_permissions_tool` as a deserialization alias.
Limit headings to 500 UTF-8 bytes on a single line and fall back to bundled text for missing or invalid values. Complete catalog approval messages retain precedence.
## Testing
Add coverage for heading validation, serialization aliases, policy gates, fallback behavior, and complete-message precedence. Integration tests verify that model switches append the selected model's headings without rewriting history or repeating unchanged instructions.
GitOrigin-RevId: 3311591a464cc086783fe25271a352dd5a229790
## What changed
- Add `ConnectorRuntimeManager::without_shared_snapshot()` to keep connector catalogs local to each client and disable disk caching.
- Use the thread's own Apps discovery when snapshot sharing is disabled, preserving its catalog across turns.
- Expose `apps_tools_from_mcp_client()` to read a ready, unshared Apps client's tools without starting or waiting for discovery.
## Testing
Extend coverage for shared-cache eligibility and verify that an unshared thread waits for its own Apps discovery and uses its own tools on successive model requests, even when a peer finishes discovery first.
GitOrigin-RevId: d57ddacaede047e67c09d378c86c1aa3577ca56f
## Why
Memory consolidation inherits the session's service tier. Give background consolidation its own setting, defaulting to Standard.
## What changed
Add `memories.consolidation_service_tier` with a default of `"default"` and apply it to the consolidation agent. Normalize recognized tier aliases, including `"fast"` to `"priority"`, and expose the setting in the configuration schema.
## Testing
Add coverage for configuration parsing and alias normalization, Standard consolidation despite an Ultrafast session tier, and an explicit `"flex"` consolidation override.
GitOrigin-RevId: 58031ad56196152d212531d56e7fe9813f83317e
## What changed
Allow `model_catalog_url` under `[model_providers.openai]` while preserving the built-in provider's identity, authentication, and inference routing through `openai_base_url`. Reject other non-default overrides and require an HTTPS catalog URL on an allowed ChatGPT host, without embedded credentials.
Apply the same catalog override merging when checking managed provider requirements for existing threads, preserving each thread's inference endpoint even if user defaults change.
Document the required Codex catalog format, credential forwarding, and existing discovery and access restrictions.
## Testing
Add coverage for allowed and rejected catalog destinations, preservation of built-in provider settings, and managed requirement changes. Add integration coverage for API key and ChatGPT authentication that verifies separate catalog and inference routing.
GitOrigin-RevId: 4d91ada518ebd2a09cbf6ded0ddf4b11c92b77dd
## Why
HTTP response streams retained copies of completed output items even when inference tracing was disabled, keeping their payloads alive until the response completed.
## What changed
Collect completed output items only when inference tracing or WebSocket last-response bookkeeping needs them. Make the last-response sender optional and create its channel only for WebSocket requests.
## Testing
Add a regression test that verifies an untraced HTTP stream releases a completed item's shared payload after delivery, while the response remains open. Update existing tracing and WebSocket bookkeeping tests for the optional sender.
GitOrigin-RevId: 7695534c6d00f3002005324c43c23bf733f01804
## What changed
Add `CodexThread::replace_dynamic_tools` to replace the tool catalog without starting a turn. Updates apply to subsequent model steps and future turns while preserving in-flight calls; an empty list clears the catalog.
Require host opt-in through `allow_persistent_dynamic_tool_updates` for persistent threads. Persist replacements before publishing them, restore the latest catalog on resume, and preserve replacements and explicit clears through compaction. Keep explicit resume overrides separate from the persisted catalog.
## Testing
Add coverage for pending calls during replacement, replacement and clearing across restart, compaction recovery, resume overrides, submission ordering after caller cancellation, host opt-in, and persistence failures.
GitOrigin-RevId: fd2634906895a9b72ae7892ddd9435dced45eccb
## What changed
Expose the backend-provided `generationId` on image-generation items and preserve it through legacy completion events, thread history reconstruction, and rollout migration. Default missing IDs to `None` for older history.
Update the generated JSON schemas, TypeScript types, and Python models to include the nullable field.
## Testing
Extend coverage for ID serialization round trips, completed-item notifications, persisted and resumed history, and legacy rollout migration.
GitOrigin-RevId: a7f06faa7ac6833417b257a535d2901c03b79135
## What changed
Expose `literature_search_query` in standalone `web.run` when the active model lists it in `experimental_supported_tools`. Support query text, domain filters, and recency filters, and reject literature search calls for models without the capability.
Pass the active model information to extensions for each step and report literature queries as search actions in web search events.
## Testing
Add coverage for capability rejection, model-dependent schema exposure, app-server round trips with a turn-level model override, and a scenario that searches for a paper and opens the returned reference.
GitOrigin-RevId: 87ef1fadd80ca7bfc2a7642c4e443f9b9f2e0e00
## What changed
Add optional `TurnInputRequest.user_item_uuid` and carry it through turn starts and steering to the recorded user message. When supplied, use `msg_input_<uuid>` as the Responses item ID; inputs without a UUID retain generated IDs. Raw response items ignore this field.
Add `ResponseItemId` helpers to construct and recognize the reserved `msg_input_` namespace.
## Testing
Add regression coverage for steered inputs with identical text and client IDs, checking that each supplied UUID survives recording and UI item IDs remain distinct. Add a scenario test verifying that the assigned ID reaches Responses and persists in subsequent request history alongside an identical ordinary prompt.
GitOrigin-RevId: 440e306bfdaa7facb69b21930c5244a83db2fdae
## What changed
Keep original scheduled messages and heartbeat provenance during synchronous transcript collection. Add `prepare_sync_context` to compress repeated heartbeat instructions across the complete collected transcript before selecting the requested suffix, preserving reference numbering in full and delta prompts.
Route synchronous reviewer prompts through this preparation step and reject composition of unprepared scheduled messages. Move transcript profiles and REPL evidence rendering into the preparation module. Asynchronous heartbeat compression remains in collection.
## Testing
Extend heartbeat transcript coverage to compare prepared full and delta prompts with the existing rendering and verify that direct composition rejects unprepared scheduled messages.
GitOrigin-RevId: c9824b8862021301a1f537a74f83bc7eb20aa0ad
## What changed
Move planned-action construction from the core Guardian prompt builder into the exported `codex_guardian_reviewer::build_guardian_planned_action` helper. Keep the existing action kinds, MCP tool and connector descriptions, and reason selection unchanged.
GitOrigin-RevId: 119323523ed2e199c998c1139fb8e6255bc730d9
## What changed
Move executor path-convention selection from the approval caller into `ReviewAction::from_approval_action`, using the supplied `GuardianReviewContext`. Preserve Windows and POSIX rendering based on the executor OS, with path inference for legacy executors that do not report an OS.
## Testing
Update the command approval test to exercise `ReviewAction` with Windows, Linux, and legacy executor contexts, checking both approval JSON and assessment working directories.
GitOrigin-RevId: b13fad1d4033976cb94b899b62a3779c97e171fa
## What changed
Move approval-request assessment and analytics mappings, target item lookup, and turn ID selection into `codex-guardian-reviewer`. Export the helpers from the crate and re-export them in core, preserving existing behavior.
## Testing
Move the apply-patch serialization test into the reviewer crate. It checks that Windows paths are preserved and patch text is omitted from the assessment action.
GitOrigin-RevId: 3a011db0bc51c653bb0877c4b0cb52eb61c63244
## What changed
Add the disabled-by-default `omit_skill_descriptions` feature flag to omit descriptions from model-visible host, executor, and cloud skill catalogs while retaining names and locators. Adapt catalog introductions, usage instructions, and alias budget accounting to the selected policy.
Persist the description policy in catalog snapshots and refresh usage instructions when it changes. Ignore description-only edits in catalog deltas while descriptions are omitted.
## Testing
Add coverage for alias token budgets, policy changes with older snapshots, and suppressed description-only updates. Extend scenario snapshots to cover plugin visibility and catalog deduplication with descriptions omitted.
GitOrigin-RevId: 5bf1dcee8c04059c7a78998affc1efcdef8ec883
## What changed
- Retain captured root conversation sources until synchronous preparation applies request limits. Keep asynchronous snapshots limited to the bounded messages.
- Move approval-reason truncation and tool-description escaping and truncation from composition into preparation.
- Extract synchronous image preparation and admission into `guardian-context`, preserving the prompt token cap and model context-window checks.
GitOrigin-RevId: d3a864c77d8f16bd3e77df436d80c4d2738c43cc
## What changed
Move approval request serialization and model-visible action formatting from `codex-core` into `codex-guardian-reviewer`, expose both helpers publicly, and re-export them from core. Preserve the existing JSON shapes and formatting.
## Testing
Move the MCP tool-call and network-access trigger serialization tests into the reviewer crate, retaining coverage for omitted optional fields and Windows path rendering.
GitOrigin-RevId: c0093c023b2f4c1be16bac481f1173a5da6ac25c
## What changed
Export `GuardianPromptItems` and `compose_guardian_prompt` from `codex-guardian-reviewer`, and delegate prompt composition from core to the new helper. Core continues to collect context sections and choose the history version.
Preserve full and delta transcript selection, empty-transcript placeholders, cursor accounting, and Node REPL evidence sequencing.
GitOrigin-RevId: e3e2b2a87a42709ddb5441f47cbb7461b0130579
## What changed
Export `GuardianApprovalRequest`, `GuardianNetworkAccessTrigger`, and
`GuardianMcpAnnotations` from `codex-guardian-reviewer` and re-export them in
core. Move environment lookup and scope classification with the request type,
preserving existing classification and credential-safe debug output.
GitOrigin-RevId: 5253a5124d5e370a1230b46aa7dc48b24b1c902e
## What changed
Move Guardian session-source detection, the reviewer name constant, and parent-thread prompt cache key generation into `codex-guardian-reviewer`. Re-export them from core, preserving existing behavior.
## Testing
Move the existing prompt cache key test into the reviewer crate, retaining coverage for parent-thread scoping, key length, and missing or inapplicable inputs.
GitOrigin-RevId: c14b3c030ba93676a0799aba56b23a676dbbc56d
## What changed
Extend the incremental Guardian review scenario with historical encrypted `send_message` calls in the default and custom tool namespaces. Add snapshots and assertions that both reviews retain each call and parent reply exactly once, in order, preserving call IDs, namespaces, and encryption metadata with a paired evidence-only output.
Verify that successive reviews reuse the same thread and append to the previous input. The scenarios use mocked ciphertext and decisions to check transport behavior, not backend decryption or authorization outcomes.
GitOrigin-RevId: a0927531b6cdae496fefa2a7f35032f7413807b6
## What changed
Add mock guardian approvals for setup handoffs in `handoff_scenario` when native multi-agent history is enabled. Match only the trailing `function_call_output` so historical handoffs cannot consume the final command's separate denial response.
Require five approval requests with native history enabled and none with it disabled.
GitOrigin-RevId: 7b14c97aa85adc63c4d3bf95014c44831eae1c85
## What changed
Make tool descriptions precede namespace descriptions in Code Mode execution and tool search. Mark `code_mode_tool_description_first` as removed and simplify cached definitions to depend only on the input schema budget.
## Testing
Update Code Mode integration assertions and catalog snapshots to expect tool descriptions before namespace descriptions.
GitOrigin-RevId: 1344d4c264d1d2b18ee0388a064b6813ca81f770
## What changed
Export `token_usage_delta` from `codex-guardian-reviewer` and use it in the core review session. Preserve the existing calculation, including clamping negative deltas to zero.
## Testing
Move the existing nonnegative token usage delta test alongside the helper in the reviewer crate.
GitOrigin-RevId: fcfce24b8d50a63dcd1122a4204939bc6750feda
## What changed
Add `GuardianScope::for_multi_agent_tool` and use it for both tool review requirements and asynchronous scoring. Classify `spawn_agent`, `send_message`, `followup_task`, and `interrupt_agent` as native tools when they match the configured namespace, accepting default namespace spellings when none is configured. Keep review enablement checks at each call site.
GitOrigin-RevId: 9180ccdc629532f6c171d899dc5881e15b447141
## What changed
- Recognize `policy_capacity_unavailable` in HTTP and streaming API errors and expose it as `policyCapacityUnavailable` through the app-server protocol and generated bindings.
- Display “Policy-restricted capacity is unavailable. Please try again later.” while preserving overload retry behavior based on `Retry-After`.
- Handle the new error in TUI overload rendering, reviewer retries, and error telemetry.
## Testing
Add coverage for HTTP and streaming error classification, retry advice, WebSocket terminal error attribution, protocol serialization, and TUI rendering with and without an upstream message.
GitOrigin-RevId: 1b9536fba1e11cc891dcf689aee35410d82f5412
## What changed
When `GuardianNativeMultiAgentHistory` is enabled, require Guardian review for wrapped Multi-Agent V2 calls to `spawn_agent`, `send_message`, `followup_task`, and `interrupt_agent`.
Include these calls in Guardian's asynchronous native-tool scoring scope, honoring the configured multi-agent namespace and preserving that configuration across overrides.
## Testing
Extend the incremental-review scenario with an encrypted, namespaced `send_message` call. Verify that the encrypted parent reply remains present exactly once in both Guardian reviews and the subsequent agent request, and update the snapshot to cover native tool-call review evidence.
GitOrigin-RevId: 2af9561b0d69af4a8f878fa32365ce845368e02d
## Why
Native multi-agent calls must stay intact for backend decryption during Guardian review.
## What changed
When `guardian_native_multi_agent_history` is enabled for multi-agent v2, preserve `spawn_agent`, `send_message`, `followup_task`, and `interrupt_agent` function calls in the configured tool namespace as native transcript items. Use `functions` when no namespace is configured.
Include the paired evidence-only output in token accounting, and omit native evidence whole when limits require it.
## Testing
Extend transcript tests to cover native call ordering, paired-output budgeting, and whole-item omission. Add a native-history snapshot alongside the existing text-history delegation handoff scenario.
GitOrigin-RevId: e7d9b4f5ea772890bf14b79fdb53b0f9d0a842e4
## What changed
Register `guardian_native_multi_agent_history` as an under-development feature,
disabled by default, and expose it in the configuration schema.
GitOrigin-RevId: d434750c1aa0cc78b36d1d661a4df0def5426f02
## Why
Custom OpenAI base URLs caused Codex backend authentication to lose Responses headers and Guardian metadata when the URL did not match a recognized backend route.
## What changed
Apply model-scoped Responses headers, Guardian reviewer and classifier markers, and eligible ChatGPT parent credit requests based on the OpenAI provider and authentication configuration, regardless of the base URL. Keep API-key requests and other providers excluded from this behavior.
## Testing
Extend coverage for custom `/v1` URLs and direct `/api/codex` endpoints with ChatGPT, header-based, and API-key authentication, including HTTP fallback and WebSocket requests.
GitOrigin-RevId: cc4bb5b4a1a6596904ecf8965237c7a4f09e112b
## What changed
Pass `PreparedExecCommand` through the process manager and runtime instead of copying command, shell, environment, and permission fields into separate request structs. Move approval requirement evaluation and approval action construction onto the prepared command, and use it for shell snapshots and launch failure events.
GitOrigin-RevId: 3e9f87e0974218fea6fffc0f7740616d696b9800
## What changed
Move argument parsing, working-directory and shell resolution, permission validation, and command lifetime setup into `unified_exec::preparation` with crate-visible parsed and prepared command types.
Update `ExecCommandHandler` to use the extracted preparation flow, preserving implicit skill event emission between parsing and preparation.
GitOrigin-RevId: 79e9be5d87c1b0ab905afdf1e343ac141fb7d4c6
## What changed
- Add `CoreToolRuntime::requires_guardian_review` to review opted-in function calls before dispatch and stop execution when review rejects the call.
- Deliver proposed calls as native review evidence, preserving encrypted argument metadata. When hooks rewrite arguments, review the updated arguments and discard the original encrypted replay content.
- Deduplicate native call evidence against reviewer history and include paired evidence outputs in input budgets.
- Keep review accounting and metrics for native calls without emitting action-specific client assessment events, and omit approval payloads from debug output.
## Testing
Extend regression coverage for rewritten native calls across review retries, denied calls that never execute their handlers, oversized native evidence, and review metrics without client events.
GitOrigin-RevId: c3b22c0220344581c2b38f233001abaed49cba34
## Why
Verified answers omitted because of request-size limits must be marked as missing. Legacy checkpoint messages also need to match retained excerpts after truncation so workers receive only one copy.
## What changed
Move root message preparation into `codex-guardian-context::prepare_root_context`, including handoff selection, shared message limits, legacy matching, and incomplete-context markers. Pass original legacy text into preparation and mark verified answers incomplete when oversized records are excluded.
## Testing
Add a handoff regression scenario for an ID-less checkpoint with a retained excerpt and an oversized verified answer. Assert that the worker receives one capped user message, both incomplete-evidence markers, and an incomplete authorization version.
GitOrigin-RevId: bd3f5fe164ae2fddba440e9269fd63b52e8b12f8
## What changed
- Replace Rust `1.99.1` pins with `1.95.0` across Cargo, Bazel, CI, release workflows, and Windows setup, and update the Bazel lockfile.
- Adjust Rust APIs, borrowing, and lint-related code for the pinned toolchain, and box MCP tool dispatch to bound compiler async-layout query depth.
- Update the halfwidth sound mark snapshot and allow optional symbol hash suffixes in the code mode host package smoke test.
GitOrigin-RevId: 75460c3d0465dc601811e619bce373a4f7ff2912
## What changed
Add optional `approvals.granular` catalog fields for the policy introduction and prompted and rejected category lists. Preserve template strings verbatim in `GranularApprovalMessages`, leaving validation and rendering to consumers.
Document the 500-byte template limit, required `{{ policy }}` and `{{ categories }}` markers, and fallback rules.
## Testing
Extend serialization tests to cover null, empty, and populated granular templates, and verify their preservation through model responses and base instruction overrides.
GitOrigin-RevId: ce32e41a9b37e2e7c45ac9b079708a640c99326c
## What changed
Add the `features.simplified_selection_hints` flag, disabled by default. When enabled, replace detailed plugin capability instructions with a concise developer message identifying the selected plugin and directing the model to use its available skills and tools.
Continue omitting hints for plugins without available capabilities and bounding hint length. Preserve the existing detailed instructions when the flag is disabled.
## Testing
Cover plugin refresh in an existing thread with the flag enabled and disabled. Verify that explicit skill prompts precede the simplified plugin hint and that the hint uses the developer role.
GitOrigin-RevId: 64fc3347bd49fffe675f343a4733ced1b4e5e88c
## Why
Rust formatting should include source files that Cargo cannot discover, while keeping command lines within Windows limits.
## What changed
- Update `scripts/format.py` to discover tracked and non-ignored untracked Rust files with Git and invoke `rustfmt` in batches of 64 files. Use the workspace configuration and skip recursive child-module formatting.
- Pass `SystemDrive` and `LOCALAPPDATA` to Windows Bazel tests.
- Extend the spelling allowlist, remove a manifest feature exception, and clean up an unused import, an empty MCP scenario module, and its snapshot.
GitOrigin-RevId: c0d625fda9290ead6a63919325d65f7352abfa40
## What changed
Allow omitting `gateway_oauth.delivery` to send the gateway token as `Authorization: Bearer …`. Custom headers and cookies remain available when a provider also needs its own credentials.
Reject conflicting provider authentication and duplicate delivery headers with actionable errors. For gateway bearer delivery, require `model_catalog_url` to share the scheme, host, and port of `base_url`. Consistently reject request header collisions across auth application methods.
## Testing
Add coverage for omitted delivery, provider credential conflicts, and model catalog origin validation. Update inference tests to verify gateway-only authentication while signed in and combined ChatGPT and gateway credentials in separate headers.
GitOrigin-RevId: 8564d3998511b3f6c55f3b65a01ed9056a90417e
## What changed
Collect original retained instructions, assistant context, and verified answers before applying synchronous request limits in `guardian-context` preparation. Carry original record sizes through collection, preserving whole-record omission notices, source ordering, and host assistant-context framing.
Apply retained transcript source size checks during synchronous transcript preparation as well.
## Testing
Extend regression assertions for synchronous/asynchronous retained-context parity and equivalent verified-answer omissions when oversized records are filtered during preparation.
GitOrigin-RevId: 9042ee3fe60ebd455f6656b4008efb2acc44ac3d
## What changed
- Move plugin install and uninstall errors and telemetry classification helpers into `codex-rs/core-plugins/src/manager_errors.rs`, preserving the existing public re-exports and error classifications.
- Box the realtime text routing future to bound the handoff future's size while retaining the admission permit across the await.
GitOrigin-RevId: ccf0c3bb93a8c69a0a618b96cd1525da87003176
## What changed
- Add `gpt-6.1-terra` to the bundled catalog and GCP Vertex model picker.
- Add `openai.gpt-6.1-terra` to Amazon Bedrock catalogs, including Global and US cross-region runtime variants.
- Use `gpt-6.1-terra` as GCP's preferred memory consolidation model.
## Testing
Extend catalog, model picker, thread startup, and Vertex request tests to cover GPT-6.1 Terra.
GitOrigin-RevId: edb3be3c4286c191349cdee742005ed3c67029e8
## What changed
Keep inherited incremental tool declarations in MCP-disabled prediction forks and `thread/sample` requests, and append a developer notice that MCP tools are unavailable. Skip tool declaration updates within the inherited context window; new windows still declare the current tools.
Add the default-enabled `prediction_preserve_parent_tools` feature flag and its configuration schema entries.
## Testing
Update core snapshots and app-server tests to cover inherited declarations, the MCP-unavailable notice, preservation across MCP refresh, and continued parent tool availability.
GitOrigin-RevId: 3adf96c7d0e0093986dbdbd540bd47674a2c105c
## What changed
Accept `[network_proxy]` in `requirements.toml`, retaining `[experimental_network]` as a compatibility alias. Reject both names in a single layer and normalize them before merging across layers so precedence and source tracking are preserved.
Honor the new spelling during credential broker project discovery, display `network_proxy` in debug configuration, and update diagnostics and documentation. Document the distinction between managed requirements, proxy feature configuration, and permission profiles. Policy publishers must keep using `[experimental_network]` until all clients and policy validation layers support the new spelling.
## Testing
Add coverage for alias equivalence, duplicate rejection, cross-layer precedence and sources, and broker project discovery. Exercise both spellings in proxy configuration, network approval routing, app-server requirements responses, and Windows machine policy validation.
GitOrigin-RevId: 5c5bb82bd4b6b1431aa97f27248775820eae6302
## Why
MCP completion events can truncate large results and omit content removed for model consumption. Rollout traces need the original response, including content and metadata, as well as results for calls denied before reaching the backend.
## What changed
- Record MCP runtime results before callbacks, content filtering, and UI truncation, and record skipped calls as failures.
- Stop deriving MCP runtime completion records from protocol events to avoid duplicate records.
- Treat MCP runtime payloads as opaque results when reducing agent interactions.
- Buffer JSON trace writes and explicitly flush them.
## Testing
Extend app-server tests to verify that traces retain oversized MCP text, audio, structured content, and metadata while UI results remain truncated, and that denied calls produce exactly one failed runtime result with the denial reason.
GitOrigin-RevId: 57880770545eb5869420d7dc6fa4a15e481cae9d
## What changed
When `features.untrusted_as_tool_output` is enabled, emit untrusted additional context as `function_call_output` items in the `agent` namespace before user input. Use the explicit context name when supplied; otherwise derive a bounded name from the source key, falling back to `untrusted_content`.
Deduplicate these outputs by name and truncated value so name changes are emitted while changes confined to truncated content are ignored. Application context remains in developer messages, and disabling the feature preserves the existing message format.
## Testing
Extend core and app-server tests and add request-history snapshots covering both formats, turn start and steering, derived names, truncation, and deduplication.
GitOrigin-RevId: cd5e2f4b3121714acb7f4baf5607f3c27d408cae
## What changed
- Separate native child spawn preparation from submission and move forked-history loading and filtering into `agent::history`.
- Group resolved session startup dependencies in `ThreadStartupContext` and extract helpers for effective spawn configuration and inherited environment requests.
- Share loaded-recipient message submission and capture only the content needed for communication logging, avoiding a clone of the full message.
GitOrigin-RevId: 4417ba6ca35e9f293064600f0a54932de514636f
## What changed
Add `WorldStateSectionContribution::with_window_prefix()` to render extension developer fragments as separate messages before retained history when building a context window, including after compaction. Updates within an existing window still append to history. Ignore fragments with other roles for sections that opt in.
## Testing
Add a unit test for prefix placement and duplicate avoidance, plus local and remote compaction scenarios covering live updates, persisted replacement history, and rejection of non-developer fragments.
GitOrigin-RevId: 8621457c6eec17f1585ca8e6fa154b1bf3257330
## What changed
Collect REPL evidence as shared immutable response records, then select and render it during synchronous context preparation. Share response selection logic between text and multimodal layouts while preserving rendering limits, omission markers, and image order.
Use the shared `NodeReplResponse` type in core evidence storage and preserve existing snapshots through copy-on-write image eviction.
## Testing
Extend tests to verify shared response identity during collection, unchanged rendering after preparation, and snapshot stability when storage pressure evicts images.
GitOrigin-RevId: 8564d3732faa5adf01a66bca523d4c03cfbe1822
## What changed
Add session options for client- or server-owned output truncation and optional `max_output_tokens` budgets for wait and terminate observations. Introduce `TerminateRequest` and update session implementations and callers to use it.
Extend the stdio and gRPC schemas with optional truncation settings and budgets, and define the `bounded-output-v1` capability. Existing sessions retain client-side truncation; providers reject server mode by default. This change defines the contracts without enabling server-side truncation.
## Testing
Extend protocol serialization coverage for `outputTruncation: "server"` while preserving existing wire shapes when optional fields are absent.
GitOrigin-RevId: 22145897ea16d385c5ba9e7549109f7a669fceb4
goodman-oai