85 Commits

• [codex] Split tool handlers by tool name (#20687)

  ## Why
  
  Tool registration used to bind a tool name to a handler externally,
  which left ownership split between the registry plan and the handler
  implementation. Some built-in handlers also multiplexed multiple in-core
  tools by switching on the invoked tool name internally.
  
  This moves the registry identity onto the handler itself and makes
  built-in multi-tool areas use separate concrete handlers, so each
  registered handler instance owns exactly one tool name and one dispatch
  path.
  
  ## What Changed
  
  - Added `ToolHandler::tool_name()` and changed
  `ToolRegistryBuilder::register_handler` to derive the registry key from
  the handler.
  - Split built-in multiplexed handlers into concrete per-tool handlers
  for unified exec, shell/local shell/container exec, MCP resources, goal
  tools, and agent job tools.
  - Kept name-carrying handler instances only where the runtime target is
  inherently external or dynamic, such as MCP tools, dynamic tools, and
  unavailable placeholders.
  - Updated `ToolHandlerKind` and registry-plan construction so plan
  entries map directly to concrete handler registrations.
  
  ## Verification
  
  - `cargo test -p codex-tools tool_registry_plan`
  - `cargo test -p codex-core --lib tools::registry_tests`
  - `just fix -p codex-tools`
  - `just fix -p codex-core`

  pakrym-oai · 2026-05-05 13:46:45 -07:00
  f593323ef1
• Route process tools to selected environments (#20647)

  ## Why
  When a turn exposes multiple selected environments, shell-style tools
  need a model-facing way to identify the intended target environment and
  handlers need to resolve that target before parsing cwd-relative
  permission fields or launching processes.
  
  This PR scopes that rollout to process tools. Filesystem-oriented tools
  such as `apply_patch`, `view_image`, and `list_dir` are intentionally
  left for follow-up slices.
  
  ## What Changed
  - Adds an `include_environment_id` option to shell-style tool schema
  builders.
  - Exposes optional `environment_id` on `shell`, `shell_command`, and
  `exec_command` only when `ToolEnvironmentMode::Multiple` is active.
  - Adds a shared handler helper that parses `environment_id` and
  `workdir` from JSON function-call arguments and returns the selected
  `Environment` plus effective absolute cwd.
  - Uses that helper in `shell`, `shell_command`, and `exec_command`
  handling so process execution uses the selected environment filesystem
  and cwd.
  - Changes `ExecCommandRequest` to carry a required resolved `cwd`,
  removing the process-manager fallback to the primary turn cwd for new
  exec commands.
  - Leaves `write_stdin` unchanged because it targets an existing process
  id, not a new environment.
  
  ## Testing
  - Added unit coverage for process-tool schema exposure, selected
  environment resolution, primary fallback, no-environment handling,
  unknown environment ids, and resolving cwd-relative permission paths
  against the selected environment cwd.
  - Added a remote-suite e2e coverage case for `exec_command` routing
  across explicit zero environments, one local environment, and
  local+remote environments.
  - Ran `just fmt` and `git diff --check`.
  
  ---------
  
  Co-authored-by: Codex <noreply@openai.com>

  starr-openai · 2026-05-05 12:12:03 -07:00
  78421face0
• tools: remove unused experimental list_dir tool (#21170)

  ## Why
  `list_dir` still carries a full spec/handler/test path, but nothing in
  the current model catalog advertises it via
  `experimental_supported_tools`. That leaves us maintaining an
  environment-backed tool surface that is effectively unused.
  
  ## What changed
  - delete the `list_dir` handler and its tests from `codex-core`
  - remove the `list_dir` spec builder, handler kind, and registry wiring
  from `codex-tools`
  - clean up the remaining internal README and registry tests so they no
  longer mention the removed tool

  jif-oai · 2026-05-05 13:11:07 +02:00
  70807730f5
• 1- Add model service tiers metadata (#20969)

  ## Why
  
  The model list needs to carry display-ready service tier metadata so
  clients can render tier choices with stable IDs, names, and
  descriptions. A raw speed-tier string list is not enough for richer UI
  copy or future tier labels.
  
  ## What changed
  
  - Added `ModelServiceTier` to shared model metadata with string `id`,
  `name`, and `description` fields.
  - Added `service_tiers` to `ModelInfo` and `ModelPreset`, preserving
  empty defaults for older cached model payloads.
  - Exposed `serviceTiers` on app-server v2 `Model` responses and threaded
  it through TUI app-server model conversion.
  - Marked legacy `additional_speed_tiers` / `additionalSpeedTiers`
  metadata as deprecated in source and generated schema output.
  - Regenerated app-server protocol JSON schema and TypeScript fixtures,
  including `ModelServiceTier.ts`.
  
  ## Verification
  
  - Ran `just write-app-server-schema`.
  - Did not run local tests per repo instruction; relying on PR CI.
  
  ---------
  
  Co-authored-by: Codex <noreply@openai.com>

  Ahmed Ibrahim · 2026-05-05 09:51:18 +03:00
  9d579813bb
• Use MCP server instructions in deferred namespace descriptions (#21053)

  ## Why
  
  MCP servers can provide `instructions` that explain what their tools are
  for. Directly exposed MCP namespaces already use those instructions when
  a connector description is not available, but deferred `tool_search`
  results did not preserve that fallback. The direct path falls back from
  connector metadata to server instructions, while the deferred path only
  carried `connector_description` and otherwise fell back to generic
  namespace text.
  
  That meant a plain MCP server could provide useful model-facing guidance
  and still appear as `Tools in the X namespace.` whenever it was
  discovered lazily through `tool_search`.
  
  ## What changed
  
  - Store one model-facing `namespace_description` on `ToolInfo`, using
  connector descriptions for connector-backed tools and server
  instructions for plain MCP servers.
  - Thread that namespace description through the `tool_search` source
  list, search indexing, and returned namespace metadata.
  - Add an end-to-end regression test for deferred non-app MCP search
  results exposing server instructions as the namespace description.
  
  ## Verification
  
  - `cargo test -p codex-tools
  search_tool_description_lists_each_mcp_source_once --lib`
  - `cargo test -p codex-core --test all
  tool_search_uses_non_app_mcp_server_instructions_as_namespace_description`

  sayan-oai · 2026-05-04 19:36:07 +00:00
  b9e8df47da
• Prepare selected environment plumbing (#20669)

  ## Why
  This is a prep PR in the multi-environment process-tool stack. It
  separates ownership/config cleanup from the behavior change that teaches
  process tools to route by selected environment, so the follow-up PR can
  focus on model-facing `environment_id` behavior.
  
  ## Stack
  1. https://github.com/openai/codex/pull/20646 - `EnvironmentContext`
  rendering for selected environments
  2. https://github.com/openai/codex/pull/20669 - selected-environment
  ownership and tool config prep (this PR)
  3. https://github.com/openai/codex/pull/20647 - process-tool
  `environment_id` routing
  
  ## What Changed
  - keep the resolved turn environment list wrapped in
  `ResolvedTurnEnvironments` through `TurnContext` instead of unwrapping
  it back to a raw `Vec`
  - add `TurnContext::resolve_path_against` so cwd-relative path
  resolution has one shared helper
  - replace the old tool config boolean with `ToolEnvironmentMode::{None,
  Single, Multiple}`
  
  ## Testing
  - Tests not run locally; this prep refactor is covered by GitHub CI for
  the stack.
  
  Co-authored-by: Codex <noreply@openai.com>

  starr-openai · 2026-05-04 17:55:49 +00:00
  905987c08f
• [tool_suggest] More prompt polishes. (#20566)

  Tool suggest still misfires when model needs tool_search, updating the
  prompts to further disambiguate it:
  
  - [x] rename it from `tool_suggest` to `request_plugin_install`
  - [x] rephrase "suggestion" to "install" in the tool descriptions.
  - [x] disambiguate "the tool" vs "the plugin/connector". 
  
  Tested with the Codex App and verified it still works.

  Matthew Zeng · 2026-05-02 04:22:12 +00:00
  f88701f5c8
• Gate multi-agent v2 tools independently of collab (#20246)

  ## Why
  
  `multi_agents_v2` is meant to be independently gated from the older
  `collab` feature. The tool registry still treated the
  collaboration-style agent tools as `collab`-only, so enabling
  `multi_agents_v2` without `collab` omitted the v2 agent tools. Review
  and guardian sub-sessions also need to keep agent spawning disabled even
  when the outer session has `multi_agents_v2` enabled.
  
  ## What changed
  
  - Include the collab-backed agent tools when either `multi_agents_v2` or
  `collab` is enabled.
  - Explicitly disable `multi_agents_v2` for review and guardian review
  sub-sessions, matching the existing `spawn_csv` and `collab`
  restrictions.
  - Add a registry test that enables `multi_agents_v2`, disables `collab`,
  and verifies the v2 agent tools are present while legacy `send_input`
  and `resume_agent` remain hidden.
  
  ## Testing
  
  - Added
  `test_build_specs_multi_agent_v2_does_not_require_collab_feature`.

  jif-oai · 2026-04-30 10:23:31 +02:00
  c37f7434ba
• Reduce the surface of collaboration modes (#20149)

  Collaboration modes were slightly invasive both into ThreadManager
  construction and ModelProvider

  pakrym-oai · 2026-04-29 17:22:41 -07:00
  fedcefe9da
• stop blocking unified_exec on Windows (#19435)

  ## Summary
  - remove the Windows-specific unified-exec environment block from tool
  selection
  - keep `unified_exec` default-off on Windows unless the feature is
  explicitly enabled
  - normalize model-provided `shell_type = unified_exec` to
  `shell_command` when the feature is disabled
  - drop obsolete tests tied to the removed environment gate and keep the
  feature-flag regression coverage
  
  ## Why
  Now that the session/long-lived process backend is implemented for the
  Windows sandbox, we don't need to hard disable it anymore. We will be
  rolling out slowly using a feature gate.
  
  ## Impact
  This allows manual Windows opt-in in CLI and app-backed flows while
  preserving the existing default-off behavior for Windows users.
  
  ---------
  
  Co-authored-by: canvrno-oai <kbond@openai.com>
  Co-authored-by: Codex <noreply@openai.com>

  iceweasel-oai · 2026-04-29 16:06:33 -07:00
  13dbcda28f
• [tool_suggest] Improve tool_suggest triggering conditions. (#20091)

  ## Summary
  - Tighten `tool_suggest` guidance so it prefers explicit plugin install
  requests, while still allowing a connector install when the relevant
  plugin is already installed and a needed connector from that plugin is
  missing.
  - Tell the model not to call `tool_suggest` in parallel with other
  tools.
  
  ## Testing
  - `cargo test -p codex-tools tool_suggest`
  - `cargo test -p codex-core tool_suggest`

  Matthew Zeng · 2026-04-29 13:41:12 -07:00
  8ce48f9968
• Delete multi_agent_v2 followup_task interrupt parameter (#20139)

  Messages sent with `followup_task` already arrive at their target
  recipient promptly (at message boundaries while sampling, or after the
  pending tool call completes) -- having `interrupt` is not worth the
  added complexity.

  Andrey Mishchenko · 2026-04-28 23:19:48 -07:00
  857146b328
• feat: disable capabilities by model provider (#19442)

  ## Why
  
  Unsupported features must fail closed and Codex must not expose
  OpenAI-hosted fallback paths when the active provider cannot support
  them. In practice, Bedrock should not surface app connectors, MCP
  servers, tool search/suggestions, image generation, web search, or JS
  REPL until those paths are explicitly supported for that provider.
  
  This PR moves that decision into provider-owned capability metadata
  instead of scattering Bedrock-specific checks across callers.
  
  ## What changed
  
  - Adds `ProviderCapabilities` to `codex-model-provider`, with default
  support for existing providers and a Bedrock override that disables
  unsupported launch surfaces.
  - Adds `ToolCapabilityBounds` to `codex-tools` so provider capability
  limits can clamp otherwise-enabled tool config.
  - Applies capability bounds when building session and review-thread tool
  config.
  - Routes MCP/app connector configuration through
  `McpManager::mcp_config`, which filters configured MCP servers and app
  connectors based on the active provider.
  - Updates app-server MCP list/read paths to use the filtered MCP config.
  - Adds coverage for default provider capabilities, Bedrock disabled
  capabilities, and optional tool-surface clamping.
  
  ## Testing
  
  built locally and verified that bedrock responses api now return without
  errors calling unsupported tools.

  Celia Chen · 2026-04-28 17:51:30 -07:00
  f8fe96d548
• Support disabling tool suggest for specific tools. (#20072)

  ## Summary
  - Add `disable_tool_suggest` to app and plugin config, schema, and
  TypeScript output
  - Exclude disabled connectors and plugins from tool suggestion discovery
  - Persist "never show again" tool-suggestion choices back into
  `config.toml`
  - Update config docs and add coverage for connector and plugin
  suppression
  
  ## Testing
  - Added and updated unit tests for config persistence and tool-suggest
  filtering
  - Not run (not requested)

  Matthew Zeng · 2026-04-29 00:19:34 +00:00
  ebdf3a878c
• Make MultiAgentV2 wait minimum configurable (#20052)

  ## Why
  
  MultiAgentV2 `wait_agent` currently clamps short waits to a fixed 10
  second minimum. That default is still useful for preventing tight
  polling loops, but it is too rigid for environments that need faster
  mailbox wake-up checks or a larger minimum to discourage frequent
  polling.
  
  This PR makes the minimum wait timeout configurable from the existing
  MultiAgentV2 feature config section, so operators can tune the behavior
  without changing the legacy multi-agent tool surface.
  
  ## What Changed
  
  - Added `features.multi_agent_v2.min_wait_timeout_ms`.
  - Defaulted the new setting to the existing 10 second floor.
  - Validated the configured value as `1..=3600000`, matching the existing
  one hour maximum wait bound.
  - Applied the configured minimum to MultiAgentV2 `wait_agent` runtime
  clamping.
  - Plumbed the configured minimum into the `wait_agent` tool schema,
  including the effective default when the minimum is above the normal 30
  second default.
  - Regenerated `core/config.schema.json`.
  
  ## Verification
  
  - `cargo test -p codex-features`
  - `cargo test -p codex-tools`
  - `cargo test -p codex-core --lib multi_agent_v2`
  - `just fix -p codex-core`

  jif-oai · 2026-04-28 22:36:44 +02:00
  34d71d43eb
• permissions: make runtime config profile-backed (#19606)

  ## Why
  
  This supersedes #19391. During stack repair, GitHub marked #19391 as
  merged into a temporary stack branch rather than into `main`, so the
  runtime-config change needed a fresh PR.
  
  `PermissionProfile` is now the canonical permissions shape after #19231
  because it can distinguish `Managed`, `Disabled`, and `External`
  enforcement while also carrying filesystem rules that legacy
  `SandboxPolicy` cannot represent cleanly. Core config and session state
  still needed to accept profile-backed permissions without forcing every
  profile through the strict legacy bridge, which rejected valid runtime
  profiles such as direct write roots.
  
  The unrelated CI/test hardening that previously rode along with this PR
  has been split into #19683 so this PR stays focused on the permissions
  model migration.
  
  ## What Changed
  
  - Adds `Permissions.permission_profile` and
  `SessionConfiguration.permission_profile` as constrained runtime state,
  while keeping `sandbox_policy` as a legacy compatibility projection.
  - Introduces profile setters that keep `PermissionProfile`, split
  filesystem/network policies, and legacy `SandboxPolicy` projections
  synchronized.
  - Uses a compatibility projection for requirement checks and legacy
  consumers instead of rejecting profiles that cannot round-trip through
  `SandboxPolicy` exactly.
  - Updates config loading, config overrides, session updates, turn
  context plumbing, prompt permission text, sandbox tags, and exec request
  construction to carry profile-backed runtime permissions.
  - Preserves configured deny-read entries and `glob_scan_max_depth` when
  command/session profiles are narrowed.
  - Adds `PermissionProfile::read_only()` and
  `PermissionProfile::workspace_write()` presets that match legacy
  defaults.
  
  ## Verification
  
  - `cargo test -p codex-core direct_write_roots`
  - `cargo test -p codex-core runtime_roots_to_legacy_projection`
  - `cargo test -p codex-app-server
  requested_permissions_trust_project_uses_permission_profile_intent`
  
  
  
  
  ---
  [//]: # (BEGIN SAPLING FOOTER)
  Stack created with [Sapling](https://sapling-scm.com). Best reviewed
  with [ReviewStack](https://reviewstack.dev/openai/codex/pull/19606).
  * #19395
  * #19394
  * #19393
  * #19392
  * __->__ #19606

  Michael Bolin · 2026-04-26 13:29:54 -07:00
  4d7ce3447d
• Add goal model tools (3 / 5) (#18075)

  Adds the model-facing goal tools on top of the app-server API from PR 2.
  
  ## Why
  
  Once goals are persisted and exposed to clients, the model needs a
  small, constrained tool surface for goal workflows. The tool contract
  should let the model inspect goals, create them only when explicitly
  requested, and mark them complete without giving it broad control over
  user/runtime-owned state.
  
  ## What changed
  
  - Added `get_goal`, `create_goal`, and `update_goal` tool specs behind
  the `goals` feature flag.
  - Added core goal tool handlers that validate objectives and token
  budgets before mutating persisted state.
  - Constrained `create_goal` to create only when no goal exists, with
  optional `token_budget` only when a budget is explicitly provided.
  - Tightened the `create_goal` instructions so the model does not infer
  goals from ordinary task requests.
  - Constrained `update_goal` to expose only goal completion; pause,
  resume, clear, and budget-limited transitions remain user- or
  runtime-controlled.
  - Registered the goal tools in the tool registry and kept them out of
  review contexts where they should not appear.
  
  ## Verification
  
  - Added tool-registry coverage for feature gating and tool availability.
  - Added core session tests for create/get/update behavior, duplicate
  goal rejection, budget validation, and completion-only updates.

  Eric Traut · 2026-04-24 20:54:40 -07:00
  32ace07ac5
• Remove js_repl feature (#19410)
  Curtis 'Fjord' Hawthorne · 2026-04-24 17:49:29 -07:00
  8a559e7938
• feat: surface multi-agent thread limit in spawn description (#19360)

  ## Summary
  - Thread `agent_max_threads` into `ToolsConfig` and
  `SpawnAgentToolOptions`.
  - Render the configured `max_concurrent_threads_per_session` value in
  the MultiAgentV2 `spawn_agent` description.
  - Cover the description text in `codex-tools` unit tests and
  `codex-core` tool spec tests.
  
  ## Validation
  - `just fmt`
  - `cargo test -p codex-tools`
  - `cargo test -p codex-core spawn_agent_description`
  - `git diff --check`
  
  ## Notes
  - `cargo test -p codex-core` was also attempted, but unrelated
  environment-sensitive tests failed with the active local environment.
  Examples: approvals reviewer defaults observed `AutoReview` instead of
  `User`, request-permissions event tests did not emit events, and
  proxy-env tests saw `http://127.0.0.1:50604` from the active proxy
  environment.
  
  Co-authored-by: Codex <noreply@openai.com>

  jif-oai · 2026-04-24 15:13:54 +02:00
  deb4509302
• guide Windows to use -WindowStyle Hidden for Start-Process calls (#19044)

  Sometimes codex runs `Start-Process` to start up a service or something
  similar, which launches a user-visible powershell window that probably
  doesn't get cleaned up. This instruction change encourages it to do so
  using a hidden window.
  
  This was reported in
  https://openai.slack.com/archives/C09K6H5DZC4/p1776741272870519
  
  One caveat is that this change won't do anything to cleanup these
  processes, but it will stop them from polluting the user's visible
  workspace
  
  ---------
  
  Co-authored-by: Codex <noreply@openai.com>

  iceweasel-oai · 2026-04-23 21:36:17 +00:00
  2e228969be
• [tool search] support namespaced deferred dynamic tools (#18413)

  Deferred dynamic tools need to round-trip a namespace so a tool returned
  by `tool_search` can be called through the same registry key that core
  uses for dispatch.
  
  This change adds namespace support for dynamic tool specs/calls,
  persists it through app-server thread state, and routes dynamic tool
  calls by full `ToolName` while still sending the app the leaf tool name.
  Deferred dynamic tools must provide a namespace; non-deferred dynamic
  tools may remain top-level.
  
  It also introduces `LoadableToolSpec` as the shared
  function-or-namespace Responses shape used by both `tool_search` output
  and dynamic tool registration, so dynamic tools use the same wrapping
  logic in both paths.
  
  Validation:
  - `cargo test -p codex-tools`
  - `cargo test -p codex-core tool_search`
  
  ---------
  
  Co-authored-by: Sayan Sisodiya <sayan@openai.com>

  pash-openai · 2026-04-21 14:13:08 +08:00
  dc1a8f2190
• [codex] prefer inherited spawn agent model (#18701)

  This updates the spawn-agent tool contract so subagents are presented as
  inheriting the parent model by default. The visible model list is now
  framed as optional overrides, the model parameter tells callers to leave
  it unset and the delegation guidance no longer nudges models toward
  picking a smaller/mini override.
  
  Fixes reports that 5.4 would occasionally pick 5.2 or lower as
  sub-agents.

  Thibault Sottiaux · 2026-04-20 22:34:08 +00:00
  54bd07d28c
• Update image outputs to default to high detail (#18386)

  Do not assume the default `detail`.

  pakrym-oai · 2026-04-18 11:01:12 -07:00
  53b1570367
• enable tool search over dynamic tools (#18263)

  ## Summary
  
  - Normalize deferred MCP and dynamic tools into `ToolSearchEntry` values
  before constructing `ToolSearchHandler`.
  - Move the tool-search entry adapter out of `tools/handlers` and into
  `tools/tool_search_entry.rs` so the handlers directory stays focused on
  handlers.
  - Keep `ToolSearchHandler` operating over one generic entry list for
  BM25 search, namespace grouping, and per-bucket default limits.
  
  ## Why
  
  Follow-up cleanup for #17849. The dynamic tool-search support made the
  handler juggle source-specific MCP and dynamic tool lists, index
  arithmetic, output conversion, and namespace emission. This keeps source
  adaptation outside the handler so the search loop itself is smaller and
  source-agnostic.
  
  ## Validation
  
  - `just fmt`
  - `cargo test -p codex-core tools::handlers::tool_search::tests`
  - `git diff --check`
  - `cargo test -p codex-core` currently fails in unrelated
  `plugins::manager::tests::list_marketplaces_ignores_installed_roots_missing_from_config`;
  rerunning that single test fails the same way at
  `core/src/plugins/manager_tests.rs:1692`.
  
  ---------
  
  Co-authored-by: pash <pash@openai.com>

  sayan-oai · 2026-04-18 02:07:59 +08:00
  6991be7ead
• Launch image generation by default (#17153)

  ## Summary
  - Promote `image_generation` from under-development to stable
  - Enable image generation by default in the feature registry
  - Update feature coverage for the new launch-state expectation
  - Add the missing image-generation auth fixture field in a tool registry
  test
  
  ## Testing
  - `just fmt`
  - `cargo test -p codex-features`
  - `cargo test -p codex-tools` currently fails:
  `test_full_toolset_specs_for_gpt5_codex_unified_exec_web_search` needs
  its expected default tool list updated for `image_generation`

  Won Park · 2026-04-16 09:05:38 -07:00
  85203d8872
• [code mode] defer mcp tools from exec description (#17287)

  ## Summary
  - hide deferred MCP/app nested tool descriptions from the `exec` prompt
  in code mode
  - add short guidance that omitted nested tools are still available
  through `ALL_TOOLS`
  - cover the code_mode_only path with an integration test that discovers
  and calls a deferred app tool
  
  ## Motivation
  `code_mode_only` exposes only top-level `exec`/`wait`, but the `exec`
  description could still include a large nested-tool reference. This
  keeps deferred nested tools callable while avoiding that prompt bloat.
  
  ## Tests
  - `just fmt`
  - `just fix -p codex-code-mode`
  - `just fix -p codex-tools`
  - `cargo test -p codex-code-mode
  exec_description_mentions_deferred_nested_tools_when_available`
  - `cargo test -p codex-tools
  create_code_mode_tool_matches_expected_spec`
  - `cargo test -p codex-core
  code_mode_only_guides_all_tools_search_and_calls_deferred_app_tools`

  sayan-oai · 2026-04-17 00:01:14 +08:00
  9c6d038622
• Update ToolSearch to be enabled by default (#17854)

  ## Summary
  - Promote `Feature::ToolSearch` to `Stable` and enable it in the default
  feature set
  - Update feature tests and tool registry coverage to match the new
  default
  - Adjust the search-tool integration test to assert the default-on path
  and explicit disable fallback
  
  ## Testing
  - `just fmt`
  - `cargo test -p codex-features`
  - `cargo test -p codex-core --test all search_tool`
  - `cargo test -p codex-tools`

  Matthew Zeng · 2026-04-15 22:01:05 -07:00
  77fe33bf72
• Support original-detail metadata on MCP image outputs (#17714)

  ## Summary
  - honor `_meta["codex/imageDetail"] == "original"` on MCP image content
  and map it to `detail: "original"` where supported
  - strip that detail back out when the active model does not support
  original-detail image inputs
  - update code-mode `image(...)` to accept individual MCP image blocks
  - teach `js_repl` / `codex.emitImage(...)` to preserve the same hint
  from raw MCP image outputs
  - document the new `_meta` contract and add generic RMCP-backed coverage
  across protocol, core, code-mode, and js_repl paths

  Curtis 'Fjord' Hawthorne · 2026-04-15 14:43:33 -07:00
  9e2fc31854
• Fix empty tool descriptions (#17946)

  ## Summary
  - Ensure direct namespaced MCP tool groups are emitted with a non-empty
  namespace description even when namespace metadata is missing or blank.
  - Add regression coverage for missing MCP namespace descriptions.
  
  ## Cause
  Latest `main` can serialize a direct namespaced MCP tool group with an
  empty top-level `description`. The namespace description path used
  `unwrap_or_default()` when `tool_namespaces` did not include metadata
  for that namespace, so the outbound Responses API payload could contain
  a tool like `{"type":"namespace","description":""}`. The Responses API
  rejects that because namespace tool descriptions must be a non-empty
  string.
  
  ## Fix
  - Add a fallback namespace description: `Tools in the <namespace>
  namespace.`
  - Preserve provided namespace descriptions after trimming, but treat
  blank descriptions as missing.
  
  ### Issue I am seeing
  This is what I am seeing on the local build.
  <img width="1593" height="488" alt="Screenshot 2026-04-15 at 10 55 55
  AM"
  src="https://github.com/user-attachments/assets/bab668ba-bf17-4c71-be4e-b102202fce57"
  />
  
  ---------
  
  Co-authored-by: Sayan Sisodiya <sayan@openai.com>

  Shijie Rao · 2026-04-15 18:14:43 +00:00
  78ce61c78e
• register all mcp tools with namespace (#17404)

  stacked on #17402.
  
  MCP tools returned by `tool_search` (deferred tools) get registered in
  our `ToolRegistry` with a different format than directly available
  tools. this leads to two different ways of accessing MCP tools from our
  tool catalog, only one of which works for each. fix this by registering
  all MCP tools with the namespace format, since this info is already
  available.
  
  also, direct MCP tools are registered to responsesapi without a
  namespace, while deferred MCP tools have a namespace. this means we can
  receive MCP `FunctionCall`s in both formats from namespaces. fix this by
  always registering MCP tools with namespace, regardless of deferral
  status.
  
  make code mode track `ToolName` provenance of tools so it can map the
  literal JS function name string to the correct `ToolName` for
  invocation, rather than supporting both in core.
  
  this lets us unify to a single canonical `ToolName` representation for
  each MCP tool and force everywhere to use that one, without supporting
  fallbacks.

  sayan-oai · 2026-04-15 21:02:59 +08:00
  0df7e9a820
• Always enable original image detail on supported models (#17665)

  ## Summary
  
  This PR removes `image_detail_original` as a runtime experiment and
  makes original image detail available whenever the selected model
  supports it.
  
  Concretely, this change:
  - drops the `image_detail_original` feature flag from the feature
  registry and generated config schema
  - makes tool-emitted image detail depend only on
  `ModelInfo.supports_image_detail_original`
  - updates `view_image` and `code_mode`/`js_repl` image emission to use
  that capability check directly
  - removes now-redundant experiment-specific tests and instruction
  coverage
  - keeps backward compatibility for existing configs by silently ignoring
  a stale `features.image_detail_original` entry
  
  The net effect is that `detail: "original"` is always available on
  supported models, without requiring an experiment toggle.

  Curtis 'Fjord' Hawthorne · 2026-04-14 08:15:56 -07:00
  f030ab62eb
• chore: refactor name and namespace to single type (#17402)

  avoid passing them both around, unify on a type. this now also keys
  `ToolRegistry`.
  
  tests pass

  sayan-oai · 2026-04-11 23:06:22 +00:00
  1325bcd3f6
• feat: description multi-agent v2 (#17338)
  jif-oai · 2026-04-10 15:31:32 +01:00
  d39a722865
• Add output_schema to code mode render (#17210)

  This updates code-mode tool rendering so MCP tools can surface
  structured output types from their `outputSchema`.
  
  What changed:
  - Detect MCP tool-call result wrappers from the output schema shape
  instead of relying on tool-name parsing or provenance flags.
  - Render shared TypeScript aliases once for MCP tool results
  (`CallToolResult`, `ContentBlock`, etc.) so multiple MCP tool
  declarations stay compact.
  - Type `structuredContent` from the tool definition's `outputSchema`
  instead of rendering it as `unknown`.
  - Update the shared MCP aliases to match the MCP draft `CallToolResult`
  schema more closely.
  
  Example:
  - Before: `declare const tools: { mcp__rmcp__echo(args: { env_var?:
  string; message: string; }): Promise<{ _meta?: unknown; content:
  Array<unknown>; isError?: boolean; structuredContent?: unknown; }>; };`
  - After: `declare const tools: { mcp__rmcp__echo(args: { env_var?:
  string; message: string; }): Promise<CallToolResult<{ echo: string; env:
  string | null; }>>; };`

  Vivian Fang · 2026-04-10 11:41:44 +00:00
  7bbe3b6011
• preserve search results order in tool_search_output (#17263)

  we used to alpha-sort tool search results because we were using
  `BTreeMap`, which threw away the actual search result ordering.
  
  Now we use a vec to preserve it.
  
  ### Tests
  Updated tests

  sayan-oai · 2026-04-09 18:15:10 -07:00
  04fc208b6d
• [mcp] Expand tool search to custom MCPs. (#16944)

  - [x] Expand tool search to custom MCPs.
  - [x] Rename several variables/fields to be more generic.
  
  Updated tool & server name lifecycles:
  
  **Raw Identity**
  
  ToolInfo.server_name is raw MCP server name.
  ToolInfo.tool.name is raw MCP tool name.
  MCP calls route back to raw via parse_tool_name() returning
  (tool.server_name, tool.tool.name).
  mcpServerStatus/list now groups by raw server and keys tools by
  Tool.name: mod.rs:599
  App-server just forwards that grouped raw snapshot:
  codex_message_processor.rs:5245
  
  **Callable Names**
  
  On list-tools, we create provisional callable_namespace / callable_name:
  mcp_connection_manager.rs:1556
  For non-app MCP, provisional callable name starts as raw tool name.
  For codex-apps, provisional callable name is sanitized and strips
  connector name/id prefix; namespace includes connector name.
  Then qualify_tools() sanitizes callable namespace + name to ASCII alnum
  / _ only: mcp_tool_names.rs:128
  Note: this is stricter than Responses API. Hyphen is currently replaced
  with _ for code-mode compatibility.
  
  **Collision Handling**
  
  We do initially collapse example-server and example_server to the same
  base.
  Then qualify_tools() detects distinct raw namespace identities behind
  the same sanitized namespace and appends a hash to the callable
  namespace: mcp_tool_names.rs:137
  Same idea for tool-name collisions: hash suffix goes on callable tool
  name.
  Final list_all_tools() map key is callable_namespace + callable_name:
  mcp_connection_manager.rs:769
  
  **Direct Model Tools**
  
  Direct MCP tool declarations use the full qualified sanitized key as the
  Responses function name.
  The raw rmcp Tool is converted but renamed for model exposure.
  
  **Tool Search / Deferred**
  
  Tool search result namespace = final ToolInfo.callable_namespace:
  tool_search.rs:85
  Tool search result nested name = final ToolInfo.callable_name:
  tool_search.rs:86
  Deferred tool handler is registered as "{namespace}:{name}":
  tool_registry_plan.rs:248
  When a function call comes back, core recombines namespace + name, looks
  up the full qualified key, and gets the raw server/tool for MCP
  execution: codex.rs:4353
  
  **Separate Legacy Snapshot**
  
  collect_mcp_snapshot_from_manager_with_detail() still returns a map
  keyed by qualified callable name.
  mcpServerStatus/list no longer uses that; it uses
  McpServerStatusSnapshot, which is raw-inventory shaped.

  Matthew Zeng · 2026-04-09 13:34:52 -07:00
  d7f99b0fa6
• Fix missing fields (#17149)

  Fix missing `image_generation_tool_auth_allowed` in two locations.

  canvrno-oai · 2026-04-08 13:53:53 -07:00
  58ad79b60e
• release ready, enabling only for siwc users (#17046)

  **Disabling Image-Gen for Non-SIWC Codex Users**
  
  We are only enabling image-gen feature for SIWC Codex users until there
  comes a fix in ResponsesAPI to omit output from responses.completed, to
  prevent the following issues:
  
  1. websocket blows up due to heavier load (images) than before (text) 
  2. http parser streams through n^2 of n-base64 bytes (sum of base64s of
  all images generated in turn) that causes long delays in
  turn_completion.

  Won Park · 2026-04-08 11:22:39 -07:00
  e003f84e1e
• Configure multi_agent_v2 spawn agent hints (#17071)

  Allow multi_agent_v2 features to have its own temporary configuration
  under `[features.multi_agent_v2]`
  
  ```
  [features.multi_agent_v2]
  enabled = true
  usage_hint_enabled = false
  usage_hint_text = "Custom delegation guidance."
  hide_spawn_agent_metadata = true
  ```
  
  Absent `usage_hint_text` means use the default hint.
  
  ```
  [features]
  multi_agent_v2 = true
  ```
  
  still works as the boolean shorthand.

  pakrym-oai · 2026-04-08 08:42:18 -07:00
  4c07dd4d25
• Render namespace description for tools (#16879)
  Vivian Fang · 2026-04-08 02:39:40 -07:00
  d47b755aa2
• Render function attribute descriptions (#16880)
  Vivian Fang · 2026-04-08 02:10:45 -07:00
  9091999c83
• Support anyOf and enum in JsonSchema (#16875)

  This brings us into better alignment with the JSON schema subset that is
  supported in
  <https://developers.openai.com/api/docs/guides/structured-outputs#supported-schemas>,
  and also allows us to render richer function signatures in code mode
  (e.g., anyOf{null, OtherObjectType})

  Vivian Fang · 2026-04-08 01:07:55 -07:00
  ea516f9a40
• Add regression tests for JsonSchema (#17052)

  Tests added for existing JsonSchema in
  `codex-rs/tools/src/json_schema_tests.rs`:
  
  - `parse_tool_input_schema_coerces_boolean_schemas`
  - `parse_tool_input_schema_infers_object_shape_and_defaults_properties`
  - `parse_tool_input_schema_normalizes_integer_and_missing_array_items`
  - `parse_tool_input_schema_sanitizes_additional_properties_schema`
  -
  `parse_tool_input_schema_infers_object_shape_from_boolean_additional_properties_only`
  - `parse_tool_input_schema_infers_number_from_numeric_keywords`
  - `parse_tool_input_schema_infers_number_from_multiple_of`
  -
  `parse_tool_input_schema_infers_string_from_enum_const_and_format_keywords`
  - `parse_tool_input_schema_defaults_empty_schema_to_string`
  - `parse_tool_input_schema_infers_array_from_prefix_items`
  -
  `parse_tool_input_schema_preserves_boolean_additional_properties_on_inferred_object`
  -
  `parse_tool_input_schema_infers_object_shape_from_schema_additional_properties_only`
  
  Tests that we expect to fail on the baseline normalizer, but pass with
  the new JsonSchema:
  
  - `parse_tool_input_schema_preserves_nested_nullable_type_union`
  - `parse_tool_input_schema_preserves_nested_any_of_property`

  Vivian Fang · 2026-04-07 18:18:54 -07:00
  fa5119a8a6
• Use model metadata for Fast Mode status (#16949)

  Fast Mode status was still tied to one model name in the TUI and
  model-list plumbing. This changes the model metadata shape so a model
  can advertise additional speed tiers, carries that field through the
  app-server model list, and uses it to decide when to show Fast Mode
  status.
  
  For people using Codex, the behavior is intended to stay the same for
  existing models. Fast Mode still requires the existing signed-in /
  feature-gated path; the difference is that the UI can now recognize any
  model the model list marks as Fast-capable, instead of requiring a new
  client-side slug check.

  pash-openai · 2026-04-07 17:55:40 -07:00
  80ebc80be5
• chore: keep request_user_input tool to persist cache on multi-agents (#17009)
  jif-oai · 2026-04-07 16:53:31 +01:00
  4cc6818996
• chore: hide nickname for debug flag (#17007)
  jif-oai · 2026-04-07 11:31:13 +01:00
  99f167e6bf
• chore: send_message and followup_task do not return anything (#17008)
  jif-oai · 2026-04-07 11:26:36 +01:00
  68e16baabe
• feat: drop agent ID from v2 (#17005)
  jif-oai · 2026-04-07 10:56:01 +01:00
  2a8c3a2a52
• chore: debug flag to hide some parameters (#17002)
  jif-oai · 2026-04-07 10:42:19 +01:00
  e2bb45bb24
• Disable env-bound tools when exec server is none (#16349)

  ## Summary
  - make `CODEX_EXEC_SERVER_URL=none` map to an explicit disabled
  environment mode instead of inferring from a missing URL
  - expose environment capabilities (`exec_enabled`, `filesystem_enabled`)
  so tool building can gate behavior explicitly and future
  multi-environment work has a clearer seam
  - suppress env-backed tools when the relevant capability is unavailable,
  including exec tools, `js_repl`, `apply_patch`, `list_dir`, and
  `view_image`
  - keep handler/runtime backstops so disabled environments still reject
  execution if a tool path somehow bypasses registration
  
  ## Testing
  - `just fmt`
  - `cargo test -p codex-exec-server`
  - `cargo test -p codex-tools
  disabled_environment_omits_environment_backed_tools`
  - `cargo test -p codex-tools
  environment_capabilities_gate_exec_and_filesystem_tools_independently`
  - remote devbox Bazel build via `codex-applied-devbox`:
  `//codex-rs/cli:cli`

  starr-openai · 2026-04-06 17:22:06 -07:00
  a504d8f0fa