Draft
Conversation
Refactor EnvironmentManager to own a keyed environment registry with explicit default and local lookups. Keep remote exec-server connections lazy at environment use sites and preserve disabled agent environment access separately from internal local environment access. Co-authored-by: Codex <noreply@openai.com>
Use EnvironmentManager::new for args-struct construction and keep from_env methods as the env-var factory entrypoints. Co-authored-by: Codex <noreply@openai.com>
Return concrete default and local environments from EnvironmentManager now that the manager always installs local and default entries. Keep arbitrary ID lookup optional. Co-authored-by: Codex <noreply@openai.com>
Keep the lazy remote exec-server client wrapper alongside ExecServerClient and import it from the client module for environment-backed exec and filesystem use. Co-authored-by: Codex <noreply@openai.com>
Use EnvironmentManager::new with EnvironmentManagerArgs for runtime-path-aware construction and keep from_env only for the no-args env-var factory. Co-authored-by: Codex <noreply@openai.com>
Add high-level EnvironmentManager docs for local/remote initialization, default environment selection, disabled agent access, and lazy remote connection behavior. Co-authored-by: Codex <noreply@openai.com>
Drop the unused local_environment helper and keep local lookups on the generic get_environment API. Co-authored-by: Codex <noreply@openai.com>
Clarify that SessionServices carries an Arc handle to the process-level EnvironmentManager rather than owning a session-specific manager. Co-authored-by: Codex <noreply@openai.com>
Restore CODEX_EXEC_SERVER_URL=none semantics by making EnvironmentManager::default_environment return None when environment access is disabled. Remove the separate disabled-for-agent flag and derive tool availability from the optional default environment. Add an end-to-end tool exposure test for CODEX_EXEC_SERVER_URL=none. Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Create one lazy exec-server client per remote environment and pass clones into the remote process and filesystem backends. This keeps ExecServerClient as the connected-client type while avoiding duplicate websocket clients for one environment. Co-authored-by: Codex <noreply@openai.com>
Make EnvironmentManagerArgs::default() own CODEX_EXEC_SERVER_URL parsing so production entrypoints can keep using EnvironmentManager::new with struct update syntax for runtime paths. Add explicit test defaults so test managers do not depend on the process environment. Co-authored-by: Codex <noreply@openai.com>
Drop the networked integration test for CODEX_EXEC_SERVER_URL=none omitting environment-backed tools. Lower-level coverage already verifies disabled environments omit those tools. Co-authored-by: Codex <noreply@openai.com>
Make EnvironmentManagerArgs carry ExecServerRuntimePaths for production construction and route test-only unsandboxed setup through explicit _for_tests helpers. Use the manager local environment for MCP and app-server filesystem fallbacks instead of constructing a fresh default environment. Co-authored-by: Codex <noreply@openai.com>
Remove the app-server-client re-export now that environment-manager construction owns CODEX_EXEC_SERVER_URL reading directly in exec-server. Co-authored-by: Codex <noreply@openai.com>
Add a connector loading helper that accepts the existing EnvironmentManager and switch app-server paths to use it. Keep the config-only helper as a temporary fallback for callers such as TUI that do not yet pass the manager through. Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Codex <noreply@openai.com>
Allow thread/start to configure sticky environment selections that are used by turns when no per-turn override is supplied. Per-turn environments continue to take precedence, while omitted thread selections preserve the existing default behavior. Co-authored-by: Codex <noreply@openai.com>
Cover sticky thread environment selections and turn-level overrides through the app-server v2 thread/start and turn/start JSON-RPC flow. The matrix mirrors the manual smoke cases for omitted, empty, local, remote, and local plus remote selections. Co-authored-by: Codex <noreply@openai.com>
Group thread-start options for lint-friendly callsites and update generated v2 schema for sticky environment selections. Co-authored-by: Codex <noreply@openai.com>
starr-openai
changed the base branch from
starr/sticky-env-split-2-config-toml
to
starr/sticky-env-split-1-api-thread
starr-openai
force-pushed
the
starr/sticky-env-split-1-api-thread
branch
from
b810587 to
a578544
Compare
Reject named turn and sticky environments when environment access is disabled, while preserving explicit empty environment selections as no-environment turns. Validate sticky environment state before storing it so invalid selections cannot poison future turns. Co-authored-by: Codex <noreply@openai.com>
starr-openai
force-pushed
the
starr/sticky-env-split-1-api-thread
branch
from
a578544 to
309cfc8
Compare
starr-openai
force-pushed
the
starr/sticky-env-split-3-downstream-tools
branch
7 times, most recently
from
682e310 to
c00ff07
Compare
Route apply_patch, unified exec, view_image, tool schemas, and model-visible environment context through selected thread environments. Update downstream tests and fixtures for environment-aware tool behavior. Co-authored-by: Codex <noreply@openai.com>
starr-openai
force-pushed
the
starr/sticky-env-split-3-downstream-tools
branch
from
c00ff07 to
8d7c1a5
Compare
starr-openai
force-pushed
the
starr/sticky-env-split-1-api-thread
branch
6 times, most recently
from
a5e0d6c to
23ff063
Compare
starr-openai
force-pushed
the
starr/sticky-env-split-1-api-thread
branch
8 times, most recently
from
ce102d0 to
a1ec341
Compare
starr-openai
added a commit
that referenced
this pull request
Apr 24, 2026
## Summary - add sticky environment selections to app-server v2 thread/start and turn/start request flow - carry thread-level selections through core session/thread state - add app-server coverage for sticky selections and turn overrides ## Stack 1. This PR: API and thread persistence 2. #18898: config.toml named environment loading 3. #18899: downstream tool/runtime consumers ## Validation - Not run locally; split only. --------- Co-authored-by: Codex <noreply@openai.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Stack
Validation