Conversation naming #8991

pap-openai · 2026-01-09T18:42:36Z

Session renaming:

/rename my_session
/rename without arg and passing an argument in customViewPrompt
AppExitInfo shows resume hint using the session name if set instead of uuid, defaults to uuid if not set
Names are stored in CODEX_HOME/sessions.jsonl

Session resuming:

codex resume lookup for CODEX_HOME/sessions.jsonl first entry matching the name and resumes the session

JaviSoto

Looks good!!

Random suggestion I just thought of: it could be useful to give the agent a way to set the name as well via a tool call, so you can tell the agent to name the conversation as well? (you could even add something to your AGENTS.md with instructions on how you want the agent to name your sessions, or to rename them in certain cases etc)

JaviSoto · 2026-01-09T18:59:04Z

codex-rs/cli/src/main.rs

 #[derive(Debug, Parser)]
 struct ResumeCommand {
-    /// Conversation/session id (UUID). When provided, resumes this session.
+    /// Conversation/session id (UUID) or session name. UUIDs take precedence if it parses.


codex-rs/cli/src/main.rs

codex-rs/core/src/rollout/list.rs

codex-rs/core/src/codex.rs

codex-rs/core/src/util.rs

codex-rs/tui2/src/lib.rs

codex-rs/tui/src/lib.rs

JaviSoto · 2026-01-09T19:18:00Z

codex-rs/protocol/src/protocol.rs

+pub struct SessionMetaUpdatedEvent {
+    /// Name left as session_id instead of thread_id for backwards compatibility.
+    pub session_id: ThreadId,
+    #[serde(default, skip_serializing_if = "Option::is_none")]


jif-oai

I would def go to the index file

can we try to use thread instead of session as much as possible

jif-oai · 2026-01-12T14:09:06Z

codex-rs/core/src/rollout/recorder.rs

    Ok(())
 }

+async fn rewrite_session_name(


Those functions should not be needed. We should use the same mechanism as for the model used during the rollout for example

codex-rs/core/src/rollout/list.rs

codex-rs/core/src/rollout/recorder.rs

codex-rs/protocol/src/protocol.rs

… name

pap-openai · 2026-01-13T01:28:29Z

@jif-oai @JaviSoto moved to an index file instead of SessionMeta so we don't touch rollout files. We're reading a CODEX_HOME/session_index.jsonl from most recent renames to oldest. We're appending new names even if previous session had a name to avoid searching for the entry with name + rewriting file and keep write O(1).

Implemented your comments (adding tests on resume, move to thread, thread_name or thread_id naming)

pap-openai · 2026-01-26T15:16:48Z

now showing

I'm wondering if it won't be confusing though given the composer is still showing (we're still in the TUI) compared to when we exit (in which case we'll show codex resume anyway if a name is set). Should we keep @jif-oai ?

slkzgm · 2026-01-27T12:45:15Z

Hi! Thanks for putting this together — it addresses the core pain point of making sessions easier to find via codex resume.

I’d like to contribute a small “minimal integration” improvement that stays within the existing flow:

add a more explicit /name <…> command (possibly as an alias to /rename) to name the current session from inside the conversation,
surface the name directly in the codex resume list/picker to improve recognizability,
and if codex resume <name> is ambiguous (multiple matches), fall back to the picker rather than adding complex collision logic.

Would appreciate maintainer guidance on whether you prefer /name vs /rename, and whether codex resume <name> should require unique matches or just fall back to the picker on ambiguity.

If this approach sounds good, I’m happy to send a small follow-up PR (or base changes off your branch) with the tweaks + tests.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 84324c0d41

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-01-28T16:23:00Z

codex-rs/core/src/rollout/session_index.rs

+    let line = serde_json::to_string(entry).map_err(std::io::Error::other)?;
+    file.write_all(line.as_bytes()).await?;
+    file.write_all(b"\n").await?;


Write session_index entries atomically

The append path writes the JSON line and the newline as two separate write_all calls. If two Codex sessions/processes rename threads concurrently (same codex_home), the writes can interleave (e.g., json1json2\n\n), producing malformed JSONL that scan_index_from_end* will skip and making resume by name fail. Write the line plus newline in a single write (or lock the file) to keep each entry atomic under concurrent appends.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2eb7459e9b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-01-28T16:31:33Z

codex-rs/app-server/src/codex_message_processor.rs

+        self.outgoing
+            .send_response(request_id, ThreadSetNameResponse {})
+            .await;


Don’t acknowledge rename before it succeeds

The handler sends a success response immediately after enqueueing Op::SetThreadName, but the core implementation can still fail later (e.g., persistence disabled or a write error when appending to the session index), in which case it emits an Error event and never sends ThreadNameUpdated (see codex-rs/core/src/codex.rs around lines 2953–2979). That means JSON-RPC clients that rely on the response will think the rename succeeded even though it didn’t persist, and resume-by-name will silently fail. Consider waiting for the ThreadNameUpdated/error event (similar to thread_rollback) or pre-checking persistence and returning an error response.

Useful? React with 👍 / 👎.

…ession-naming-clean

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5d90138ad1

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-01-29T18:54:39Z

codex-rs/core/src/util.rs

+pub fn resume_command(thread_name: Option<&str>, thread_id: Option<ThreadId>) -> Option<String> {
+    let resume_target = thread_name
+        .filter(|name| !name.is_empty())
+        .map(str::to_string)
+        .or_else(|| thread_id.map(|thread_id| thread_id.to_string()));


Guard resume hints against UUID-looking thread names

Because resume_command always prefers thread_name, a thread renamed to something that parses as a UUID will produce a hint like codex resume <name>. Both TUI and exec resume resolution treat UUID-looking strings as IDs (Uuid::parse_str branches in tui/src/lib.rs and exec/src/lib.rs), so that hint will resume a different thread or fail if the UUID doesn’t exist. Consider falling back to the actual thread_id when the name parses as a UUID (and doesn’t match the thread’s id), or add an explicit disambiguation flag for name-based resume.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 38b9d8bdb0

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-01-29T19:10:48Z

codex-rs/tui/src/bottom_pane/chat_composer.rs

                    self.personality_command_enabled,
                    self.windows_degraded_sandbox_active,
                )
-                && cmd == SlashCommand::Review
+                && matches!(cmd, SlashCommand::Review | SlashCommand::Rename)
            {


Update chat composer docs for new /rename arg handling

This change expands the chat-composer command-with-args path to include /rename, which modifies the state machine behavior. Per codex-rs/tui/src/bottom_pane/AGENTS.md, any chat-composer state-machine change must be reflected in the module docs and docs/tui-chat-composer.md. Those docs were not updated here, so the documented slash-command dispatch behavior is now out of sync and will be misleading to future maintainers. Please sync the docs with the new /rename handling.

Useful? React with 👍 / 👎.

…ession-naming-clean

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 68997282f5

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-01-29T19:48:58Z

codex-rs/app-server/src/codex_message_processor.rs

+        self.outgoing
+            .send_response(request_id, ThreadSetNameResponse {})
+            .await;


Return an error when rename fails in core

The thread/name/set handler sends a success response immediately after enqueueing Op::SetThreadName, but the core handler can still fail later (e.g., persistence disabled or session_index.jsonl write errors) and emits an Error event instead of a ThreadNameUpdated event (see codex-rs/core/src/codex.rs around the set_thread_name error branches). In those cases clients get {} even though the rename didn’t persist, so UIs may update state incorrectly unless they separately watch for error notifications. Consider waiting for a ThreadNameUpdated/Error event to reply, or pre-check persistence and IO failures so the RPC response accurately reflects success.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e021ed2627

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-01-30T09:25:22Z

codex-rs/tui/src/lib.rs

+            let is_uuid = Uuid::parse_str(id_str).is_ok();
+            let path = if is_uuid {
+                find_thread_path_by_id_str(&config.codex_home, id_str).await?
+            } else {
+                find_thread_path_by_name_str(&config.codex_home, id_str).await?


Avoid UUID-like thread names breaking name resume

Because resume selection treats any argument that parses as a UUID as an ID (Uuid::parse_str branch here), a user can rename a thread to a UUID-like string and then be unable to resume it by name (the lookup never reaches find_thread_path_by_name_str, and the printed codex resume <name> hint will resolve by ID instead). This only shows up when the chosen name parses as a UUID, but it makes the new rename feature unreliable in that case and can even resume the wrong thread if that UUID exists. Consider rejecting UUID-like names on rename or falling back to name lookup when the UUID path isn’t found.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e7abfec081

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-01-30T10:22:48Z

codex-rs/app-server/src/codex_message_processor.rs

+        self.outgoing
+            .send_response(request_id, ThreadSetNameResponse {})
+            .await;


Don’t acknowledge rename before it succeeds

The app-server replies success immediately after enqueuing Op::SetThreadName, but the actual rename can still fail later (e.g., session persistence disabled or I/O error appending the session index in codex-rs/core/src/codex.rs around the set_thread_name handler). In those cases the client will already have received {} as a success response even though an Error event is emitted and the name is not updated, which can leave the UI/state inconsistent. Consider deferring the response until a ThreadNameUpdated event arrives for this request, or translating the failure Error into a request error response.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-01-30T10:22:48Z

codex-rs/core/src/util.rs

+pub fn resume_command(thread_name: Option<&str>, thread_id: Option<ThreadId>) -> Option<String> {
+    let resume_target = thread_name
+        .filter(|name| !name.is_empty())
+        .map(str::to_string)
+        .or_else(|| thread_id.map(|thread_id| thread_id.to_string()));


Avoid suggesting resume-by-name for UUID-like names

resume_command always prefers thread_name over thread_id, but the CLI/TUI resume logic treats any UUID-parsing string as an ID (Uuid::parse_str in codex-rs/tui/src/lib.rs around the resume/fork path selection). If a user renames a thread to a UUID-looking string, the resume hint will suggest codex resume <name>, but the CLI will interpret it as an ID and either resume the wrong session or fail if that ID doesn’t exist. You may want to detect UUID-looking names and fall back to the actual thread_id (or disallow such names) so the hint matches the real resume behavior.

Useful? React with 👍 / 👎.

pap-openai added 10 commits January 9, 2026 11:45

adding session naming with /rename and metadata storage in SessionMeta

0a6d2d5

fix Restore writer file handle on rename failure

bfc7a3e

changing attribute to name + support command with arg

27bbc02

fix

41e7e02

Adding session_configured_event & /status output of session name

9f6aac5

+docs

4d6e184

resume work wip

3fd2ccc

Merge branch 'resume-name' into session-naming-clean

bbab77f

safe session_name + always showing session name in /status

a0f9f68

adding appexitinfo hint codex resume <name> if set

eebd0a2

JaviSoto reviewed Jan 9, 2026

View reviewed changes

jif-oai reviewed Jan 12, 2026

View reviewed changes

pap-openai added 12 commits January 12, 2026 22:43

uses thread_name instead of session_name

32d4278

adds shell escaping for exit hint info instead of restrictive session…

aa4d062

… name

moving to index file instead of sessionmeta

3186b17

removing name from SessionMeta

8d57df0

ThreadNameUpdated instead of SessionMetaUpdated

d9c675b

moving crate up

1e568e6

reverting changes in codex-rs/core/src/rollout/list.rs

b385eb8

revert rollout changes

2da5bf2

adding test for resume command with id

e62459e

adding hint on uuid or name on exit

b8de956

naming clarity

9de9d3e

documenting public function + thread_id rename

61a0477

pap-openai requested a review from jif-oai January 13, 2026 01:31

app-server v2 + thread_id

00ec8d1

pap-openai changed the title ~~WIP: Conversation naming~~ Conversation naming Jan 13, 2026

etraut-openai added the oai PRs contributed by OpenAI employees label Jan 14, 2026

Merge main into session-naming-clean

4c32273

persistence enabled smaller scope

8458bee

pap-openai added 3 commits January 26, 2026 15:32

adding tests for multiple or no entries

51b32f6

adding hint for resume on renaming

b859ba1

Merge remote-tracking branch 'origin' into session-naming-clean

4aa8c6a

pap-openai added 2 commits January 28, 2026 16:40

Merge branch 'main' into session-naming-clean

35e4a0d

add rename tooltip

84324c0

chatgpt-codex-connector bot reviewed Jan 28, 2026

View reviewed changes

Merge branch 'main' into session-naming-clean

2eb7459

chatgpt-codex-connector bot reviewed Jan 28, 2026

View reviewed changes

pap-openai added 3 commits January 28, 2026 17:42

fix two atomic writes

00c061f

Merge branch 'session-naming-clean' of github.com:openai/codex into s…

99e0835

…ession-naming-clean

Merge branch 'main' into session-naming-clean

5d90138

chatgpt-codex-connector bot reviewed Jan 29, 2026

View reviewed changes

Merge branch 'main' into session-naming-clean

38b9d8b

chatgpt-codex-connector bot reviewed Jan 29, 2026

View reviewed changes

pap-openai added 2 commits January 29, 2026 19:33

fix

e71c628

Merge branch 'session-naming-clean' of github.com:openai/codex into s…

6899728

…ession-naming-clean

chatgpt-codex-connector bot reviewed Jan 29, 2026

View reviewed changes

Merge branch 'main' into session-naming-clean

e021ed2

chatgpt-codex-connector bot reviewed Jan 30, 2026

View reviewed changes

pap-openai added 2 commits January 30, 2026 09:29

fix

112e5b6

Merge branch 'main' into session-naming-clean

e7abfec

chatgpt-codex-connector bot reviewed Jan 30, 2026

View reviewed changes

pap-openai enabled auto-merge (squash) January 30, 2026 10:23

Merge branch 'main' into session-naming-clean

9998ca8

pap-openai merged commit 1ef5455 into main Jan 30, 2026
32 checks passed

pap-openai deleted the session-naming-clean branch January 30, 2026 10:40

github-actions bot locked and limited conversation to collaborators Jan 30, 2026

Conversation naming #8991

Conversation naming #8991

Uh oh!

Conversation

pap-openai commented Jan 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

JaviSoto left a comment

Choose a reason for hiding this comment

Uh oh!

JaviSoto Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

JaviSoto Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

jif-oai left a comment

Choose a reason for hiding this comment

Uh oh!

jif-oai Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

pap-openai commented Jan 13, 2026

Uh oh!

pap-openai commented Jan 26, 2026

Uh oh!

slkzgm commented Jan 27, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Jan 29, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Jan 29, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Jan 29, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Jan 30, 2026

Choose a reason for hiding this comment

pap-openai commented Jan 9, 2026 •

edited

Loading