mirror of https://github.com/openai/codex.git synced 2026-04-29 17:06:51 +00:00

Files

David de Regt eaf78e43f2 Add sorting/backwardsCursor to thread/list and new thread/turns/list api (#17305 )

To improve performance of UI loads from the app, add two main
improvements:
1. The `thread/list` api now gets a `sortDirection` request field and a
`backwardsCursor` to the response, which lets you paginate forwards and
backwards from a window. This lets you fetch the first few items to
display immediately while you paginate to fill in history, then can
paginate "backwards" on future loads to catch up with any changes since
the last UI load without a full reload of the entire data set.
2. Added a new `thread/turns/list` api which also has sortDirection and
backwardsCursor for the same behavior as `thread/list`, allowing you the
same small-fetch for immediate display followed by background fill-in
and resync catchup.

2026-04-17 11:49:02 -07:00

src

Add sorting/backwardsCursor to thread/list and new thread/turns/list api (#17305 )

2026-04-17 11:49:02 -07:00

Cargo.toml

ci: verify codex-rs Cargo manifests inherit workspace settings (#16353 )

2026-03-31 21:59:28 +00:00

README.md

Redirect debug client output to a file (#17234 )

2026-04-14 09:53:17 -07:00

README.md

WARNING: this code is mainly generated by Codex and should not be used in production

codex-debug-client

A tiny interactive client for codex app-server (protocol v2 only). It prints all JSON-RPC lines from the server and lets you send new turns as you type.

Usage

Start the app-server client (it will spawn codex app-server itself):

cargo run -p codex-debug-client -- \
  --codex-bin codex \
  --approval-policy on-request \
  --output-file /tmp/app-server-server-json.jsonl

You can resume a specific thread:

cargo run -p codex-debug-client -- --thread-id thr_123

CLI flags

--codex-bin <path>: path to the codex binary (default: codex).
-c, --config key=value: pass through --config overrides to codex.
--thread-id <id>: resume a thread instead of starting a new one.
--approval-policy <policy>: untrusted, on-failure (deprecated), on-request, never.
--auto-approve: auto-approve command/file-change approvals (default: decline).
--final-only: only show completed assistant messages and tool items.
--output-file <path>: write raw server JSONL to this file instead of stdout.
--model <name>: optional model override for thread start/resume.
--model-provider <name>: optional provider override.
--cwd <path>: optional working directory override.

Interactive commands

Type a line to send it as a new turn. Commands are prefixed with ::

:help show help
:new start a new thread
:resume <thread-id> resume a thread
:use <thread-id> switch active thread without resuming
:refresh-thread list available threads
:quit exit

The prompt shows the active thread id. Client messages (help, errors, approvals) print to stderr; raw server JSON prints to stdout so you can pipe/record it unless --final-only is set. Pass --output-file <path> to record raw server JSONL to a file instead of stdout.

Notes

The client performs the required initialize/initialized handshake.
It prints every server notification and response line as it arrives.
Approvals for item/commandExecution/requestApproval and item/fileChange/requestApproval are auto-responded to with decline unless --auto-approve is set.