ReMe/integrations/codex

ReMe memory for Codex

中文说明

Start a ReMe service, then install the ReMe plugin in Codex. This is the supported integration for this host. The plugin bundles automatic recall, completed-turn recording, MCP tools, and the reme-memory skill in one installation. ReMe manages durable memory and consolidation in your workspace.

The plugin and its marketplace are maintained in integrations/codex/ and installed from this checkout. Keep the ReMe service running while using the plugin.

Start ReMe → Install the ReMe plugin → Use memory in Codex

Requirements

• Python 3.11+ available as python3 in the host's environment, with FastMCP >= 3.4.2 (included with ReMe).
• Codex CLI 0.145.0 or a compatible client supporting local plugins, command hooks, and hook trust review.
• A graphical host supporting the openai/settings MCP extension for the native settings UI. CLI 0.145.0 supports hooks and tools but does not render this UI.
• A running ReMe HTTP service exposing search, auto_memory, and health_check, with MCP enabled.
• A working model configuration on the ReMe server for memory extraction. Default BM25 recall does not need embeddings.

1. Start ReMe

pip install "reme-ai[core]"
reme start workspace_dir=/absolute/path/to/workspace service.backend=http

The default endpoint is http://127.0.0.1:2333, with MCP at /mcp. Keep the unauthenticated HTTP service on loopback or behind a protected proxy. Different workspaces isolate different memory scopes; sharing one workspace between hosts is intentional cross-agent recall.

2. Install the plugin

Register this repository-local marketplace, then install its plugin:

codex plugin marketplace add /absolute/path/to/ReMe/integrations/codex
codex plugin add reme@reme-codex

Start Codex and review and trust the installed plugin's hooks in /hooks, then open a new conversation. The installation loads the bundled hooks, MCP connection, and skill together. After plugin updates, reinstall it and review any changed hook definitions.

3. Use memory in Codex

In a new conversation, ask “Check whether ReMe is available.” The installed plugin's tools check the running service; /mcp shows the plugin's connection status. Then use Codex normally: the plugin recalls relevant memory before prompts and records completed turns in batches of five. For an explicit query, ask “What did we decide about Project Juniper? Cite the memory sources.”

Automatic recording sends completed user/assistant text to the configured ReMe service. Set auto_memory to false for recall-only use.

Configure through MCP settings

Open the installed ReMe MCP server's settings in a host that supports openai/settings. The native form exposes the service address, automatic recall and memory switches, batch size, recall limits, timeouts, and timezone. Save changes, then use Check connection to call health_check on the saved address without sending conversation content. The settings remain available when ReMe is offline, so a wrong address can be corrected there.

Settings are atomically saved to ${CODEX_HOME:-$HOME/.codex}/reme/config.json, outside the plugin cache. The next MCP tool call and hook invocation use the new values without a reconnect or reinstall. An in-flight call keeps its original endpoint and settings. Disabling auto_memory pauses capture and queued retries; explicit memory tools remain available.

Option	Default	Meaning
mcp_url	http://127.0.0.1:2333/mcp	MCP address for both explicit tools and automatic hooks; custom paths are supported
auto_recall	true	Search before a user prompt
auto_memory	true	Queue and submit completed turns; false also disables retries
recall_limit	5	Search result limit
recall_min_score	0	Minimum recall score
recall_timeout	5	Foreground MCP timeout in seconds, at most 10
request_timeout	600	Memory-write and MCP request timeout in seconds, at most 600
memory_interval	5	Completed turns per batch; use 1 for immediate per-turn writes
shutdown_timeout	2	Total best-effort exit drain budget in seconds, at most 2
context_max_chars	8000	Maximum recalled payload characters
timezone	Asia/Shanghai	Daily batch timezone; match the ReMe workspace

The JSON file remains user-owned and can also be edited directly; only overrides are required. See config.example.json for a complete example. Remove a field from the file to restore its default. Unknown fields and invalid values fail validation; hooks skip work and log hook_failed instead of falling back to another service. UI updates reject invalid values without replacing the saved file.

Existing configurations keep working. The deprecated api_url field is accepted on read but is no longer used and is removed on the next UI save. All operations use mcp_url. Standard /mcp endpoints retain existing queue and receipt identities. Pending queues from a custom legacy api_url remain untouched: they are not automatically retargeted. Address changes never forward old pending conversations to the new service; switch back to the original endpoint to retry its queue.

Both the host-managed MCP server and short-lived hooks require FastMCP in the host's python3 environment. If ReMe runs elsewhere, install it with python3 -m pip install "fastmcp>=3.4.2". The plugin does not launch ReMe. Its native form follows the pinned OpenAI MCP Extensions settings contract. Clients without that extension can still use tools and hooks, but have no native settings form.

Lifecycle and failure behavior

• UserPromptSubmit calls search synchronously with a short timeout. Results are bounded and wrapped in <reme-context> as untrusted historical evidence. A failure does not reject the prompt.
• Synchronous Stop hooks extract only the completed user/final-assistant text from the hook's local transcript, then submit auto_memory batches through MCP. Tool output, reasoning, and injected ReMe context are excluded. The server does not need access to host files.
• Session and message IDs are stable and host-scoped. Repeated Stops are deduplicated. Per-session process locks serialize writes; batches never mix dates. Subagent events and hook continuations are excluded. No extra background daemon, package download, or ReMe process is started.
• Queues and acknowledgement receipts live under the host's reme/queue/, separate from the plugin cache and namespaced by endpoint. A failed or timed-out write stays queued. SessionStart retries pending work synchronously; SessionEnd attempts a short drain.
• Codex 0.145.0 skips hooks marked async, so this adapter deliberately uses synchronous hooks. A due memory batch or a startup retry can delay completion by up to request_timeout. Claude Code uses its own native asynchronous adapter.
• Shutdown and delivery are best-effort. A host killed before the Stop hook captures a turn can lose that turn. A lost MCP acknowledgement can cause a retry; stable message IDs prevent duplicate source messages, but memory extraction is at least once, not exactly once. Short residual batches survive for the next session when shutdown has insufficient time.
• Queue files contain source conversation text; retain them until delivery completes. Logs in reme/hooks.log contain event/error types only. Disabling auto_memory retains pending files.
• Daily auto_dream consolidation remains owned by the ReMe service. Unlike OpenClaw's long-lived Gateway adapter, these short-lived hook processes do not run their own scheduler.

Verify and troubleshoot

1. In Codex, ask the installed ReMe plugin to check the connection; confirm its status in /mcp.
2. For a quick test, set memory_interval to 1, then start a new conversation and ask the agent to remember a synthetic fact, such as “Project Juniper reviews are on Thursday.”
3. Wait for the write and inspect reme/hooks.log for memory_saved and the ReMe workspace's daily/.
4. Start a separate conversation and ask for that fact. Confirm recall is from ReMe, with source paths.

The installed plugin provides both automatic memory and explicit search/read and health checks. If its tools are missing, check that the plugin is installed and enabled, the service is running, and service.jobs exposes the required jobs. If its hooks do not load, use a host version supporting the plugin requirements and reinstall or reload the plugin.

Capture currently reads Codex rollout JSONL event_msg records (user_message and final agent_message). The hook requires transcript_path and last_assistant_message. Missing or unsupported transcripts are skipped rather than guessed. Codex transcript formats are not a stable public interface. Check capture_skipped or hook_failed events when recall works but recording does not.

In Codex 0.145.0, an unattended codex exec MCP call can return user cancelled MCP tool call when approval cannot be collected. Verify explicit tools in an interactive session and approve the specific call. Trusted automatic hooks act as MCP clients themselves; they do not request per-call model tool approval.

Source and validation

Run pytest tests/unit/test_coding_agent_plugins.py tests/unit/test_codex_plugin_mcp.py -v from the ReMe checkout. Tests isolate host state and mock service calls. Native hook behavior is described in the Codex hook reference. ReMe is developed at agentscope-ai/ReMe under Apache-2.0.