Persistent memory for OpenAI Codex CLI — powered by Supermemory
Codex forgets every session. codex-supermemory wires Supermemory into Codex CLI's
hooks system so your coding agent remembers your stack, preferences, prior decisions,
and the lessons learned across every project — automatically.
- 🧠 Automatic recall — relevant memories are injected into every prompt via the
UserPromptSubmithook. - 💾 Automatic capture — conversations are stored incrementally (every N turns) and
at session end via the
Stophook. - 🏷️ Project + user scoping — memories are tagged per-project and per-user so context never leaks across repos.
- 🔒 Privacy-aware — anything wrapped in
<private>...</private>is redacted before being sent to Supermemory. - ⚡ Zero-config install — one command sets up
~/.codex/config.tomland~/.codex/hooks.jsonfor you. - 🪶 No runtime deps in hooks — the hook scripts are pre-bundled with esbuild for fast cold starts.
- 🔧 Fallback skills — explicit
/supermemory-search,/supermemory-save, and/supermemory-forgetcommands available when hooks don't cover your use case.
-
Install the hooks:
npx codex-supermemory install
-
Start Codex CLI. On your first prompt, a browser window will open to authenticate with Supermemory automatically.
Alternatively, authenticate manually:
- Use
$supermemory-logininside Codex - Or set
export SUPERMEMORY_CODEX_API_KEY="sm_..."in your shell profile
- Use
-
That's it — memory is active.
Codex CLI supports a hooks system that lets external scripts run at specific
lifecycle events. codex-supermemory registers two hooks:
| Hook | Event | What it does |
|---|---|---|
recall |
UserPromptSubmit |
Captures new turns (every N prompts), then searches Supermemory for relevant memories and your profile, injecting them into the prompt as additionalContext. |
flush |
Stop |
Captures any remaining turns at session end so the final conversation turns are never lost. |
Incremental capture: Memories are saved every N turns (default: 3) during the session. This means memories from earlier in your session are immediately available for recall in the same session. The flush hook ensures any trailing turns are captured when the session ends.
The installer:
- Enables the
codex_hooksfeature flag in~/.codex/config.toml - Registers the hooks in
~/.codex/hooks.json - Copies pre-bundled hook scripts to
~/.codex/supermemory/ - Installs skills to
~/.codex/skills/
The hooks are tolerant: if Supermemory is unreachable, the API key is missing, or anything else fails, they exit cleanly without breaking your Codex session.
| Variable | Purpose |
|---|---|
SUPERMEMORY_CODEX_API_KEY |
Your Supermemory API key (browser auth is preferred). |
SUPERMEMORY_DEBUG |
Set to any truthy value to enable debug logging to ~/.codex-supermemory.log. |
Drop this file in to override defaults:
| Key | Type | Default | Description |
|---|---|---|---|
apiKey |
string |
— | API key (env var takes precedence, browser auth is preferred). |
similarityThreshold |
number |
0.6 |
Minimum similarity score for retrieved memories. |
maxMemories |
number |
5 |
Max memories injected per prompt. |
maxProfileItems |
number |
5 |
Max profile items considered. |
injectProfile |
boolean |
true |
Whether to fetch and inject the user profile. |
containerTagPrefix |
string |
"codex" |
Prefix for auto-generated container tags. |
userContainerTag |
string |
auto | Override the user container tag. |
projectContainerTag |
string |
auto (per-cwd) | Override the project container tag. |
filterPrompt |
string |
(sensible) | Filter prompt used by Supermemory's stateful filter. |
debug |
boolean |
false |
Enable debug logging. |
autoSaveEveryTurns |
number |
3 |
Save memories every N turns (incremental capture). |
signalExtraction |
boolean |
false |
Enable signal-based filtering (only capture turns with keywords like "prefer", "decided"). |
signalKeywords |
string[] |
(defaults) | Keywords that trigger signal extraction. |
signalTurnsBefore |
number |
3 |
Include N turns before a signal for context. |
User and project tags are auto-derived from your git config user.email and the
current working directory (both hashed) when not explicitly set.
When signalExtraction is enabled, only conversation turns containing signal keywords
(like "prefer", "decided", "remember", "bug", "fix") are captured. This reduces noise
but may miss some context. Disabled by default — all turns are captured.
npx codex-supermemory install # set up hooks + config + skills
npx codex-supermemory uninstall # remove hooks + config (keeps your memories)
npx codex-supermemory status # show current install statusThese Codex skills are available as explicit commands when you need more control:
| Skill | Usage | Description |
|---|---|---|
/supermemory-search |
/supermemory-search <query> |
Search memories manually. |
/supermemory-save |
/supermemory-save <content> |
Save a specific memory explicitly. |
/supermemory-forget |
/supermemory-forget <content> |
Remove a memory. |
/supermemory-login |
/supermemory-login |
Re-authenticate with Supermemory. |
Skills are fallback commands — the hooks handle most use cases automatically.
Anything wrapped in <private>...</private> is replaced with [REDACTED] before
being sent to Supermemory. Use this for secrets, tokens, or anything you'd rather
not have stored.
MIT