CDX Manager

The right AI coding account, when you need it. CDX Manager keeps your Codex and Claude accounts in separate profiles, shows which ones have quota, and helps you pick the next usable session from your terminal.

When one account hits a limit mid-task, you should not have to guess which login still works or rebuild your context from memory. CDX shows the available accounts and can prepare a transcript-backed handoff to another session.

Get started in a minute

Install with npm (Node.js 18+ and Python 3.9+):

npm install -g cdx-manager
cdx add codex work
cdx add claude personal
cdx status
cdx next
cdx work

cdx next recommends a session and prints the launch command; cdx work opens your isolated work profile. CDX also supports Antigravity and local Ollama sessions. Prefer pipx or uv? See installation and setup.

What you can do

CDX checks authentication before interactive launches. Handoff keeps the source transcript in place and asks the receiving agent to verify the conversation and workspace before acting. No account switch or prompt submission happens just because you run cdx next.

Cache efficiency and fix rounds

cdx history --json records uncached input, cache reads, cache writes, output tokens, and usage.cache_hit_ratio beside the existing token totals. The ratio is cache reads divided by all input tokens, excluding output. Missing provider splits stay null, and text views show cache n/d instead of claiming a zero hit rate.

cdx stats --json adds cache_usage sums and an input-weighted cache_hit_ratio to each session, project (--by project), model (--by model), and total. cache_usage preserves unknown sums as null; existing numeric aggregate fields keep their previous behavior. Ratios include only runs with measured input and cache reads. Coverage is explicit in measured_runs and measured_input_tokens. run_types.cold and run_types.continued expose counts, sums, and ratios for comparison. Text views show values such as cache 72 % next to tokens.

Reuse a completed run's provider conversation for another fix round:

cdx run work --cwd /path/to/repo --prompt "Implement the change" --reasoning-effort high --json
cdx run --continue-run RUN_ID --cwd /path/to/repo --prompt "Fix the remaining test failure" --json

--continue-run resumes the exact Claude session or Codex thread and pins the previous run's recorded model and effort, even if the profile settings changed. The profile is inferred from the parent; an explicit session or provider must match. An explicit model or effort must also match. Runs and history record parent_run_id and conversation_id. Detached continuations use the cdx child supervisor with the same pins. Codex continuation usage is measured between exact transcript snapshots before and after the run, since exec resume can report conversation totals. If either boundary is unavailable, per-run usage stays unknown with usage_attribution: "continuation_usage_unavailable" in history instead of counting earlier turns again.

Conservative assumption: a continuation requires a finished Claude or Codex run, a locally available provider transcript, and recorded model and effort settings. Missing runs, profiles, expired sessions, and unavailable settings produce a clear error; the caller can retry without --continue-run for a cold run. --failover and --isolate-workspace cannot be combined with continuation. Capacity recovery and provider model fallback are disabled for continuations so they cannot silently switch model or account. Provider cache retention remains provider-controlled, so resuming does not guarantee a cache hit.

Codex model capacity recovery

cdx run (including --detach) automatically recovers from Codex's "Selected model is at capacity" failure. It retries the same prompt, profile, and model after 1 and 2 seconds, then tries the profile's fallback_model list in configured order, removing duplicates. Configure alternatives with cdx set work --fallback-model MODEL_A,MODEL_B.

After exhausting those models, CDX selects another ready Codex profile using the normal ranking, including profile priority. Disabled, logged-out, never, and ask-first profiles are excluded from automatic selection. To choose an ask-first profile explicitly, name it in a new command (cdx run NAME ...). Each successor uses --model if supplied, otherwise its own configured/default model, followed by its own fallback list. Each model gets at most three attempts; at most four profile switches are allowed. Settings are not changed. Each attempt retains the command's timeout; the backoff adds at most three seconds per model.

Recovery does not require --failover, which still controls account quota exhaustion separately. Other provider failures and timeouts do not trigger capacity recovery. Attempts restart the prompt in the same workspace; files from earlier attempts remain there. Each retry, model fallback, profile fallback, and exhausted budget is logged to stderr and persisted in recovery_events in the run registry and final JSON (cdx run-status / cdx run-report). Events include archived attempt output paths. Recovery keeps one run ID and reports the final effective model. If no eligible alternative remains or the profile switch budget is reached, the run fails with capacity_recovery_exhausted.

Documentation

• Installation and first steps — prerequisites, npm/pipx/uv, Windows, configuration.
• Features and handoff — routing, isolated accounts, automation, recovery.
• Sessions and notifications — alerts, launch settings, history.
• Command and JSON reference — commands, selection, usage fields, programmatic contracts.
• Operations — backup, disk cleanup, Windows notes, troubleshooting, release procedure.
• Technical overview — isolation, status sources, project and data layout.
• Tray companion testing — local development loops.

Contributions are welcome; see CONTRIBUTING.md. CDX Manager is MIT licensed.