Project management for Codex

How a task moves

1. Context: the goal, the references and the acceptance criteria are versioned text that Codex reads before it plans; each saved change is a new revision.
2. Plan: Codex writes the plan in the project, with phases, risks, dependencies and acceptance criteria for each task, and you approve it before production.
3. Work: Codex claims a task through the Spectra MCP server in its config.toml, works in your repository on your computer and reports with the files. A cycle runs up to five executors, each under its own lease.
4. Record: every file Codex uploads is stored unchanged, with its SHA-256, beside the project's versions, decisions and open points.
5. Review: you go through the round, what Codex reported and the files, and decide what the next round asks.

How Codex connects

The MCP endpoint below is public and the same for every project; what gives access is the token issued for your project. Add this to Codex's config.toml:

[mcp_servers.spectra]
url = "https://ukvjmdbfkvonkjlwpydh.supabase.co/functions/v1/customer-agent-connector"
bearer_token_env_var = "SPECTRA_EXECUTOR_TOKEN"

Codex sends the token it reads from SPECTRA_EXECUTOR_TOKEN as a bearer token: the file names the variable, never its value. A chat connection keeps its token in SPECTRA_AGENT_TOKEN, so the two can share one computer. Spectra issues the token per project, with an active plan or a trial released by Spectra; it is valid for 24 hours, 7 days or 30 days and can be revoked. Provider credentials never enter Spectra.

model = "gpt-6.1-sol"
model_reasoning_effort = "high"

Codex reads .codex/config.toml only for trusted projects; otherwise ~/.codex/config.toml. model and model_reasoning_effort start the executor on GPT-6.1 Sol at high effort; the server refuses a claim that declares less.

Execution on Opus 5.5 or GPT 6.1 Sol, which claim every task; the chat on Fable 5.1 or Astra answers report requests. In new projects the Laboratory gives tasks only to a connection that declares an executor model; the declaration comes from the client.

What is recorded

The plan and each of its revisions, the rounds, what Codex reports, your decisions, the checkpoints and the files. A file cannot be changed after upload and keeps its SHA-256; each is up to 8 MiB, with 48 files per task.

What Codex reports stays its declaration: Spectra records the report as sent and never sees your repository or your terminal.

What it is not

It is not application hosting, not a cloud runtime and not a proprietary AI model. Spectra does not run your code and has no cluster: the scale is your machine's, or that of the infrastructure your agent reaches. The agent connects with a per-project token, revocable and with an expiry; provider credentials never enter the platform.

What you need

A Spectra account and Codex on your computer. Writing the brief and preparing the project need no plan; connecting your agent and releasing rounds need an active plan or a trial released by Spectra.

Prices are on the plans page