Storybloq now works with Codex

Storybloq v1.2.0 is live. The headline change: native support for the Codex CLI. The same .story/ project memory that works under Claude Code now works under Codex too.

We argued in the three camps essay that the repository is the substrate for agentic engineering work, and the agent is just the runtime. v1.2.0 is what that looks like in practice.

What ships

One command for new users:

npm install -g @storybloq/storybloq && storybloq setup --client all

That installs the Storybloq skill in both Claude Code and Codex, and registers MCP with whichever clients are on your PATH. Hooks get wired up too: Claude gets PreCompact, SessionStart, and Stop; Codex gets SessionStart and Stop, and may ask you to approve them with /hooks on first use. If you only want one client, swap all for claude or codex.

Once installed, project context is one keystroke away in either tool. In Claude Code, /story. In Codex, $story. Both load the same data: project status, the last three handovers, the lessons digest, recent commits.

In a Storybloq project, all 43 MCP tools are exposed to both agents. Codex setup pre-approves the read-only tools so they run without a prompt; mutating tools still go through Codex approval. Same .story/ data, same dashboard, same handovers.

Why it matters

The two agents do not share a live session. Each one starts fresh, reads .story/, does its work, writes back. The substrate is shared. The runtime is not.

The Mac app shows the same tickets, the same issues, the same handovers regardless of which agent produced them. You can pick the agent that fits the work, not the agent your tooling is wired for.

If the repository is the substrate, the agent should be a choice you can revisit, not a commitment.