Blog
How we run a fleet of coding agents, what we built to steer them, and what we learned along the way.
The software factory
Codecast turns what your product reports, from errors and analytics to failing checks and evals, into a small set of problems, has a flow your team owns prove and fix each one, and brings you one card per change. Shipped fixes are watched, and anything that comes back reopens. Here is the whole loop on our own product, with one change from error to card.

5 min readYour coding agent can check the page in your own Chrome
Codecast lets a coding agent open pages in the Chrome you already use, with your logins, in a background tab of its own. Its screenshots and the errors the page logged come back into the conversation, so you see what it saw without switching windows.
4 min readYour old Claude Code sessions, on the day you install Codecast
By the time most people try Codecast, they have months of Claude Code sessions sitting in files on their laptop. When you sign in on a machine, Codecast now uploads all of them in the background. The ones from the past week go in your inbox. The rest are in search and open in full, from the first day.
5 min readAnswering Claude Code from your phone
You start Claude Code in a terminal on your laptop, give it a job, and walk away. Codecast's iPhone app shows you that session, lets you answer its questions and approve its edits, and sends your next message to the same terminal, as if you had typed it there.
5 min readClaude Code, Cursor and Codex on the same project
Plenty of people run more than one coding agent on a project: Cursor for one piece, Claude Code for another, Codex for a second look. Codecast puts the sessions from each in one list, shows what each one said and did, and can move a conversation from Claude Code to Codex so the second agent picks up with the whole history.
6 min readFinding, naming and reopening your Claude Code sessions
Claude Code keeps every session as a file on the machine that ran it, named by a random id. Codecast collects those sessions, and the ones from Codex, Cursor and Gemini, from every computer you use into one list. You can search what was said in them, give them names, read the whole transcript with the code it changed, and send the next message to any of them.
5 min readUndo, with a history you can see
Running several agents means a lot of small moves: pin this one, file that one as done, rename a third, stash a fourth. In Codecast every one of them can be taken back with ⌘Z, and ⌘⌥Z opens the list of what you did in this window, with a button to go back to any point in it.
5 min readThe chart in the reply
Some findings are a shape: sixteen numbers where four stand out, five versions of one line of UI. In Codecast an agent can put that shape in its reply as a chart, a table or a mockup, drawn right in the conversation, so you see it before you read the paragraph explaining it.
6 min readThe questions your agents save for you
Run several agents and each one will, sooner or later, need you to choose. Codecast lets an agent write that choice up as a card, put it in one queue with every other agent's, and wait. You answer the lot in one sitting, and each agent picks up from your answer.
6 min readThe task your agent reports to
A chat thread is a bad place to keep track of work. Codecast gives every piece of work a page of its own, where the agents working on it leave their notes, their evidence and their status, and where a plan gathers the tasks that belong together.
14 min readCodecast in the cloud
Every cloud feature, end to end: a host in your own AWS account that starts from your uncommitted work, carries your agent config and logins, mirrors its edits back to your laptop, takes whole batches of running sessions mid-flight, and sleeps when nothing is happening.
6 min readThe tree at the end of every turn
A transcript shows the edits an agent made with its editing tools. It never sees what a shell command, a formatter or a person changed. Codecast now records the whole working tree at the end of every turn, as git commits you can diff and rewind to.
4 min readThe codecast field manual: running a company of agents
The org your agents report into, the inbox that sorts them by who acts next, moving a live conversation from Claude to Codex without losing a word, and ten things only codecast does. Thirteen chapters, with the real product on every page.
7 min readFewer, bigger jumps
When you work with coding agents, tokens are cheap and your attention is not. Ask for a leap that lands near the goal, close the gap in shrinking corrections, and spend the time it buys on the next leap.
5 min readOne repository, twenty checkouts
Running several agents at once is easy until two of them edit the same file. Codecast gives each one its own worktree, with the env files, dependencies and a port of its own, from one switch when you start the session.
6 min readWhat your team sees
Sharing your Claude Code sessions with a team is a great idea right up to the session where you pasted a customer's data. Codecast answers with two dials, a preview before every share, and one settings page that shows the whole picture.
6 min readThe pull request that knows its sessions
When agents write most of the code, a pull request is the end of a conversation you were not in. Codecast keeps the two attached: every PR carries its checks, its reviews, and the sessions that made it, and a review can wake the agent that owns it.
6 min readAgents that talk to each other
A Claude Code agent shipped a change that silently stalled 27 transcripts. A Codex agent in the same checkout found it, fixed it, and told the first one. No human relayed a word.
5 min readA URL for everything your agent makes
Reports, dashboards, design proposals — agents produce them daily, and chat transcripts bury them. Codecast turns a file into a live page with versions, comments, and a link you can actually send.
5 min readThis post wrote itself (on a schedule)
Codecast triggers run full agent sessions on a timer. The proof is this blog: last week's post and this one were both written, unattended, by runs of the same weekly trigger.
5 min readYour agents forget. Your team doesn't have to.
Every agent session is a problem being solved out loud, and then the terminal closes. Codecast keeps the record searchable — so the next agent, or the next person, starts from the answer.
5 min readAn inbox for your agents
Agents don't fail loudly. They finish, or stall, and wait for you to notice. The inbox makes the waiting visible — every session, every machine, sorted by who needs you.
6 min readgit blame for AI agents
When an agent writes the line, the author column goes blank. Codecast's blame fills it back in, with the conversation that wrote it.