standup
You and your agents were busy all day. Doing what?
It’s 6:21pm, you have a dozen Claude Code sessions mid-flight, a handful of branches and no clear memory of the morning. How did we get here? Did we get anything done today? Manager asking for update. Client asking for status? You need to pick up your kid from acting classes in 1h. Standup goes through your work and your agents’ work and writes the summary for you.
Install in Claude Code
/plugin install standup --marketplace <owner>/<repo>
Not published yet. This command works once the repository is public. It reads files on your own disk and uploads nothing.
Codex, Cursor, Gemini CLI and opencode run the same two skills from your skills folder.
Or paste this into your agent
Read https://standup-site.pages.dev/llms.txt to learn what Standup reads and what it writes. Then install it for the agent you are. In Claude Code, run /plugin install standup --marketplace <owner>/<repo>. In any other agent, copy the eod-review and eow-review skills from that repository into your skills folder. When it is installed, run eod-review for today and show me the report.
This is Wednesday 11 November 2026, the day this page was built.
EOD Wed 2026-11-111
- Standup Site2 ✅ → scaffolded, homepage written, sun dial and wordmark in
- Switchboard Quota (E-080) 🔁3 → rc-quota merged, eight review findings fixed
- Switchboard Records ✅ → E-077 to E-080 marked shipped
- wcgw Starter ✅ → copywriting skill and a pre-commit copy review for new sites
Standup site is on a branch, not deployed.4
- The weekday is checked against the calendar before anything is written.
- Eight commits in three repos, grouped into four lines.
- 🔁 marks a merge and ✅ marks work finished here, so nobody reads a merge as new work.
- The last line says where the work stands. Built is not the same as shipped.
EOD Wed 2026-11-11
| switchboard | 4 |
|---|---|
| standup-site | 2 |
| wcgw-starter | 2 |
| output6 | 3.98M |
|---|---|
| fresh input6 | 13.98M |
| cache reads7 | 787.16M |
| total | 805.12M |
| opus | 3.30M output | $1,523 |
|---|---|---|
| fable | 680k output | $375 |
Estimated at API list prices≈ $1,8998
- Commits come from git, from every branch of every repo in one folder.
- Output and fresh input are what the models wrote and what they read for the first time.
- 98% of the tokens are cache reads, billed at about a tenth of the input price.
- An estimate at API list prices. On a subscription you pay none of it per token.
"What did you work on today?" is a hard question now.
You started three things before lunch. An agent finished two of them, and another ran in a worktree you forgot about. The work happened. Your memory of it did not.
It reads Claude Code, Codex, Grok, Gemini CLI, opencode and Cowork.
What shipped comes from git, so it covers the work whoever wrote it, you or any agent. Tokens and cost come from those six, including sessions Superconductor or Conductor ran in a separate worktree. Cursor and Antigravity keep no token counts on disk, so the report counts their sessions.
Work, consulting and side projects stay apart.
Point it at the folder that holds all your repos and one report covers every project. Name your groups, say work, consulting and personal, and each gets its own section with its own commits and cost. The update you send to one never mentions another.
You get four to six lines you can say out loud.
Commits from every branch of every repo you point it at, grouped into things a person would recognise. Paste it into Slack, or keep it as a catch-up with your own day.
On Friday, ask for the week.
eow-review covers Monday to Sunday, day by day. Monday's work is still there on Friday, which is more than most of us can say.
It also says what the agents cost.
Tokens are counted per folder, branch and day, dead ends included. The dollar figure is an estimate at API list prices, and the report labels it that way. On a subscription you pay none of it per token.