catch-report

1# catch-report
2 
3The product claim is two brains: one does the work, another checks it. `sno catch-report` puts numbers on that
4claim from records already on this machine. It changes nothing.
5 
6- **Mutual review** is every review the `peer-review` skill recorded: a different model read the work and
7  reported problems. `sno catch-report run` prints those numbers.
8- **Self-check** is an agent finding its own error inside one conversation. `sno catch-report self` reads the
9  agents' own conversations (`~/.claude/projects`, `~/.codex/sessions`), picks the moments where an agent
10  reacted to a failing command or test, or admitted a mistake, and asks that agent's own model (claude
11  judges claude conversations, codex judges codex conversations) whether the agent found the error itself or
12  was told. It then prints the two side by side.
13 
14Prerequisites: bash 4+, GNU date, find and grep, and jq. `self` also needs the `claude` and/or `codex` command
15line, logged in.
16 
17## Use
18 
191. To show the user what the two brains caught, run `sno catch-report brief --since 7d` (or `12h`, `30m`, `30d`, or a
20   date). It measures both and prints one page: a headline with the problems the second model caught and the moments in which the agents caught their own mistakes (kept apart, never added together), the second brain's
21   numbers with its biggest catches quoted by title, the agents' own catches in their own words, and how to read
22…