UX pass: timing, honest summaries, dashboard-style inspect, richer banner

[DENNIS-CODES]

Summary

Scoped implementation of the terminal-polish feedback from the "before we release" review. This covers the achievable-now bucket of that feedback within the existing scrolling-output model. It deliberately does not build a live-redrawing TUI (ratatui), a web dashboard + auto-open browser, doctor --fix, an interactive arrow-key replay picker, or config profiles (profile create/use) — those are separate features and architectural decisions, not polish, and are worth their own scoping conversation rather than a rushed bolt-on here.

Also deliberately did not fabricate a "Health Score: 98/100" or per-callback "Latency" — there's no real formula behind a health score, and no legitimate timestamp to compute latency from (we never see when the STK push was initiated, only when the callback arrives).

doctor

• Every check now times itself and prints its duration, e.g. ✓ OAuth round trip (612ms)
• A conversational line ("Authenticating with Daraja...") precedes each check as it runs, instead of a batch of results dumped at the end
• Sandbox/callback reachability wording no longer reads as ambiguous when the response is a non-2xx status (e.g. 404) — the detail text now says explicitly that any response confirms the network path, not that the route serves anything
• Summary block shows real numbers (pass count, environment, total time) instead of a bare "N/M checks passed"

inspect

• Startup is now a small dashboard block (heavy rule, URL, waiting state) instead of a couple of loose lines
• Each callback prints as a labeled "Incoming STK Callback" card (status, amount, receipt, masked phone 254712••••78) ahead of the raw JSON, which is now dimmed underneath for anyone who wants the full payload
• A running tally (Totals so far: N payments · N callbacks · N errors) prints after every event via shared Arc<Mutex<Stats>> state, so the terminal has a sense of history instead of being a flat scrolling log

Banner

• The "M" mark stays; the dotted frame now also shows which environment you're pointed at (Environment: sandbox), highlighted in yellow when it's production since that's real money

Test plan

• cargo build --all-targets, cargo clippy --all-targets -- -D warnings, cargo fmt --all -- --check, cargo test all pass
• Verified locally: full doctor run against real (and deliberately broken) sandbox credentials — timing, progress lines, and summary all render correctly, including correct singular/plural grammar in the failure count
• Verified inspect against success/failure/malformed payloads: card layout, phone masking, and counters all increment correctly (including singular "1 payment" vs plural "2 payments")
• Confirmed replay --corrupt's diff highlighting still works on top of the new card layout
• Confirmed banner renders correctly for both sandbox and production environments

https://claude.ai/code/session_01BE4uE4wMaDgqN43zXvbqHH

---

Generated by Claude Code

[Copilot]

Pull request overview

This PR implements terminal UX polish across doctor, inspect, and the startup banner: timing/progress output for checks, a more “dashboard-like” callback inspector with per-event cards and running totals, and an environment indicator in the banner (with a production highlight). It also updates RUNNING.md to reflect the new output and behaviors.

Changes:

• Add per-check timings, live “doing this now” progress lines, clearer reachability wording, and a real summary block to doctor.
• Rework inspect startup + callback rendering into a dashboard/card layout, and introduce running stats printed after each event.
• Extend the startup banner to display the active environment (highlighting production), and update documentation examples accordingly.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/main.rs	Passes the configured environment into the banner printer.
src/commands/inspect.rs	Adds inspector dashboard output, callback cards, and shared running totals state.
src/commands/doctor.rs	Adds timing/progress output for checks and a richer end-of-run summary.
src/banner.rs	Prints an environment line and highlights production.
RUNNING.md	Updates docs and examples to match the new terminal output.

Comments suppressed due to low confidence (1)

src/commands/inspect.rs:133

• With callbacks counted per request, the stats update inside the recognized STK branch should not re-increment callbacks, and failures should contribute to errors (the PR description/test plan says counters increment for success/failure/malformed payloads). As written, Outcome::Failure callbacks don't increment errors.

            {
                let mut s = stats.lock().unwrap();
                s.callbacks += 1;
                if outcome == Outcome::Success {
                    s.payments += 1;
                }
            }

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.