feat(ui): customizable between-turn statusline

[emal-avala]

Summary

Adds ui.statusline so operators can reshape (or suppress) the divider the REPL draws before each prompt after the first turn.

Config surface

[ui.statusline]
enabled  = true                                # default
template = "{model} · turn {turn} · ${cost}"   # optional override

Placeholders:

Token	Value
{model}	Current API model
{turn}	Completed turns this session
{tokens}	Cumulative token count
{cost}	Session cost in USD (4-digit precision)
{cwd}	Working directory
{session_id}	Active session id

• Unknown placeholders pass through verbatim — lets templates reference names this runtime doesn't recognize yet without silently dropping text.
• {{ / }} produce literal braces.
• Unterminated {… is preserved verbatim so the operator can spot the typo.
• enabled = false suppresses the divider entirely — useful for scripted sessions where the divider adds noise to captured output.

Defaults preserved

With no config or enabled = true, template = None the existing model · turn N · T tokens · $0.0000 line is unchanged byte-for-byte.

Tests (10 new)

Schema (3): default round-trip, TOML parse, disabled flag round-trip.

Template rendering (7):

• All six built-in placeholders substitute
• {{ / }} escape to literal braces
• Unknown placeholder passed through (forwards-compat)
• Unterminated { preserved
• Cost uses 4-digit precision
• No-placeholder string unchanged
• Stray } preserved

Full schema suite: 107 pass, 0 fail. Clippy clean under -D warnings. CLI crate builds cleanly with the repl.rs wiring.

Test plan

• cargo test -p agent-code-lib --lib config::schema
• cargo build -p agent-code
• cargo clippy --workspace --tests --no-deps -- -D warnings
• cargo fmt --all --check

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.