feat: add --format=json to seven step + hook commands

[max-sixty]

Extends structured-output coverage beyond list/switch/remove/merge to the remaining commands where JSON is useful for scripting and Claude Code integration: step rebase, step push, step commit, step squash, step relocate, step copy-ignored, and hook show.

JSON shapes follow the existing pattern (additive on stdout; human prose stays on stderr) and use stable snake_case outcome discriminators where the result is one of several variants.

JSON shapes

Command	Payload
step rebase	{target, outcome: "rebased"|"fast_forwarded"|"up_to_date"}
step push	{target, outcome: "fast_forwarded"|"up_to_date"|"merge_commit", commits, merge_sha?}
step commit	{commit, message, stage_mode} (resolved mode, not raw flag)
step squash	{outcome: "squashed"|"no_commits_ahead"|"already_single_commit"|"no_net_changes", commit?, message?, stage_mode?, target?}
step relocate	{dry_run, entries: [{branch, from, to}], skipped: [{branch, reason}]}
step copy-ignored	{outcome, dry_run, from, to, entries: [{path, kind}], files, bytes}
hook show	[{type, source, name, template, needs_approval, expanded?}]

Notable refactors

• RebaseResult::Rebased now carries {target, fast_forward}.
• SquashResult::Squashed now carries {sha, message, stage_mode}.
• CommitOutcome returned from commit_staged_changes and CommitOptions::commit carries {sha, message, stage_mode} — the resolved mode that was actually applied (CLI flag merged with config defaults), not the raw Option<StageMode> from clap. Without this, --format=json would emit null whenever the user didn't pass --stage.
• handle_push / handle_no_ff_merge return PushResult { target, commit_count, outcome } with outcome variants FastForwarded / UpToDate / MergeCommit { merge_sha }. The merge-commit SHA was previously discarded.
• relocate::GatherResult carries template_error_branches (was: opaque count). Text mode already showed these in stderr; JSON now surfaces them as skipped entries with reason: "template_error" so automation can detect a broken worktree-path config rather than reading entries: [] as success.

Behavioral guards

--show-prompt is rejected when combined with --format=json for step commit / step squash — show-prompt emits raw LLM-prompt text that would corrupt a JSON consumer's stdout.

Reviewed by Codex

Two rounds. Round 1 caught the relocate template-error case (fixed in this PR). Round 2 reported no remaining correctness issues.

Tests

13 new integration tests covering the JSON shapes (parsed into serde_json::Value and asserted against concrete keys). Full suite: 1604 / 1604 passing locally.

This was written by Claude Code on behalf of @max-sixty

[worktrunk-bot]

A few observations, nothing blocking.

Display abbreviation regression in commit_hash. The two display sites in src/commands/commit.rs and src/commands/step_commands.rs switched from git rev-parse --short HEAD to a fixed 7-char slice of the full SHA. --short honors core.abbrev and auto-extends for ambiguous prefixes; the slice does neither. Default-config users see no change, but anyone with a custom core.abbrev or a repo with 7-char collisions will see a quieter, possibly-ambiguous hash in Committed changes @ … / Squashed @ …. Cheapest preserve-behavior fix is keeping a separate git rev-parse --short HEAD call alongside the full one. Snapshots redact the hash, so the change isn't caught by tests.

--format flag descriptions are thinner than the existing pattern. wt switch/wt remove/wt merge each include a one-liner describing what the JSON contains (e.g. "JSON prints structured result to stdout after removal completes"). The new commands all use Output format (text, json) only. Worth a brief sentence per command so the shape isn't only discoverable via the PR body.

Missing test for step push --no-ff --format=json. fast_forwarded and up_to_date are covered, but the merge_commit outcome (with merge_sha) — the new field that was previously discarded — isn't asserted anywhere. That's the one variant most worth pinning down.

Heads-up on overlap with #2557. Both PRs touch src/cli/step.rs and src/commands/commit.rs, and #2557 hides --show-prompt while this PR adds a runtime guard against --show-prompt --format=json. The two are compatible, but landing them in either order will need a small rebase — worth noting which goes first.

[worktrunk-bot]

CI fails to compile across all platforms (lint, feature-check, code-coverage, test linux/macos/windows, affected tests). Single error, one site:

error[E0308]: mismatched types
   --> src/commands/hook_commands.rs:413:13
413 |             project_config.as_ref(),
    |             ^^^^^^^^^^^^^^^^^^^^^^^ expected `Option<&ProjectConfig>`, found `Option<&&ProjectConfig>`

LoadedConfigs::project is already Option<&ProjectConfig> (see src/config/loaded.rs:48), so .as_ref() adds an unwanted second &. Drop it. Happy to push the one-character fix if you'd like — otherwise the inline suggestion below applies.

[worktrunk-bot]

Display-abbreviation regression flagged previously is fixed in commit.rs and step_commands.rs, and test_step_push_no_ff_json pins down the new merge_sha field.

One same-pattern follow-up outside the diff but in the same file the refactor touches: src/commands/worktree/push.rs:444 still uses &merge_sha[..merge_sha.len().min(7)] for the Merged to @ <hash> line on --no-ff — same core.abbrev / ambiguous-prefix issue as the two sites just fixed. Pre-existing, so not a regression introduced here; happy to push a one-line fix as a follow-up if you'd like.

Lint is failing on a lychee 502 to github.com/max-sixty/worktrunk/tree/main/dev/opencode-plugin.ts — transient GitHub-side flake, should clear on rerun.

[worktrunk-bot]

CI is finished — required tests all pass. Two non-required checks need a note:

codecov/patch (88.07%) appears to be a false positive. The lines codecov reports as missed are actually covered in the cobertura.xml the workflow uploaded. Spot check from the code-coverage-report artifact for this same HEAD:

Line	Codecov hits	Local cobertura hits
src/main.rs:216 (if format == SwitchFormat::Json && (args.show_prompt …))	0	25
src/main.rs:217 (the bail! arm)	0	1
src/main.rs:219 (the step_commit call)	0	24
src/main.rs:165–166 (HookCommand::Show arm)	0	17

This pattern holds across all the lines codecov flags in main.rs / step_commands.rs / hook_commands.rs. The codecov totals match (96.68%, 46780/48385 — same as local), but the per-line hits_head for the new lines is wrong. There were four uploads against this commit and the patch check ran ~30s after the first; looks like a multi-upload merge race on codecov's side, not a real gap.

lint failed on a lychee 502 to github.com/max-sixty/worktrunk/tree/main/dev/opencode-plugin.ts — already noted, transient.

Approval stands — the underlying coverage is fine. Worth a rerun if you want green checks before merging.