[Mikola Lysenko (mikolalysenko)]

LLM Description written by Claude Code:claude-opus-5-5

Sweep of the next audited, unswept rows of coverage/files.tsv in crates/socket-patch-cli/src/commands/: hosted_bundle.rs, hosted_unwind.rs, list.rs, lock_cli.rs, mod.rs and pypi_reinstall.rs (clean). No other open PR touches these files. This follows the architecture audit's per-file notes (discussion #1830, living document Part 4).

Changes

hosted_unwind.rs (+ rollback.rs)

• Deleted the dead HostedLegOutcome::unsupported field. Nothing ever wrote to it. Removing it also removes rollback's hosted_unsupported event loop and its exit-code conjunct (−11 lines in rollback.rs). The hosted_unsupported code appears nowhere else: not in the contract and not in any test.
• Merged the two flush_error.is_none() blocks into one. Warning order is unchanged.
• A dry run now skips the bun live_user_patch_keys read, since only the wet run uses it.
• The module doc now names scan's withdrawn-patch caller. Removed 3 bare issue references.

hosted_bundle.rs

• Bundle option fields are now moved with take() instead of cloned.
• The stdin schema in the module doc was missing policySha256; it is now listed.

list.rs

• Removed the Store as Source alias and its 6-line comment.
• Replaced the PatchRecord literal with ledgers::uuid_only_record.
• Replaced 3 RunWarning literals with RunWarning::new.
• Removed a stale comment about read_manifest.
• Inlined vendor_state_lenient, which had one caller and lived in mod.rs.
• Moved a misplaced fixture doc onto vendor_entry.
• Deleted ordering_is_deterministic_across_builds. It built the envelope twice from the same HashMap, so both builds iterate in the same order and the test can't fail.

mod.rs

• Moved mod user_patches up with the other module declarations.
• hosted_state_from_pins now uses uuid_only_record.
• Tests fixed (3):
  • an_uncovered_manifest_record_beside_a_ledger_is_ambiguous asserted only .contains("--mode").
  • a_malformed_manifest_is_not_a_hosted_project asserted only is_err().
  • a_hosted_pin_over_a_ledger_entry_is_ambiguous asserted only is_err().
  • All three now compare the whole usage error, including the named purl. The expected strings are written out as literals.
  • hosted_pins_beside_a_ledger_are_ambiguous now does the same.
  • Red→green: I temporarily changed the (true, true) arm to return "<root> --mode". The old asserts still passed. The new tests failed (an_uncovered_… and a_malformed_…), and they pass again after the revert.

lock_cli.rs

• Inlined emit, held_hint and held_message into their only callers.
• The human branch of acquire_or_emit now prints through format_lock_error, so it shares one formatter with the sites that build their own envelope.
• Removed the 24-line lifecycle essay; core apply_lock already documents the lifecycle.
• Removed history notes about the "historical" try-once, --break-lock and the acquire-mkdirs change.
• Fixed the doc overclaim on error_envelope_propagates_dry_run_and_command.
• Tests:
  • The 3 held_message tests became one, lock_failure_renders_the_waited_budget.
  • Deleted held_hint_depends_on_whether_we_already_waited, a duplicate of format_lock_error_matches_acquire_or_emit.
  • waiting_message is still a separate function, because its status-line text is tested only through it.

Not done here, because each would reach files that #1698 touches or would change output:

• acquire_or_emit(&GlobalArgs) (T09/Acquire apply.lock through one lock_cli helper that takes &GlobalArgs; the sequence is open-coded 9 times and renders lock_held 3 ways #1858)
• the patch_server_origins forwarder (callers in get.rs / vendor.rs)
• global_scope_flag vs ui::global_scope_arg (they print different text)
• hosted_bundle's json! error and its hand-built ApiClient (both would change output or behavior)

Size

git diff --stat: 6 files, +160 / −300.

• Production: about −105
• Tests: about −35

Behavior change

None. Output bytes, JSON, exit codes and error codes are unchanged. Removing the dead hosted_unsupported branch changes nothing, because that list was always empty.

Verification

• cargo clippy --workspace --all-features -- -D warnings: clean.
• cargo test -p socket-patch-cli --lib: 936 passed.
• --test rollback, remove, in_process_rollback_hosted, in_process_rollback_vendored, scan_get_mode_from_project_state, e2e_safety_lock, cli_parse_list, command_module_layering and hosted_memory_parity: all pass.
• No platform-specific code was touched.

/code-review high

Six findings. I fixed three:

• acquire_or_emit now reuses format_lock_error.
• The lock_failure doc no longer overclaims.
• The merged test is renamed.

I did not fix three:

• list --json doesn't report an unreadable vendor ledger in warnings[]. This is pre-existing; I only moved the code. Fixing it would change JSON output, so it belongs in its own issue.
• The deleted determinism test "wasn't a tautology." It was: it reused one manifest instance, so both builds iterate the same HashMap.
• The mode tests copy the production message. That is deliberate: expected values are written independently of the code under test.

Risk

Low. The changes are mechanical inlining and deletions, and the touched paths are covered by the suites above.

🤖 Generated with Claude Code

https://claude.ai/code/session_011f9aeQq8rjUYApktQhf6WR

---

Note

Low Risk
Mechanical refactors and deletion of dead rollback paths in CLI listing, lock, and hosted helpers; PR asserts output and exit codes stay the same aside from never-fired hosted_unsupported.

Overview
Architecture-audit cleanup across several socket-patch-cli command modules: mostly doc trims, shared helpers, and dead-code removal with no intended user-visible behavior change.

Hosted unwind and rollback drop the never-populated HostedLegOutcome::unsupported field and rollback’s hosted_unsupported JSON events / exit-code check. Hosted restore dry runs skip the bun live_user_patch_keys read; post-restore warning logic is merged into one flush_error-gated block.

List and mod centralize uuid-only patch stubs via ledgers::uuid_only_record, use RunWarning::new, rename ledger Store usage (no Source alias), and inline vendor-ledger lenient loading into list (stderr warning unchanged; --json still omits that warning). Mode-resolution tests now assert full --mode usage strings instead of partial checks.

Lock CLI inlines lock-failure formatting so human-mode acquire_or_emit goes through format_lock_error with lock_failure; redundant tests and long historical comments are removed.

Hosted bundle documents policySha256 on stdin and moves optional bundle fields into scan options with take() instead of cloning.

^{Reviewed by Cursor Bugbot for commit 1d557d4. Configure here.}

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

BugBot review

---

Generated by Claude Code

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 1d557d4. Configure here.}

[Mikola Lysenko (mikolalysenko)]

Burn-down agent: labeled Ready for review at 1d557d4. CI 82/82 green on this head; mergeable, no conflicts. Bugbot reviewed 1d557d4: no findings; no open review threads. Slack announcement not sent this run (Slack posting unavailable).

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

Burn-down agent: labeled Ready for review at 1d557d417a6c10181fe2e9606eb9e330e3a17ec1.

• CI: 32/32 run checks green (50 skipped by path filters); no main-wide failures.
• Bugbot: reviewed 1d557d417a, no new issues; no unresolved review threads.
• Mergeable: yes, no conflicts; waiting only on human approval.
• Slack: not announced (Slack send unavailable in this run); the next run retries.

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

Final review brief: approver-ready

What it does: It is a behavior-preserving cleanup of five CLI command modules: it inlines single-use helpers, dedupes tests and trims docs. It also deletes the HostedLegOutcome::unsupported field. Nothing on main ever writes that field, so the hosted_unsupported event loop and its exit-code check in rollback.rs could never run.

Risk: low. Every user-visible string, error code, exit code and JSON field compares identical to main. The only runtime difference is that a dry-run unwind skips one read-only lockfile scan whose result it never used.

Look here

• hosted_unwind.rs:77: the dry run now skips live_before. :123-165:`` the two post-flush blocks are merged, and their warnings still come out in the same order.
• rollback.rs:2226: the dead hosted_unsupported field and its exit-code check are removed.
• lock_cli.rs:19-41: human output now goes through format_lock_error, which prints byte-identical text. The trimmed lifecycle docs are still in socket_patch_core::patch::apply_lock.
• list.rs:411-423: the lenient vendor-ledger read is inlined, with the same warning and the same silent gating.
• hosted_bundle.rs:125-148:``` .clone() became `.take()` on fields `build_input` never reads.

Verified

• I read the diff line by line against main.
• cargo test -p socket-patch-cli --lib: 936/936 pass.
• Integration suites: in_process_rollback_hosted 41, in_process_rollback_all_ecosystems 28, e2e_safety_lock 34, cli_parse_list 69, command_module_layering 3, hosted_memory_engine 49, all passing.
• Deleted tests: what they checked is still covered by remaining tests, or the test could never fail.
• CI: 32 success, 50 skipped, 0 failing.
• Bugbot found no issues on 1d557d4. There are no open threads, and CHANGELOG.md is untouched.

Changes I made: none.

Open questions (non-blocking)

• Commit 09a66fa is empty and names pypi_reinstall, which the PR doesn't touch. The squash merge hides this.
• The PR title doesn't mention the rollback.rs edit.

Auto-merge is armed, so approving sends this straight to the merge queue.

---

Generated by Claude Code