test(mcp): prove tracedecay_multi_str_replace behavior

[ScriptedAlchemy]

Summary

• Prove tracedecay_multi_str_replace through a production JSON-RPC tools/call, not call_tool_for_test.
• Lock the caller-visible result: preview diff and message, applied file bytes, idempotent replay, and refusals that leave the file unchanged.

Motivation

The tool is all-or-nothing and apply is bound to the preview expected_state. These tests send the arguments a client sends and assert the JSON-RPC payload and the bytes left on disk.

Changes

• crates/tracedecay/tests/mcp_suite/mcp_handler_test/multi_str_replace_behavior_test.rs exercises only this tool over the production server connection:
  • dry run writes nothing and returns the exact two-line diff
  • apply then replay of the same key keeps new-a / new-b and the same receipt; replay reports durable metadata on effect.payload, not a fresh edit body
  • a reused key with different replacements is an idempotency conflict (-32603, source_edit.idempotency_conflict) and does not rewrite the file
  • an earlier replacement that inserts a later match edits the original span; the preview hunk is fn keep() {} / fn target() {} / +fn target_renamed() {}
  • a miss, a duplicate match, overlapping ranges, a path outside the project, a malformed pair, a missing path, an apply without preview state, a stale expected_state, and a multi-byte miss all leave the original bytes in place

Test plan

• cargo test -p tracedecay --test mcp_suite --features test-transport mcp_handler_test::multi_str_replace_behavior_test -- --test-threads=1 reports 3 passed; 0 failed (350 filtered out)
• No production code changed
• Do not merge

  

[changeset-bot]

⚠️ No Changeset found

Latest commit: 4882a2f

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.