[liuchao-001]

Publish the current main as 1.1.1. Part of #29.

Fixes a 1.1.0 regression (#33): after write finish --publish, the repository cache's remote-tracking ref was not advanced and the 60-second fetch record stayed valid, so a sync on the same machine within the window returned the pre-publish snapshot. Found by OpenTag's client integration tests while bumping to 1.1.0.

Only the package version changes. Validation: frozen install, pnpm check, pnpm typecheck, pnpm test (128 tests), pnpm check:package all pass.

After merge: tag v1.1.1 on the merge commit and push the tag to publish. Dependents (OpenTag, First Tree) will bump directly to 1.1.1.

🤖 Generated with Claude Code

[liuchao-001]

Hold: merge #35 (concurrent first-run mkdir ~/.context-tree race, found by OpenTag #1287's two-subtree regression test) before this PR so 1.1.1 includes both #33 and #35. This PR only changes the version, so no rebase is needed.

[yuezengwu]

Exact-head release-preparation review of 1b86aaa01e67df2809e43b60884d0b8f3b28569e, tested together with actual main 4a09b1df62669e2b4946c8bd469850b3479d597c after #35 merged as requested. The integrated tree b90bab9b850ccf004d5e0cfe812b13b265184398 differs from current main only by package.json version1.1.0 to1.1.1, retaining both the published-ref fix and concurrent-directory fix. Frozen install (host-install scripts disabled), check, typecheck,130 tests across15 files and real-tarball package e2e pass on that integration tree. No code blocker found.

This approves the version-preparation PR. No tag was pushed and no explicit npm publication was run by this review.