Stale-path relocation, index prune, derived alias (v1.0.152)

.claude/CLAUDE.md

@@ -22,6 +22,7 @@ Add symbol-aware reference lookups to codesearch via `find_impact` MCP tool. Ret
 - **`-with-csharp` release variants** — 6 release archives (3 plain + 3 with helper)
 - **Gated integration test** — `csharp_helper_integration` cargo feature for full-pipeline testing
 - **CI** — separate `csharp-integration-tests` job in `.github/workflows/ci.yml`
+- **Stale-path resilience + derived alias** — moved/renamed indexed folders no longer crash serve: `git_remote` captured at registration, `reconcile_all_paths()` best-effort relocates by matching `remote.origin.url` (bounded depth, `CODESEARCH_RELOCATE_MAX_DEPTH`, default 3) else warn+skip; `codesearch index prune` for manual cleanup. The `--alias` flag was removed (alias always = directory name). `ReposConfig::reconcile()` hardens hand-edited `repos.json` on load. See AGENTS.md for details.
 
 ## Architecture
 

.claude/commands/merge.md

@@ -80,9 +80,11 @@ release — tagging happens only in `/release`.
      `PR=$(gh pr view --json number --jq .number)`.
 
 8. **Auto-merge after CI**
-   - `gh pr merge "$PR" --auto --merge` so the PR lands automatically once required checks pass.
+   - This repo **disallows merge commits** — always use `--squash`. NEVER `--merge`
+     (it fails with "Merge commits are not allowed on this repository").
+   - `gh pr merge "$PR" --auto --squash` so the PR lands automatically once required checks pass.
    - If auto-merge is not enabled on the repo (command errors), fall back: poll
-     `gh pr checks "$PR" --watch`, then `gh pr merge "$PR" --merge` once green.
+     `gh pr checks "$PR" --watch`, then `gh pr merge "$PR" --squash` once green.
 
 ## Report
 Branch, pending release version, doc updates made, PR URL, and merge status

.claude/commands/release.md

@@ -40,9 +40,10 @@ to `develop`). Then **wait for the develop PR to actually merge** (auto-merge wa
      matching history (e.g. `Release v1.0.142 — serve responsive during warmup`).
    - Body ends with: `🤖 Generated with [Claude Code](https://claude.com/claude-code)`.
    - Capture the PR number: `RELEASE_PR=$(gh pr view develop --json number --jq .number)`.
-4. `gh pr merge "$RELEASE_PR" --auto --merge`. Wait until `state` is
+4. This repo **disallows merge commits** — always use `--squash`, never `--merge`.
+   `gh pr merge "$RELEASE_PR" --auto --squash`. Wait until `state` is
    `MERGED` (poll as in Part 1). If auto-merge is unavailable, `gh pr checks "$RELEASE_PR" --watch`
-   then `gh pr merge "$RELEASE_PR" --merge`. If CI fails, STOP.
+   then `gh pr merge "$RELEASE_PR" --squash`. If CI fails, STOP.
 
 ## Part 3 — tag the release
 1. `git fetch origin --tags && git checkout master && git pull --ff-only origin master`.

AGENTS.md

@@ -26,7 +26,10 @@ Add symbol-aware reference lookups to codesearch via `find_impact` MCP tool. Ret
 - **Gated integration test** — `csharp_helper_integration` cargo feature for full-pipeline testing
 - **CI** — separate `csharp-integration-tests` job in `.github/workflows/ci.yml`
 - **Sequential phase-2 startup** — Phase 1 warms repos sequentially, Phase 2 runs gated C# SCIP rebuilds ordered by `last_changed_unix` under `Semaphore(concurrency)` via `CSHARP_SCIP_CONCURRENCY` env (default **2**, clamp [1,4])
-- **`repos_meta` tracking** — `RepoMeta` (last_changed_unix, last_scip_indexed_unix) persisted in `repos.json` with debounced save (10s window)
+- **`repos_meta` tracking** — `RepoMeta` (last_changed_unix, last_scip_indexed_unix, git_remote) persisted in `repos.json` with debounced save (10s window)
+- **Stale-path resilience** — a renamed/moved indexed folder no longer crashes serve. `git_remote` (`remote.origin.url`) is captured at registration; on startup `ServeState::reconcile_all_paths()` best-effort relocates a missing repo by scanning the nearest existing ancestor (bounded depth, env `CODESEARCH_RELOCATE_MAX_DEPTH`, default 3) for a git root with a matching remote — exactly one match rewrites `repos.json`, otherwise warn + skip. Phase-2/Phase-3 also guard `path.exists()`. Manual cleanup via **`codesearch index prune`** (relocate-first, else unregister stale aliases)
+- **Alias is always derived** — the user-settable `--alias`/`-a` flag was removed from `index add`; the alias always equals the (sanitized) directory name via `ReposConfig::register()`. The alias remains the internal identifier (repos.json key, groups, `project` arg); only user override is gone. The `index symbol <alias>` positional is a lookup key and is retained
+- **Hand-edited `repos.json` tolerated** — `ReposConfig::reconcile()` runs in-memory on every load: drops empty-alias entries, drops orphan `repos_meta`, prunes group members referencing unknown aliases and empty groups. Never renames valid aliases, never crashes
 - **TUI C# indicator** — in status column: green `C#·` ready, yellow `C#…` indexing, red `C#!` error; footer shows helper availability; Calls column with tool call count
 - **Phase 2 & 3 TUI feedback** — Phase 2 pre-marks all queued candidates as `C#…` immediately on discovery (before semaphore slot); Phase 3 pre-warm sets `csharp_index_status = Indexing` before `batch-find-refs` and restores `Ready` after — TUI shows `C#…` throughout without touching `active_reindexes` (avoids blocking HTTP /reindex)
 - **Selective ref cache invalidation** — incremental rebuilds only purge cached refs for affected symbols, not entire cache

CHANGELOG.md

@@ -7,6 +7,39 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 
 
+## [1.0.152] - 2026-06-02
+
+### Added
+
+- **Best-effort relocation of moved/renamed repositories** — every repo's git
+  remote (`remote.origin.url`) is now captured at registration. When a
+  registered folder is renamed or moved, `codesearch serve` no longer crashes:
+  on startup it reconciles all paths, and for each missing path it scans nearby
+  folders (bounded depth, override with `CODESEARCH_RELOCATE_MAX_DEPTH`, default
+  `3`) for a git checkout with the same remote. A single unambiguous match is
+  rewritten into `repos.json`; ambiguous/absent matches are logged and skipped
+  (the dead path is never indexed). Phase-2 (C# SCIP) and Phase-3 (pre-warm)
+  also guard `path.exists()` so a stale path can never reach heavy code paths.
+- **`codesearch index prune`** — new command that relocates moved repos first,
+  then unregisters any remaining stale entries, printing a summary.
+
+### Changed
+
+- **The user-settable `--alias`/`-a` flag was removed from `index add`** — the
+  alias (the `repos.json` key, used by groups and the MCP `project` argument) is
+  now always derived from the repository directory name. In practice the alias
+  always had to equal the directory name, so a custom alias only caused
+  downstream mismatches. The `index symbol <alias>` positional (a lookup key) is
+  unchanged.
+
+### Fixed
+
+- **A hand-edited or corrupt-ish `repos.json` no longer crashes the app** — on
+  load the config is reconciled in memory: entries with empty/blank alias keys
+  are dropped, orphaned `repos_meta` is removed, and group members referencing
+  unknown aliases (and groups left empty) are pruned. Valid aliases are never
+  renamed (that would break group references).
+
 ## [1.0.146] - 2026-06-02
 
 ### Added

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "codesearch"
-version = "1.0.146"
+version = "1.0.152"
 edition = "2021"
 authors = ["codesearch contributors"]
 license = "Apache-2.0"

README.md

@@ -118,6 +118,9 @@ codesearch index rm /path/to/my-project
 
 # List registered repos
 codesearch index list
+
+# Remove stale entries (relocates moved repos first, then drops the rest)
+codesearch index prune
 ```
 
 `codesearch index add` is intended to be run from inside the repo you want to register.
@@ -312,17 +315,39 @@ Repos are registered via `codesearch index add`:
 
 ```bash
 # Register a repo (creates index + adds to ~/.codesearch/repos.json)
-codesearch index add /path/to/my-project --alias my-project
+codesearch index add /path/to/my-project
 
 # Remove a repo
 codesearch index rm /path/to/my-project
 
 # List registered repos
 codesearch index list
+
+# Clean up stale entries (relocates moved repos, drops the rest)
+codesearch index prune
 ```
 
+The repository **alias** (the key in `repos.json`, used for groups and the MCP
+`project` argument) is always derived automatically from the directory name —
+there is no `--alias` flag.
+
 Serve reads `~/.codesearch/repos.json` on startup and manages all registered repos.
 
+#### Moved or renamed repositories
+
+If you rename or move a registered folder, serve does **not** crash. On startup
+it tries to **relocate** each missing repo automatically: it captures every
+repo's git remote (`remote.origin.url`) at registration, and on a missing path
+it scans nearby folders (bounded depth, override with
+`CODESEARCH_RELOCATE_MAX_DEPTH`, default `3`) for a git checkout with the same
+remote. A single unambiguous match is rewritten into `repos.json`; otherwise the
+entry is logged and skipped (never indexed against a dead path). Run
+`codesearch index prune` to relocate what can be relocated and drop the rest.
+
+A hand-edited `repos.json` is also tolerated: empty entries, orphaned metadata,
+and group references to unknown repos are cleaned up on load rather than
+crashing.
+
 ### Groups
 
 Groups let you search across related repositories:

src/cli/mod.rs

@@ -20,10 +20,6 @@ pub enum IndexCommands {
         /// Create global index instead of local
         #[arg(short = 'g', long)]
         global: bool,
-
-        /// Alias for this repository (auto-generated from directory name if omitted)
-        #[arg(short, long)]
-        alias: Option<String>,
     },
 
     /// Remove the index (local or global, auto-detected)
@@ -49,6 +45,9 @@ pub enum IndexCommands {
         #[arg(short = 'f', long)]
         force: bool,
     },
+
+    /// Remove stale entries from repos.json (relocates moved repos first)
+    Prune,
 }
 
 /// Cache subcommands
@@ -235,10 +234,6 @@ pub enum Commands {
         #[arg(short = 'g', long)]
         global: bool,
 
-        /// Alias for this repository (only with --add)
-        #[arg(short, long)]
-        alias: Option<String>,
-
         /// Remove the index (local or global, auto-detected)
         #[arg(long, visible_alias = "rm")]
         remove: bool,
@@ -532,7 +527,6 @@ pub async fn run(cancel_token: CancellationToken) -> Result<()> {
             symbols,
             add,
             global,
-            alias,
             remove,
             keep_config,
             list,
@@ -543,11 +537,7 @@ pub async fn run(cancel_token: CancellationToken) -> Result<()> {
                     IndexCommands::Add {
                         path: add_path,
                         global,
-                        alias,
-                    } => {
-                        crate::index::add_to_index(add_path, global, alias, cancel_token.clone())
-                            .await
-                    }
+                    } => crate::index::add_to_index(add_path, global, cancel_token.clone()).await,
                     IndexCommands::Remove {
                         path: rm_path,
                         keep_config,
@@ -556,6 +546,7 @@ pub async fn run(cancel_token: CancellationToken) -> Result<()> {
                     IndexCommands::Symbol { alias, force } => {
                         trigger_symbol_reindex_via_api(&alias, force).await
                     }
+                    IndexCommands::Prune => crate::index::prune_index().await,
                 }
             } else {
                 // Flag-based backward-compat path
@@ -569,8 +560,7 @@ pub async fn run(cancel_token: CancellationToken) -> Result<()> {
 
                 if add || is_add_cmd {
                     let effective_path = if is_add_cmd { None } else { path };
-                    crate::index::add_to_index(effective_path, global, alias, cancel_token.clone())
-                        .await
+                    crate::index::add_to_index(effective_path, global, cancel_token.clone()).await
                 } else if remove || is_rm_cmd {
                     let effective_path = if is_rm_cmd { None } else { path };
                     crate::index::remove_from_index(effective_path, keep_config).await
@@ -911,22 +901,32 @@ mod tests {
     }
 
     #[test]
-    fn test_cli_index_add_accepts_alias_flag() {
-        let cli = Cli::try_parse_from([
+    fn test_cli_index_add_rejects_alias_flag() {
+        // The user-settable alias was removed; the flag must no longer parse.
+        let result = Cli::try_parse_from([
             "codesearch",
             "index",
             "add",
             "/tmp/foo",
             "--alias",
             "myrepo",
-        ])
-        .expect("cli parse should succeed");
+        ]);
+        assert!(
+            result.is_err(),
+            "'--alias' flag should no longer be accepted on `index add`"
+        );
+    }
+
+    #[test]
+    fn test_cli_index_add_parses_without_alias() {
+        let cli = Cli::try_parse_from(["codesearch", "index", "add", "/tmp/foo"])
+            .expect("cli parse should succeed");
         match cli.command {
             Commands::Index {
-                command: Some(IndexCommands::Add { alias: Some(a), .. }),
+                command: Some(IndexCommands::Add { path: Some(p), .. }),
                 ..
-            } => assert_eq!(a, "myrepo"),
-            _ => panic!("expected Index::Add subcommand with alias"),
+            } => assert_eq!(p, std::path::PathBuf::from("/tmp/foo")),
+            _ => panic!("expected Index::Add subcommand"),
         }
     }
 

src/constants.rs

@@ -192,6 +192,13 @@ pub const DEFAULT_EMBEDDING_DIMENSIONS: usize = 384;
 /// Environment variable to override repos config file path.
 pub const REPOS_CONFIG_ENV: &str = "CODESEARCH_REPOS_CONFIG";
 
+/// Environment variable to override how deep relocation scans for a moved repo.
+pub const RELOCATE_MAX_DEPTH_ENV: &str = "CODESEARCH_RELOCATE_MAX_DEPTH";
+
+/// Default bounded depth for the relocation scan (directories below the nearest
+/// existing ancestor of a stale repo path).
+pub const DEFAULT_RELOCATE_MAX_DEPTH: usize = 3;
+
 /// Environment variable to set MCP mode: "auto", "client", or "local".
 pub const MCP_MODE_ENV: &str = "CODESEARCH_MCP_MODE";