Clear Search Results Safely

indexly clear-search removes rows from the local search index database, fts_index.db. It does not delete files from disk.

Use it when indexed search results are stale, a folder was moved, a tag-based batch should be removed from search, or you want to rebuild the whole search index from a clean state.

Quick Reference

Syntax

indexly clear-search (--path PATH | --tag TAG [TAG ...] | --all) [--dry-run] [--yes]

Exactly one deletion mode is required:

• --path PATH deletes indexed entries for one matching path, directory-like prefix, or basename fallback.
• --tag TAG [TAG ...] deletes indexed entries matching any provided tag. Multiple tags use OR logic.
• --all clears all rows from the search index tables.

Common safety flags:

• --dry-run shows what would be deleted without changing the database.
• --yes skips confirmation for --path, --tag, and --all.

What Gets Deleted

clear-search deletes matching rows from these tables in fts_index.db:

• file_index: searchable FTS5 content and indexed path rows
• file_tags: path-to-tag mapping
• file_metadata: structured metadata for indexed paths

It also invalidates search cache entries in search_cache.json that reference deleted paths. For --all, the whole search cache is cleared because every cached search result becomes suspect.

clear-search does not delete:

• source files on disk
• cleaned analysis data in the separate stats database
• backups, organizer logs, or exported reports

Path Deletion

Use --path when a file or folder should disappear from search results.

Preview first:

indexly clear-search --path "V:/Hotline/CustomerA" --dry-run

Delete after reviewing the plan:

indexly clear-search --path "V:/Hotline/CustomerA"

Automated script mode:

indexly clear-search --path "V:/Hotline/CustomerA" --yes

Path matching works in this order:

1. Exact normalized path match
2. Directory-like prefix match, for example V:/Hotline/CustomerA/%
3. Basename fallback, useful for legacy rows or older imports

On Windows, paths are compared case-insensitively. UNC paths are normalized to forward-slash form, such as //server/share/folder.

Tag Deletion

Use --tag when you want to remove a tagged set from the search index.

Preview:

indexly clear-search --tag archive --dry-run

Delete:

indexly clear-search --tag archive

Multiple tags use OR logic. This deletes files matching archive or stale:

indexly clear-search --tag archive stale --dry-run

Tags are matched as comma-separated values, so review does not accidentally match preview.

Clear The Whole Search Index

Use --all only when you plan to rebuild the index:

indexly clear-search --all --dry-run
indexly clear-search --all
indexly index "/path/to/rebuild"

For unattended rebuild scripts:

indexly clear-search --all --yes
indexly index "/path/to/rebuild"

Confirmation And Reports

Before a real deletion, Indexly prints a deletion plan:

Deletion plan [del-20260507-154032-abc123]
   Reason: tag:archive
   Matching files: 247
   Planned entries: 741
   - file_index: 247 row(s)
   - file_tags: 247 row(s)
   - file_metadata: 247 row(s)
   Example paths:
     - V:/Hotline/CustomerA/report-001.pdf
     - V:/Hotline/CustomerA/report-002.pdf

If --yes is not provided, Indexly asks:

Found 247 file(s) matching your criteria. Proceed? (y/n):

Answer y or yes to continue. Any other answer cancels the operation.

Dry-Run Workflow

The safest workflow is:

indexly clear-search --path "V:/Hotline/CustomerA" --dry-run
indexly clear-search --path "V:/Hotline/CustomerA"

Use --dry-run whenever:

• the path is broad, such as a drive root or shared folder
• tags are reused across many projects
• you are cleaning a production index
• you are not sure whether a path was normalized the way you expect

Large Operations

For large deletions, Indexly processes paths in chunks and prints progress:

Deleting 5,000 paths (13 chunks)...
   Processing chunk 3/13 (~2 seconds remaining)

For very broad matches, Indexly warns before continuing. Treat warnings on drive roots, network shares, and organization-wide tags as a prompt to run --dry-run again with narrower criteria.

Logging And Audit Trail

Each deletion operation gets an operation ID such as:

del-20260507-154032-abc123

Indexly writes a manifest before deletion and summary events after deletion to the NDJSON log system:

• SEARCH_DELETE_INITIATED
• SEARCH_RESULT_DELETED
• SEARCH_DELETE_SUMMARY

Use the operation ID to connect the plan, deleted paths, and final summary in logs.

See Indexly Logging System for log locations and NDJSON format details.

Troubleshooting

The Command Says A Tag Was Not Found

If a tag does not exist, Indexly prints available tags when it can:

Tag 'archive' not found (0 files match). Available tags: reviewed, complete, pending

Check spelling with:

indexly stats
indexly tag list --file "/path/to/a/known/file"

The Path Exists But Has No Indexed Files

This means the folder exists on disk, but no matching paths are currently stored in fts_index.db.

Try:

indexly index "/path/to/folder"
indexly search "known term" --path-contains "/path/to/folder"

Cache Invalidation Fails

If Indexly cannot save the updated search_cache.json, it warns but keeps the database deletion successful:

Warning: Cache invalidation failed (...). Search cache may be stale.
Database deletion succeeded despite the cache warning.

Fix the file permission or disk space issue, then remove search_cache.json manually from the Indexly runtime directory. The cache will be recreated by future searches.

Runtime directory defaults:

• Windows: %APPDATA%/indexly
• macOS: ~/Library/Application Support/indexly
• Linux: $XDG_DATA_HOME/indexly or ~/.local/share/indexly
• Override: INDEXLY_HOME

Database Is Locked Or Corrupted

If you see a database error:

indexly doctor
indexly doctor --full-integrity
indexly update-db

Close other Indexly processes, watchers, or editors that may be holding the SQLite database open, then retry. Use --full-integrity when you need the slower read-only SQLite integrity_check, because the normal Doctor run may skip deep integrity scans on large databases.

Recovery

clear-search removes index rows, not files. To recover deleted search results, re-index the source folder:

indexly index "/path/to/folder"

If the source files were moved or renamed, index the new location. If the files no longer exist, there is nothing for Indexly to re-index.

Best Practices

• Use --dry-run first for every broad path or tag.
• Prefer folder-level paths over drive roots.
• Use specific tags for cleanup workflows, such as stale-index or archive-2026.
• Keep source-file deletion and search-index cleanup as separate steps.
• Run indexly doctor after database warnings.
• Re-index immediately after --all when users expect search to keep working.

Related Documentation

• Usage Guide
• Search Files with Indexly
• Indexing
• Tagging
• Indexly Doctor
• Database Design