IndexerLabs

Showing 25 of 62 from the live probe.

• Discover current IndexerLabs tools

  Use this when the needed IndexerLabs operation is missing from the client's focused tool list or its current schema is uncertain. Search the live server catalog, then call invoke_read or invoke_write with the returned exact tool name and arguments. This stable discovery tool is designed to survive future server-side tool additions.

• Invoke current IndexerLabs read tool

  Use this only as the stable fallback for a read-only tool discovered through indexerlabs.capabilities when that focused tool is absent from the client's snapshot. The server rechecks the target's current read-only annotation, authorization scopes, and arguments before running it.

• Invoke current IndexerLabs write tool

  Use this only when the user explicitly asks for a state change and the focused write tool discovered through indexerlabs.capabilities is absent from the client's snapshot. Read the target state first. The server rechecks the target's current write annotation, authorization scopes, session and cost gates, and arguments.

• Read MCP identity

  Read opaque IndexerLabs authorization status and granted MCP scopes. The authorizing user's email and delegated tokens are never returned.

• Read current indexing prices

  Read current server-configured book-index tier names and prices. This does not create a checkout or purchase.

• Read source PDF pages

  Extract text from up to 12 selected PDF pages after checking the signed-in user's access to the job.

• Search source PDF text

  Search text across an accessible job's source PDF and return bounded page snippets.

• Browse classified index terms

  Read the canonical classification taxonomy, full-index category counts, and a paginated flat list of entries and subentries. Common aliases are accepted: names/authors map to the broad person class, and titles/books map to the broad work class; the pipeline does not classify those as narrower subtypes. When the user asks for all matches, repeat with offset=next_offset until next_offset is null. This tool never launches classification; indexerlabs.compute.classify_entries is separately cost-beari

• Read compact index revision metadata

  Read only the current editable-index hash and entry counts. Use this compact tool to obtain expected_hash after locating targets with term_inventory or entry; use entries_snapshot only when the complete tree is needed.

• Start MCP editing session

  Open one auditable editing session for a job. Reuse it for the task, verification, and any immediate undo; close it only when that window is complete.

• List MCP editing sessions

  List this signed-in user's own active or closed MCP editing sessions.

• Close MCP editing session

  Close an MCP editing session only after verifying the result and completing any immediate undo. Closed sessions reject every subsequent write, paid-agent call, and MCP snapshot undo.

• Read MCP action audit

  Read the technical MCP call audit, optionally filtered to a session or job. Use editor.action_logs instead when the user asks what index edits or undo/redo actions occurred.

• Apply exact index edits

  Preferred fast path for exact edits: atomically batch up to 100 field updates or single-node inserts against the hash from snapshot_meta or entries_snapshot. Send only intended public fields. The result includes an undo_snapshot_id and a ready-to-call undo action.

• Delete multiple index entries

  Atomically delete up to 100 entries or subentries by stable ID against the hash from editor.snapshot_meta or editor.entries_snapshot. Deleting a parent also deletes its remaining subtree. The response returns one undo snapshot for the whole batch.

• Undo MCP index change

  Fastest undo for a write in the current active session. Pass its returned undo_snapshot_id; the restore runs only if the index still matches the exact post-change hash, so it cannot overwrite later work. For older edits, use editor.action_logs then editor.undo_history_action.

• Undo multiple MCP index changes

  Atomically undo up to 100 consecutive writes from the current editing session. Pass their returned undo snapshot IDs in any order; the server builds and verifies the hash chain, then restores the oldest before-state in one conflict-safe replacement.

• Read account balance and entitlements

  Read the signed-in IndexerLabs account profile, credit balance, and available indexing runs.

• List account transactions

  Read the signed-in account's credit and indexing transaction history. Owner-wide access remains subject to the existing API permission check.

• List accessible projects

  List the signed-in user's IndexerLabs projects and projects shared with that user.

• List jobs in a project

  List jobs the signed-in user may view inside one accessible project.

• Read project access

  Read the current viewer's access details for a project. This tool never changes sharing or permissions.

• Read job details

  Read one accessible IndexerLabs job, including status, type, page count, and viewer access.

• Read subject review

  Read the subject-indexing review bundle for an accessible job when that review exists.

• Get source PDF URL

  Create a short-lived signed URL for an accessible job's source PDF. Prefer source.read_pages or source.search when only text is needed.