> ## Documentation Index
> Fetch the complete documentation index at: https://diasporic3lee7-ci-auto-mmdc-diagram-render.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Rebuild the Knowledge Indices

> Learn when and how to rebuild the Termux Monorepo knowledge indices, including the combined pipeline and individual build scripts for AST and dependency graphs.

The Termux Monorepo maintains a set of knowledge indices that map code structure, function signatures, file relationships, and dependency graphs. You should rebuild these indices after large refactors, new component imports, or any change that alters the AST shape of the codebase. This guide covers the full rebuild pipeline, individual scripts, and how to query the results.

### When to rebuild

Rebuild the indices when any of the following occur:

* You add, remove, or rename a module or package
* Function signatures change across multiple files
* You merge a refactor that touches imports or cross-file references
* The `llm_index.jsonl`, `func_index.jsonl`, `file_graph.json`, or `deps.jsonl` files are older than the latest commit

### Combined rebuild pipeline

The fastest way to regenerate everything is the chained command used by the Central Mapper:

<CodeGroup>
  ```bash theme={null}
  map-build && map-func && fore
  ```
</CodeGroup>

This runs the AST index builder, the function index builder, and the forensic linker in sequence. Expected output:

```text theme={null}
[map-build] Indexed 1,247 files in 4.2s
[map-func] Extracted 8,903 function signatures
[fore] Linked 3,212 cross references
```

### Individual build scripts

If you only need to update one layer, run the specific script directly.

**Build all indices from scratch:**

<CodeGroup>
  ```bash theme={null}
  python3 workspace/llm_map/build_all.py
  ```
</CodeGroup>

**Build the LLM-readable index:**

<CodeGroup>
  ```bash theme={null}
  python3 workspace/llm_map/build_llm_index.py
  ```
</CodeGroup>

**Build the AST index from existing sources:**

<CodeGroup>
  ```bash theme={null}
  python3 workspace/llm_map/build_ast_index_from_existing.py
  ```
</CodeGroup>

**Build the dependency graph (fast):**

<CodeGroup>
  ```bash theme={null}
  python3 workspace/llm_map/build_graph_fast.py
  ```
</CodeGroup>

### Querying the index

After rebuilding, use `archivist.py` to search the knowledge base for files, functions, or relationships.

<CodeGroup>
  ```bash theme={null}
  python3 archwiz/archivist.py --query "function:dispatch_task"
  python3 archwiz/archivist.py --query "file:archwiz.py" --relationships
  ```
</CodeGroup>

Expected output:

```text theme={null}
[archivist] 3 matches for "function:dispatch_task"
  - archwiz/dispatch_task.py:42  dispatch_task(task_id, dry_run=False)
  - archwiz/autonomous_runner.py:88  runner.dispatch(task_id)
  - tests/test_dispatch.py:15  test_dispatch_ok()
```

<Tip>
  Use the context-relationship-graph skill for reconnaissance before making changes. Run `archivist.py --graph --file <target>` to see which files depend on your target so you can plan impact before you edit.
</Tip>

<Note>
  The index files live in `workspace/llm_map/`:

  * `llm_index.jsonl` — file-level metadata for LLM context
  * `func_index.jsonl` — function signatures and docstrings
  * `file_graph.json` — file-to-file reference graph
  * `deps.jsonl` — import and module dependency list
</Note>

For a deeper investigation of what changed after a rebuild, see the [Forensics guide](/guides/forensics). To understand how the mapper fits into the overall system, read the [Central Mapper component](/components/central-mapper) overview.