Builder: set the count limit.

Builder: set the output format.

Set the old line number.

Set the path prefix filter. Only files under this prefix will be processed. # Arguments `prefix` - Path prefix (e.g., "src/", "tests/") # Example

Builder: set the `--recursive` flag.

Builder: set short output mode.

Builder: set the stat flag.

Builder: set the view name.

Write the hash table to a writer. Writes `len() × 32` bytes: each hash in index order, raw (no compression). The caller is responsible for writing th

Add an entry to the manifest.

Add an added file to the result.

Add a dependency by node ID, looking up the hash. # Arguments `node_id` - Internal ID of the change # Returns Ok(()) if the dependency was added,

Add a graph_op to this file.

Add a line credit (will merge with adjacent ranges if possible).

Add a node to the graph.

Adds a token to this line.

Align the view pointer without materializing the working copy. This persists the new view name to `.atomic/current_view` so that subsequent `status()

Allocate a new unique inode. Each call returns a different inode value. # Example ```rust use atomic_core::output::Memory; let wc = Memory::new();

This exports the working copy state at the current (or specified) Merkle state to the given destination. # Arguments `destination` - Path to the out

Get the raw bytes of the public key

Mark this entry as a directory.

Assert that the content retrieved via overlay matches the expected bytes.

Mark the start of a new turn.

Build edges connecting a decision node to the surrounding context. The decision node gets: - An edge from the most recent goal (if any precedes the s

Compare two byte slices and produce diff operations. This function handles the actual diffing of file content. # Arguments `old_content` - Content

Compute cosine similarity between two vectors. Returns 0.0 if the vectors have different lengths, are empty, or if either has zero magnitude.

Create a simple test change with the given message.

Decode a `SessionEnvelope` from `HashedChange.metadata` bytes. Validates the magic prefix and schema version before deserializing. # Errors Returns

Derive the agent author from the user's identity. Constructs: `{agent_name}+{session_short} <{user_email}>` If the user has no email, uses the agent

Synchronous embed using a throwaway tokio runtime (for use in non-async contexts). When called from within an existing Tokio runtime (e.g., from hook

Ensure that the required metadata sections (HEADER + DEPS) have been written.

Pure decision core for [`check_push_credentials`], split out so it can be unit-tested without touching the on-disk identity store. `identity_name` —

Extract a file path from a node's detail or summary.

Format entries for default output. # Arguments `entries` - History entries to format `hash_length` - Number of hash characters to display # Returns

Format a hash for display with custom length. Like [`format_hash`], but allows specifying the truncation length. # Arguments `hash` - The hash to f

Format entries for JSON output. # Arguments `entries` - History entries to format # Returns Formatted JSON string.

Format entries for short output. # Arguments `entries` - History entries to format `hash_length` - Number of hash characters to display # Returns

Get the changelist for a view starting from a position. # Arguments `view` - The name of the view to query. `from` - The starting position (sequence

Convert command options to TrackingOptions.

Returns `true` if the outcome is [`Clean`](Self::Clean).

Returns `true` if the graph has no nodes.

Check if a file path is supported for AST extraction.

Iterate over all change hashes stored in the repository. This scans the changes directory and yields the hash of each change file found. The iteratio

Iterate through all lines with their display status. Lines are yielded in the order they would appear in a unified diff: - Unchanged lines appear onc

Load an identity by ID.

Load a full `Change` from the redb store. This reads all tables for the change and reassembles a `Change` object. Equivalent to deserializing a compl

Find the Longest Increasing Subsequence of matches by new_idx. Given matches sorted by old_idx, find the longest subsequence where new_idx values are

Create a test change with content

Marks this conflict as resolved with optional notes.

Mark a position as verified rooted.

Set maximum span count.

Set the maximum number of vertices to retrieve. This provides a safety limit for retrieval. If the graph exceeds this size, retrieval will stop and t

Merge parallel results into the aggregate vectors needed by `Repository::record()`. This extracts the `RecordedFile` values from the parallel results

Get the commit message if a header is loaded.

Get the effective model name.

Create a new name conflict. Name conflicts occur when a file has been renamed by concurrent changes.

Generate the next node ID.

Generate the next unique node ID and increment the counter.

All nodes in insertion order.

Normalize an intent ID — accept "PIMO-1", "pimo-1", or just "1".

Output graph content with semantic-merge resolution. This is the merge-aware counterpart of [`output_graph_content`]. For every SCC whose lead verte

Process files in parallel using rayon, returning per-file results. This is the core of the parallel recording pipeline (Phase 2). It takes a list of

Get the permission bits. # Returns The Unix permission bits.

Print the status in long (human-readable) format.

Process a diff operation. This is the main entry point for processing diff output. Call this for each diff operation in sequence. # Arguments `op`

Process an insertion. # Arguments `old_index` - Position in old content (insertion point) `new_index` - Starting line in new content `new_len` - Num

Provision an agent's private working tree as a copy-on-write clone of this repository's working tree. Copies every tracked-or-untracked working file

Push a node onto the graph, updating stats and last_node cursor.

PUT request with JSON body, unwrap `ApiResponse` envelope.