Module tools 

Sandboxed tools exposed to the model as OpenAI function calls. Target-crate tools (read_file, list_dir, grep, optional rg, web_fetch, and gated edit_file) resolve paths relative to the scheduler crate or public URLs; comparison tools read other schedulers under scheds/rust but never write them. Host-topology tools run fixed read-only commands or read fixed sysfs CPU cache metadata paths. Absolute paths and .. are rejected, and canonicalized targets must stay inside their configured roots.

Constants

DEFAULT_WEB_FETCH_BYTES 🔒

LOCAL_TOOL_TIMEOUT 🔒

MAX_GREP_MATCHES 🔒

MAX_READ_BYTES 🔒

MAX_RG_OUTPUT_BYTES 🔒

MAX_SCHED_READ_BYTES 🔒

MAX_TOPOLOGY_OUTPUT_BYTES 🔒

MAX_WEB_FETCH_BYTES 🔒

Functions

collapse_text 🔒

count_occurrences 🔒

cpu_cache_sizes 🔒

cpu_cache_sizes_from_root 🔒

decode_basic_entities 🔒

edit_file 🔒

execute_tool

Dispatch a tool call. allow_edit gates edit_file.

execute_tool_async

Async dispatch for tools that may perform network I/O.

fuzzy_line_match 🔒

Find a unique block of lines in content whose whitespace-trimmed text equals the trimmed lines of needle (ignoring indentation and blank lines around the block). Returns the byte range [start, end) of that block, or None if there is not exactly one match.

grep 🔒

grep_dir 🔒

grep_schedulers 🔒

html_to_text 🔒

is_disallowed_host 🔒

is_disallowed_ip 🔒

is_write_tool

True for write-capable tools (used by the controller to count edits per round).

list_dir 🔒

list_schedulers 🔒

lscpu_e 🔒

nearest_lines 🔒

Up to 5 file lines (with 1-based line numbers) that contain anchor, for error feedback when old_string doesn’t match.

normalized_crate_relative_path 🔒

numactl_hardware 🔒

openai_tools_json

The function-calling tool schema advertised to the model.

parse_prefixed_u32 🔒

read_file 🔒

read_path 🔒

read_scheduler_file 🔒

read_trimmed 🔒

resolve 🔒

Resolve a crate-relative path inside sandbox, rejecting ../absolute and escapes.

resolve_scheduler_path 🔒

rg 🔒

rg_available 🔒

rg_glob_arg 🔒

run_fixed_topology_command 🔒

scheduler_dir 🔒

strip_line_number_prefixes 🔒

Strip a leading line-number+tab prefix (as emitted by read_file’s "{:>6}\t" format) from each line. Preserves a trailing newline.

truncate_lossy 🔒

validate_public_url 🔒

web_fetch 🔒

web_fetch_content_type_allowed 🔒

web_fetch_max_bytes 🔒