SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2308.07124 · 2023

OctoPack: Instruction Tuning Code Large Language Models

arXiv · PDF · Open in the Atlas

Code that ran

9 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 8 failures of our instrument; 4 no recorded run), measured 2026-11-21; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2308.07124.json · badge: /api/ran/2308.07124.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
bigcode-project/octopack canonical (classifier label)
classifier basis: classifier_2025
6 ran · 0 no recorded run
bigcode-project/bigcode-evaluation-harness canonical (classifier label)
classifier basis: census
3 ran · 4 no recorded run
FunctionStatusWhere it lives
cpp_instruct Ran
ran; our contract draft was wrong, not the code
bigcode-project/octopack/evaluation/create/create_instructions.py
code served (permissive licence) · get_code("7655561f064db48a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re

def cpp_instruct(ex):
    """
    Prompt example:
    /*
    Check if in given vector of numbers, are any two numbers closer to each other than
    given threshold.
    >>> has_close_elements({1.0, 2.0, 3.0}, 0.5)
    false
    >>> has_close_elements({1.0, 2.8, 3.0, 4.0, 5.0, 2.0}, 0.3)
    true
    */
    #include<stdio.h>
    #include<vector>
    #include<math.h>
    using namespace std;
    bool has_close_elements(vector<float> numbers, float threshold){
    """
    prompt = ex['prompt']
    function_name = ex['entry_point']
    lines = [line.strip() for line in prompt.split('\n') if line.strip()]
    function = None
    for idx, line in enumerate(reversed(lines)):
        if function_name in line and line.strip().endswith('{'):
            function = line
            break
    if function is None:
        raise ValueError(f'Could not find function declaration. Prompt: {prompt} Function: {function_name}')
    signature = function.strip('{')
    splits = re.split('(\\/\\*|\\*\\/)', prompt)
    if len(splits) < 5:
        is_comment = False
        comment_lines = []
        for line in reversed(lines):
            if line.strip().startswith('//'):
                is_comment = True
                comment_lines.append(line.strip('//').strip())
            elif is_comment:
                break
        docstring = '\n'.join([x.strip() for x in reversed(comment_lines) if x.strip()])
    else:
        left, middle, _ = (splits[-5], splits[-3].strip('*'), splits[-1])
        docstring = '\n'.join([x.strip() for x in middle.split('\n') if x.strip()])
    ex['signature'] = signature
    ex['docstring'] = docstring
    ex['instruction'] = f'Write a C++ function `{signature}` to solve the following problem:\n{docstring}'
    return ex


# Construct a valid example dictionary with prompt and entry_point
ex = {
    'prompt': '''/*
Check if in given vector of numbers, are any two numbers closer to each other than
given threshold.
>>> has_close_elements({1.0, 2.0, 3.0}, 0.5)
false
>>> has_close_elements({1.0, 2.8, 3.0, 4.0, 5.0, 2.0}, 0.3)
true
*/
#include<stdio.h>
#include<vector>
#include<math.h>
using namespace std;
bool has_close_elements(vector<float> numbers, float threshold){
''',
    'entry_point': 'has_close_elements'
}

ARGS = {
    'ex': ex
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_diff Ran
ran; our contract draft was wrong, not the code
bigcode-project/octopack/dataset/commitpack/scrape_github.py
code served (permissive licence) · get_code("db89171e76d09770")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os
import random
import tempfile

# Create a minimal example dict that satisfies get_diff's requirements
# The function expects a dict with keys: 'commit', 'repos', 'old_file', 'new_file'

ex = {
    'commit': 'abc123def456',
    'repos': 'owner/repo1,owner/repo2',
    'old_file': 'path/to/old_file.py',
    'new_file': 'path/to/new_file.py'
}

ARGS = {
    'ex': ex
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_file_contents Ran
ran; our contract draft was wrong, not the code
bigcode-project/octopack/dataset/commitpack/scrape_github.py
code served (permissive licence) · get_code("2905235d73624835")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tempfile
import os

# Create a temporary directory to use as cwd for git operations
temp_dir = tempfile.mkdtemp()

ARGS = {
    'commit': 'HEAD',
    'old_file': 'old_file.txt',
    'new_file': 'new_file.txt',
    'repo': 'https://github.com/example/repo.git',
    'cwd': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_test_results_json_path Ran
ran; our contract draft was wrong, not the code
bigcode-project/bigcode-evaluation-harness/bigcode_eval/tasks/custom_metrics/multiple_metrics/evaluation.py
code served (permissive licence) · get_code("4ba028fb30a021b2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned PosixPath. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

from pathlib import Path

ARGS = {
    "output_dir": "/tmp/output",
    "problem_json_path": "problem_name.json",
    "input_dir": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type PosixPath — a PosixPath output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a PosixPath output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
java_instruct Ran
ran; our contract draft was wrong, not the code
bigcode-project/octopack/evaluation/create/create_instructions.py
code served (permissive licence) · get_code("ef807f76cf003934")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re

def java_instruct(ex):
    """
    Prompt example:
    import java.util.*;
    import java.lang.*;

    class Solution {
        /**
        Check if in given list of numbers, are any two numbers closer to each other than given threshold.
        >>> hasCloseElements(Arrays.asList(1.0, 2.0, 3.0), 0.5)
        false
        >>> hasCloseElements(Arrays.asList(1.0, 2.8, 3.0, 4.0, 5.0, 2.0), 0.3)
        true
        */
        public boolean hasCloseElements(List<Double> numbers, double threshold) {
    """
    prompt = ex['prompt']
    function_name = ex['entry_point']
    lines = [line.strip() for line in prompt.split('\n') if line.strip()]
    function = lines[-1]
    signature = function.strip('{').strip()
    splits = re.split('(\\/\\*|\\*\\/)', prompt)
    assert len(splits) >= 5, f'Error in splitting function declaration. Prompt: {prompt} Splits: {str(splits)}'
    left, middle, _ = (splits[-5], splits[-3].strip('*'), splits[-1])
    docstring = '\n'.join([x.strip() for x in middle.split('\n') if x.strip()])
    ex['signature'] = signature
    ex['docstring'] = docstring
    ex['instruction'] = f'Write a Java function `{signature}` to solve the following problem:\n{docstring}'
    return ex


ARGS = {
    'ex': {
        'prompt': '''import java.util.*;
import java.lang.*;

class Solution {
    /**
    Check if in given list of numbers, are any two numbers closer to each other than given threshold.
    >>> hasCloseElements(Arrays.asList(1.0, 2.0, 3.0), 0.5)
    false
    >>> hasCloseElements(Arrays.asList(1.0, 2.8, 3.0, 4.0, 5.0, 2.0), 0.3)
    true
    */
    public boolean hasCloseElements(List<Double> numbers, double threshold) {
''',
        'entry_point': 'hasCloseElements'
    }
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
python_instruct Ran
ran; our contract draft was wrong, not the code
bigcode-project/octopack/evaluation/create/create_instructions.py
code served (permissive licence) · get_code("ef34b9d45f6cbdf6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re

ARGS = {
    'ex': {
        'prompt': '''from typing import List


def has_close_elements(numbers: List[float], threshold: float) -> bool:
    """ Check if in given list of numbers, are any two numbers closer to each other than
    given threshold.
    >>> has_close_elements([1.0, 2.0, 3.0], 0.5)
    False
    >>> has_close_elements([1.0, 2.8, 3.0, 4.0, 5.0, 2.0], 0.3)
    True
    """
    pass
''',
        'entry_point': 'has_close_elements'
    }
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
remove_after_return Ran
ran; our contract draft was wrong, not the code
bigcode-project/bigcode-evaluation-harness/bigcode_eval/utils.py
code served (permissive licence) · get_code("29ee1d8c120f674a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re

def remove_after_return(code):
    """
    Takes as input a code, and removes everything that is after the return.
    That is, the first line that does not start with a space character
    """
    pattern = '[^\\n]+(\\n|$)'
    end_last_match = None
    for match in re.finditer(pattern, code):
        start_match, end_match = match.span()
        if end_last_match is not None and start_match < len(code) and (code[start_match].strip() != ''):
            return code[0:start_match]
        end_last_match = end_match
    return code

# Test code with indentation and a non-indented line (simulating return)
test_code = """    def foo():
    x = 1
    return x
print('this should be removed')
more_code = 2"""

ARGS = {
    "code": test_code
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
run_in_shell Ran
ran on a synthesized input
bigcode-project/octopack/dataset/commitpack/scrape_github.py
code served (permissive licence) · get_code("3266895adde27ef7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned CompletedProcess. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import subprocess

# The function run_in_shell takes:
# - cmd: a shell command string
# - cwd: optional working directory (defaults to None)
# - timeout: timeout in seconds (defaults to 60)

# We'll construct valid arguments for all parameters
ARGS = {
    "cmd": "echo 'Hello, World!'",
    "cwd": None,
    "timeout": 60
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type CompletedProcess — a CompletedProcess output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a CompletedProcess output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
update_code_gens Ran
ran; our contract draft was wrong, not the code
bigcode-project/bigcode-evaluation-harness/bigcode_eval/utils.py
code served (permissive licence) · get_code("ac86a6f136bea277")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import warnings
from unittest.mock import Mock, MagicMock

torch.manual_seed(0)

# Create a mock task object with required methods and attributes
class MockTask:
    def __init__(self):
        self.stop_words = ['<eos>']
    
    def postprocess_generation(self, code, sample_idx):
        return f"postprocessed_{code}_{sample_idx}"

# Create a mock tokenizer object with required methods and attributes
class MockTokenizer:
    def __init__(self):
        self.name_or_path = 'facebook/incoder-1B'
        self.eos_token = '<eos>'
        self.bos_token_id = 50256
        self.special_tokens_map = {
            'eos_token': '<eos>',
            'bos_token': '<bos>',
            'additional_special_tokens': ['<special1>', '<special2>']
        }
    
    def decode(self, token_ids, skip_special_tokens=False, clean_up_tokenization_spaces=False):
        # Simple mock decode that returns a string representation
        if isinstance(token_ids, list):
            result = ''.join([f'token_{tid}' for tid in token_ids])
        else:
            result = 'decoded_tokens'
        return result
    
    def get_prefix_tokens(self):
        # Return empty list for prefix tokens
        return []

# Create a mock task
task = MockTask()

# Create a mock tokenizer
tokenizer = MockTokenizer()

# Create other arguments
limit_start = 0
prefix = "def hello():\n    "
instruction_tokens = ['start_token', 'end_token', 'assistant_token']
postprocess = True

# Create code_gens dict - maps sample index to list of generated code strings
code_gens = {
    '0': [],
    '1': []
}

# Create gen_token_dict - maps sample index to list of token sequences
gen_token_dict = {
    '0': [
        [50256, 1, 2, 3],  # sequence with bos_token_id at start
        [4, 5, 6, 7]       # another sequence
    ],
    '1': [
        [50256, 8, 9, 10],
        [11, 12, 13, 14]
    ]
}

ARGS = {
    'task': task,
    'tokenizer': tokenizer,
    'limit_start': limit_start,
    'prefix': prefix,
    'instruction_tokens': instruction_tokens,
    'postprocess': postprocess,
    'code_gens': code_gens,
    'gen_token_dict': gen_token_dict
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cache_get Not yet run
unverified (harvested, no recorded run)
bigcode-project/bigcode-evaluation-harness/bigcode_eval/tasks/custom_metrics/multiple_metrics/evaluation.py
code served (permissive licence) · get_code("f2e0ef72cfe052c9")
How it ran

no invocation recorded.

cached_eval_script Not yet run
unverified (harvested, no recorded run)
bigcode-project/bigcode-evaluation-harness/bigcode_eval/tasks/custom_metrics/multiple_metrics/evaluation.py
code served (permissive licence) · get_code("2bceb28a9eb6ff16")
How it ran

no invocation recorded.

complete_code Not yet run
unverified (harvested, no recorded run)
bigcode-project/bigcode-evaluation-harness/bigcode_eval/utils.py
code served (permissive licence) · get_code("21177b52be3390ee")
How it ran

no invocation recorded.

compute_codexglue_code_to_text_bleu Not yet run
unverified (harvested, no recorded run)
bigcode-project/bigcode-evaluation-harness/bigcode_eval/tasks/codexglue_code_to_text.py
code served (permissive licence) · get_code("d611d112ec49cc8e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 2 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 2 later) and 11 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
remove_after_return 4 other papers
4 later
2024.findings-emnlp.86 (2024 (year only))
Tokenization Falling Short: On Subword Robustness in Large Language Models
2406.11687 (2024-06)
Tokenization Falling Short: On Subword Robustness in Large Language Models
2406.12809 (2024-06)
Can Large Language Models Always Solve Easy Problems if They Can Solve Harder Ones?
and 1 more not named here
update_code_gens 1 other paper
1 later
2410.05357 (2024-10)
Model-GLUE: Democratized LLM Scaling for A Large Model Zoo in the Wild

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-21.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Finetuning large language models (LLMs) on instructions leads to vast performance improvements on natural language tasks. We apply instruction tuning using code, leveraging the natural structure of Git commits, which pair code changes with human instructions. We compile CommitPack: 4 terabytes of Git commits across 350 programming languages. We benchmark CommitPack against other natural and synthetic code instructions (xP3x, Self-Instruct, OASST) on the 16B parameter StarCoder model, and achieve state-of-the-art performance among models not trained on OpenAI outputs, on the HumanEval Python benchmark (46.2% pass@1). We further introduce HumanEvalPack, expanding the HumanEval benchmark to a total of 3 coding tasks (Code Repair, Code Explanation, Code Synthesis) across 6 languages (Python, JavaScript, Java, Go, C++, Rust). Our models, OctoCoder and OctoGeeX, achieve the best performance across HumanEvalPack among all permissive models, demonstrating CommitPack's benefits in generalizing to a wider set of languages and natural coding tasks. Code, models and data are freely available at https://github.com/bigcode-project/octopack.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2308.07124")
get_code_for_paper("2308.07124")
have("2308.07124")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2308.07124.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2308.07124.svg)](https://syntology.ai/paper/2308.07124)

Connect an agent — have() is free.