SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2609.02459 · September 2026

CivBench: A Long-Horizon Benchmark for Tool-Mediated Agents in Civilization VI

Jakob Foerster, Rui Ponte Costa, David Andrews, Harry Coppock, Jakob Nicolaus Foerster, Austin Tudor David Andrews, Liam Wilkinson, Jamie Heagerty

arXiv · PDF · Open in the Atlas

Code that ran

4 of 7 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 0 have no contract verdict; 4 failures of our instrument; 3 no recorded run), measured 2026-11-22; 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/2609.02459.json · badge: /api/ran/2609.02459.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
lmwilki/civ6-mcp canonical (classifier label)
classifier basis: census
4 ran · 3 no recorded run
FunctionStatusWhere it lives
diary_path Ran
ran; our contract draft was wrong, not the code
lmwilki/civ6-mcp/src/civ_mcp/diary.py
code served (permissive licence) · get_code("c04843f32c8c1e37")
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.

import torch
from pathlib import Path

ARGS = {
    'civ': 'Rome',
    'seed': 42,
    'run_id': 'experiment_001'
}
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 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
get_autosave_for_turn Ran
ran; our contract draft was wrong, not the code
lmwilki/civ6-mcp/src/civ_mcp/autosave.py
code served (permissive licence) · get_code("d730d2b4fa982e5a")
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 sys

_IS_LINUX = sys.platform == 'linux'

def get_autosave_for_turn(turn: int) -> str:
    """Return the expected autosave name for a specific turn.

    On Windows: 0_MCP_NNNN
    On Linux: AutoSave_NNNN (game's own, may not be exact turn match)
    """
    if _IS_LINUX:
        return f'AutoSave_{turn:04d}'
    return f'0_MCP_{turn:04d}'


ARGS = {
    'turn': 42
}
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
merge_agent_reflections Ran
ran; our contract draft was wrong, not the code
lmwilki/civ6-mcp/src/civ_mcp/diary.py
code served (permissive licence) · get_code("0c2362f632bf48f8")
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 json
import tempfile
from pathlib import Path

# Create a temporary file with JSONL content
with tempfile.NamedTemporaryFile(mode='w', suffix='.jsonl', delete=False) as f:
    temp_path = Path(f.name)
    
    # Write some agent rows with turn information
    row1 = {
        'is_agent': True,
        'turn': 1,
        'reflections': {
            'tactical': 'initial tactic',
            'strategic': 'initial strategy'
        }
    }
    row2 = {
        'is_agent': False,
        'turn': 1,
        'reflections': {}
    }
    row3 = {
        'is_agent': True,
        'turn': 2,
        'reflections': {
            'tactical': 'turn 2 tactic'
        }
    }
    row4 = {
        'is_agent': True,
        'turn': 2,
        'reflections': {
            'planning': 'existing plan'
        }
    }
    
    f.write(json.dumps(row1, separators=(',', ':')) + '\n')
    f.write(json.dumps(row2, separators=(',', ':')) + '\n')
    f.write(json.dumps(row3, separators=(',', ':')) + '\n')
    f.write(json.dumps(row4, separators=(',', ':')) + '\n')

# Create new reflections to merge
new_reflections = {
    'tactical': 'improved tactic',
    'strategic': '',
    'tooling': 'new tool',
    'planning': 'revised plan',
    'hypothesis': 'test hypothesis'
}

ARGS = {
    'path': temp_path,
    'turn': 2,
    'new_reflections': new_reflections
}
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 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
read_diary_entries Ran
ran; our contract draft was wrong, not the code
lmwilki/civ6-mcp/src/civ_mcp/diary.py
code served (permissive licence) · get_code("0a551a14f892bbf0")
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 list, 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 json
from pathlib import Path
import tempfile

# Create a temporary directory and file with diary entries
temp_dir = tempfile.TemporaryDirectory()
temp_path = Path(temp_dir.name) / "diary.jsonl"

# Write some valid JSON Lines format diary entries
diary_entries = [
    {"date": "2024-01-01", "entry": "First day of the year"},
    {"date": "2024-01-02", "entry": "Second day", "mood": "happy"},
    {"date": "2024-01-03", "entry": "Third day"},
]

with open(temp_path, "w") as f:
    for entry in diary_entries:
        f.write(json.dumps(entry) + "\n")

ARGS = {
    "path": temp_path,
}
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 list, ndim 1, dtype family float — a list 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
list_autosaves Not yet run
unverified (harvested, no recorded run)
lmwilki/civ6-mcp/src/civ_mcp/game_launcher.py
code served (permissive licence) · get_code("fcd4bd83754e18b9")
How it ran

no invocation recorded.

load_save_from_menu Not yet run
unverified (harvested, no recorded run)
lmwilki/civ6-mcp/src/civ_mcp/game_launcher.py
code served (permissive licence) · get_code("128884353e4f8db0")
How it ran

no invocation recorded.

restart_and_load Not yet run
unverified (harvested, no recorded run)
lmwilki/civ6-mcp/src/civ_mcp/game_launcher.py
code served (permissive licence) · get_code("875ee72c758cb1c9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 7 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.

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 2026-11-22.

Repositories linked to this paper

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

Abstract

We present CivBench, an open-source benchmark for evaluating language model agents in long-horizon, tool-mediated environments through the Model Context Protocol (MCP). A single episode spans 300+ turns and produces thousands of tool calls over a large action space, requiring sustained planning, state monitoring, and execution under partial observability. The environment exposes 76 MCP tools and a narration layer that converts visual game state into structured text. We use CivBench to characterise agent behaviour across four model families in 23 admissible runs. The sample is a pilot, not a model ranking: aggregate outcomes do not reliably discriminate models at this scale. Instead, we introduce two interface-level metrics that the environment makes measurable: Proactive Monitoring Rate (PMR), capturing whether agents actively query latent strategic state, and RAG@10, capturing whether commitments stated in structured planning reflections are executed within ten subsequent turns. Across runs we observe two consistent patterns under a shared playbook protocol. Agents under-monitor strategically relevant state that is available but requires explicit querying: despite playbook guidance to query victory progress every 20 turns, agents do so only every 30 to 75 turns, and in 7 of 20 detectable defeats they failed to query within the 20 turn warning window before game end. Agents also frequently fail to execute near-term commitments stated in their own planning reflections (RAG@10 between 48.2% and 65.8% across models). Both patterns arise despite tool access and explicit guidance, and we interpret them as deviations under instruction rather than absences of capability. We release the environment, scenarios, logs, metrics, and analysis pipeline at https://github.com/lmwilki/civ6-mcp

For agents

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

get_harvested_code_for_paper("2609.02459")
get_code_for_paper("2609.02459")
have("2609.02459")

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

curl https://syntology.ai/api/ran/2609.02459.json

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

[![Syntology run record](https://syntology.ai/api/ran/2609.02459.svg)](https://syntology.ai/paper/2609.02459)

Connect an agent — have() is free.