Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add swarm-ai-research/swarm --skill run-scenariogit clone --depth 1 https://github.com/swarm-ai-research/swarmWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/swarm-ai-research/swarm/run-scenario)<a href="https://agentmods.dev/skills/swarm-ai-research/swarm/run-scenario"><img src="https://agentmods.dev/badge/skills/swarm-ai-research/swarm/run-scenario/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/swarm-ai-research/swarm/run-scenario"><img src="https://agentmods.dev/badge/skills/swarm-ai-research/swarm/run-scenario.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00014 | $0.00739 |
| Opus 5 | $0.00007 | $0.00369 |
| Sonnet 5 | $0.00003 | $0.00148 |
| Haiku 4.5 | $0.00001 | $0.00074 |
Grade A, and why
run-scenario scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run Scenario Skill
Execute a single SWARM scenario with a given seed and export all artifacts to a standardized output directory.
Prerequisites
swarm-safetypackage installed (pip install swarm-safetyorpip install -e /root/swarm-package/)- Scenario YAML file available (typically in
/root/scenarios/)
Procedure
1. Resolve the scenario path
Scenario references can be shorthand or full paths:
baseline→scenarios/baseline.yamlscenarios/baseline.yaml→ use as-is/root/scenarios/baseline.yaml→ use as-is
import os
def resolve_scenario(ref: str) -> str:
"""Resolve a scenario reference to a full path."""
candidates = [
ref,
f"scenarios/{ref}.yaml",
f"/root/scenarios/{ref}.yaml",
f"scenarios/{ref}",
]
for c in candidates:
if os.path.isfile(c):
return c
raise FileNotFoundError(f"Cannot find scenario: {ref}")
2. Run the simulation
Use the SWARM CLI to execute:
python -m swarm run <scenario_path> --seed <seed> --epochs <N> --steps <M>
Or programmatically:
from swarm.core.orchestrator import Orchestrator
from swarm.scenarios.loader import load_scenario
config = load_scenario(scenario_path)
# Override simulation parameters if needed
config["simulation"]["seed"] = seed
config["simulation"]["n_epochs"] = epochs
config["simulation"]["steps_per_epoch"] = steps
orch = Orchestrator(config)
result = orch.run()
3. Export artifacts
After the run completes, export to the output directory:
import json
import os
os.makedirs(output_dir, exist_ok=True)
# Export history.json
with open(os.path.join(output_dir, "history.json"), "w") as f:
json.dump(result.to_dict(), f, indent=2)
# Export CSV metrics
csv_dir = os.path.join(output_dir, "csv")
os.makedirs(csv_dir, exist_ok=True)
result.export_csv(csv_dir)
4. Extract key metrics
The final epoch snapshot contains summary metrics:
history = result.to_dict()
final = history["epoch_snapshots"][-1]
welfare = final["welfare"]
toxicity = final["toxicity_rate"]
print(f"Final welfare: {welfare:.3f}")
print(f"Final toxicity: {toxicity:.3f}")
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 119 lines · 14 tokens per session scan A 8388a78f8f90
run-scenario is a skill published in the GitHub repository swarm-ai-research/swarm (42 stars, last pushed yesterday), licensed MIT. It adds 14 tokens to every session and 739 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
gsmm-builder
Build or load a genome-scale metabolic model (GSMM) using COBRApy. Covers loading from BIGG, constructing minimal models from scratch, setting medium constraints, and exporting validated .json model files.
stat-result-validator
Validate statistical research outputs for formulation quality, method-to- problem alignment, theory presence, experimental evidence, fair comparison, artifact completeness, and final-claim consistency.
meta-analysis
Statistical methods for combining results across multiple studies. Use when aggregating cross-study or cross-experiment results.
mixed-precision
Use FP16/BF16 mixed precision to accelerate training and reduce memory. Use when optimizing GPU performance.
aclawdemy
The academic research platform for AI agents. Submit papers, review research, build consensus, and push toward AGI — together.
adme-property-predictor
Predict ADME (Absorption, Distribution, Metabolism, Excretion) properties for drug candidates using cheminformatics models and molecular descriptors. Evaluates drug-likeness, bioavailability, and pharmacokinetic profile to guide lead optimization and candidate selection in drug discovery.