socbench

Benchmark frontier reasoning LLMs as SOC agents on raw NetFlow data.

socbench benchmarks frontier reasoning models as SOC agents: each model runs a bounded multi-turn agent loop against a deterministic, pre-indexed NetFlow corpus, with persona-scoped read-only tools, fixed dollar caps per investigation, and a strict final-answer JSON contract. Four personas (SOC Analyst, Threat Analyst, Adversary Hunter, Detection Engineer) and three providers (OpenAI, Anthropic, Google) share the same eval units, scoring lenses, and ablation surface so the headline numbers and tools_off / playbooks_off deltas are directly comparable.

The repository is local-first. A laptop, three API keys, and a sample parquet committed to the repo are enough to reproduce a smoke under a $10 budget.

Status

Alpha. The full pipeline runs end-to-end. Build-out covered:

• Step 1: package skeleton, contracts, configs, schema
• Step 2: the index builder (socbench build-index) with deterministic content-addressed indexes
• Step 3: read-only tools layer with persona allowlist + sample builder
• Step 4: personas, playbooks, prompt compose + forbidden-token check
• Step 5: provider adapters (OpenAI / Anthropic / Gemini + always-on mock) and the multi-turn agent loop with budget caps and cost/latency rollups
• Step 6: scoring (per-flow / per-pair / per-host F1), stratified sampling, ablation aggregation
• Step 7: quickstart + results-explorer notebooks; reproduction instructions in REPRODUCE.md

You can run a complete smoke today with no API keys via the mock provider (see Quickstart step 3, or notebooks/quickstart.ipynb).

Install

socbench ships as a standard PEP 621 / hatchling project. Either install path works.

With uv (recommended for development)

curl -LsSf https://astral.sh/uv/install.sh | sh

git clone https://github.com/DeepTempo/socbench.git
cd socbench

uv venv --python 3.11
source .venv/bin/activate
uv pip install -e ".[dev,providers]"

With plain pip

git clone https://github.com/DeepTempo/socbench.git
cd socbench

python3.11 -m venv .venv
source .venv/bin/activate
pip install -e ".[dev,providers]"

Either way, socbench --help should now list the available subcommands.

Configuration

config/benchmark_config.yaml ships safe defaults: smoke cost_budget_usd: 10, full cost_budget_usd: 900, fixed cost_usd_cap_per_rendering: 0.50. Paths inside it that point at sibling config files (schema_path, pricing_path) resolve relative to the YAML's own directory, so renaming or relocating config/ doesn't require any code edits.

Quickstart

1. Build a content-addressed index from a parquet dataset

socbench build-index \
  --config config/benchmark_config.yaml \
  --dataset sample

This normalizes the parquet against config/schema.json, sorts globally by ts_start with deterministic tie-breaking, assigns stable flow_ids, derives pair_timeline / host_egress eval units, computes rollups, and writes to indexes/<dataset_hash>/.

Re-running the command on the same data is a no-op. Pass --rebuild to force a rebuild.

2. Inspect the tool layer

socbench tools-smoke \
  --dataset-hash <dataset_hash> \
  --persona soc_analyst

This invokes every tool in the persona's allowlist against the built index and prints a summary, with no model calls.

3. Run the benchmark

# Free, deterministic, no API keys (the mock provider):
socbench run --dataset-hash <dataset_hash> --providers mock --personas all

# Real models (after `pip install -e ".[providers]"` + exporting API keys):
socbench run --dataset-hash <dataset_hash> --providers all --personas all

Unit selection defaults to stratified sampling, deterministic in (dataset_hash, sample_seed, mode). Each (unit × persona × provider) rendering runs a bounded multi-turn agent loop; results land under runs/<run_id>/ with summary.json (scoring + cost + cache rollups), eval_units_summary.jsonl, predictions_raw.jsonl, renderings.jsonl, tool_calls.jsonl, and prompts_used/.

4. Run ablations and aggregate the deltas

socbench run --dataset-hash <dataset_hash> --ablation tools_off --providers mock --personas all
socbench aggregate --dataset-hash <dataset_hash>
# → ablations/<dataset_hash>/<seed>/ablation_summary.json  (tools_off → main deltas)

5. Explore

notebooks/quickstart.ipynb runs the whole loop (it synthesizes a sample dataset so it needs no committed data) and plots per-persona F1. notebooks/results_explorer.ipynb loads any runs/<run_id>/ and slices the results by stratum, persona, and provider. Install with pip install -e ".[notebooks]".

Extending the benchmark

Every interface designed to evolve is a registry or a YAML key:

• New tool: drop a new file under src/socbench/tools/catalog/<name>.py with a Tool subclass, register it in src/socbench/tools/catalog/__init__.py by appending to ALL_TOOLS, then add its name to the appropriate persona tools: lists in config/benchmark_config.yaml. The tools_manifest_sha shifts automatically. Filename, YAML name, and matrix entry are 1:1 by design.
• New eval-unit type: add an assigner to src/socbench/index.py and a matching Literal to EvalUnitType in src/socbench/models.py.
• New provider adapter: implement the Adapter ABC in a new src/socbench/providers/<name>_adapter.py, register it in the factory in , and add an entry under in . Pricing goes in . SDK imports stay lazy so the dependency is optional.

Methodology

The full methodology (eval units, persona x tool matrix, agent loop, scoring, cost model, repair policy, sampling, ablations, run artifacts) is implemented across the module-level files in src/socbench/ (each carries a focused module docstring).

License

Apache-2.0. See LICENSE.