Developer path

Try K² adversarial context in 10 minutes

Load the public adversarial corpora bundle, connect the K² MCP server to an existing generator, and inspect a cited evaluation plan before any adversarial input is generated.

Quickstart steps

Commands are illustrative until the public repo is published and the pipeline schema is finalized.

Clone and inspect the public bundle

The bundle contains public threat-pattern records, example policy scope, a synthetic target profile, past findings, and MCP examples.

git clone https://github.com/knowledge2-ai/k2-adversarial-context-demo.git

Load the adversarial corpora into K²

Use a K² project and keep credentials in environment variables or a local ignored file. Public examples use placeholders only.

export K2_API_KEY="<your-k2-api-key>"
export K2_API_HOST="https://api.knowledge2.ai"
python scripts/load_adversarial_demo_k2.py --execute

Connect your generator over MCP

The quickstart uses PyRIT as a vendor-neutral open-source example, but any MCP-capable harness can call the same plan endpoint.

docs/customer-demos/demo-adversarial-context/k2-assets/examples/mcp-config.example.json

Call get_evaluation_plan

Ask K² for a plan scoped to the synthetic SupportBot target, text-plus-image modalities, and staging environment.

get_evaluation_plan(
  target_id="target-supportbot-v2.3",
  modalities=["text", "image"],
  environment="staging"
)

Verify lineage survives handoff

Run one cited plan entry through your generator and confirm the threat, policy, target, and finding references remain attached downstream.

python scripts/pyrit_plan_smoke.py --plan-id seed-0142-regression --dry-run

MCP config snippets

All examples point at the same plan-only K² boundary. Keep live credentials out of committed files.

{
  "mcpServers": {
    "k2-adversarial-context": {
      "command": "python",
      "args": ["scripts/k2_adversarial_mcp_server.py"],
      "env": {
        "K2_API_KEY": "${K2_API_KEY}",
        "K2_API_HOST": "https://api.knowledge2.ai",
        "K2_PROJECT_ID": "${K2_PROJECT_ID}"
      }
    }
  }
}

[mcp_servers.k2-adversarial-context]
command = "python"
args = ["scripts/k2_adversarial_mcp_server.py"]
env_vars = [
  "K2_API_KEY",
  "K2_API_HOST",
  "K2_PROJECT_ID",
  "K2_ADV_THREATS_CORPUS_ID",
  "K2_ADV_POLICY_CORPUS_ID",
  "K2_ADV_TARGET_CORPUS_ID",
  "K2_ADV_FINDINGS_CORPUS_ID"
]

plan = k2.get_evaluation_plan(
    target_id="target-supportbot-v2.3",
    modalities=["text", "image"],
    environment="staging",
)
# PyRIT or your harness owns prompt generation, execution, and scoring.

Expected response

The first query should return scoped, cited plan context, not generated adversarial prompts.

Successful shape

A target profile for SupportBot v2.3 with accepted modalities and tools.
Threat-pattern candidates filtered by modality and target model class.
Policy verdicts that mark in-scope, out-of-bounds, and review-required entries.
Plan lines with threat, policy, target, and past-finding citations.

What to watch for

Generated jailbreak prompts at this stage cross the K² boundary.
Plan lines without policy citations are not ready for reviewer handoff.
Target-irrelevant modalities should be filtered before generation.

Next steps

Once the cited plan works, the useful question is how it maps to your current red-team process.

Replace the synthetic target profile with one production target you already red-team.

Freeze 10 to 20 evaluation plans before running a pilot.

Keep your existing generator, scorer, guardrails, and GRC workflow unchanged.