Cot Redteam Agent

Refusal quotes of a canary are not a finding. This CLI scores visible

The keyless mock provider never touches the network. The config below is also

(findings). That is the demo working — not a real-model vulnerability.

Exit codes: 0 clean, 1 findings, 2 config/env, 3 partial. Gate CI on

1 . Switch mock_mode to refuse for a clean pass. From a clone, the v0.6

Pick the tool that matches the job. This project does not replace a broad

and last real successful disclosure (refusal re-quotes are not counted as success).

agent behavior under adversarial inputs. It runs reproducible model attacks and

Version 0.6.0 adds the Proof-of-Action agent execution lane: a deterministic

preserves the 0.5 adaptive attacks and TUI, 0.3 benchmark, and 0.2 Python

Runs built-in and third-party attacks against one or more target models.

if the attacker provider fails. Requires attacker_model in attack_config .

any harmful request (phishing, fraud, weapons, hate speech, ...) with a

scan and exits 0 (clean) / 1 (findings) / 3 (partial) — gate CI on it.

OWASP LLM Top 10 tags in every report item (LLM01/02/05/07...).

Preserves system, developer, user, assistant, and simulated tool message roles.

Produces Markdown, CSV, and LaTeX reports with honest eligibility counts.

Provider keys are read only from named environment variables. Secrets must not