HackerSec Launches AI-Native Pentest Platform with Yaga Agent for Automated Offensive Testing
HackerSec, an offensive cybersecurity company, has launched a major update to its pentest platform featuring a new Pentest AI-Native technology. In this model, an AI agent takes over complete execution of offensive tests while human specialists only monitor the operation and validate each vulnerability.
The platform performs reconnaissance, enumeration, validations, exploitations, impact confirmation, and evidence production autonomously. Cybercriminals are already deploying AI agents to automate reconnaissance, identify vulnerabilities, and scale offensive campaigns. At the same time, organizations across sectors accelerate development with AI, producing more updates, more production code, and a constantly expanding attack surface that requires pentesting to keep pace.
In the AI-Native model, the intelligence receives the scope, investigates the environment, interprets system responses, formulates hypotheses, and adapts its strategy based on results. The agent maintains full operational context and continues exploring different paths until confirming whether a vulnerability is truly exploitable, enabling complete, continuous testing with reduced human intervention.
At the center of the platform is Yaga, the pentest agent developed from scratch by HackerSec. The technology uses a proprietary harness that coordinates tools, playbooks, multiple analysis stages, and different AI models during the same execution. Instead of depending on the reasoning of a single model, Yaga combines distinct capabilities to interpret the target, select techniques, execute authorized attacks, and technically validate findings.
Results published on YagaBench demonstrate the impact of this multi-model architecture. The production configuration of Yaga, operating with four combined models, solved 91.2% of black box scenarios, 94.6% of gray box scenarios, and 93.5% of white box scenarios. This performance highlights that the platform's advantage lies in the proprietary offensive orchestration layer developed by HackerSec.
“Our goal is to achieve, by the end of the year, more than 98% effectiveness in all AI-Native pentests with Yaga,” commented Andrew Martinez, CEO of HackerSec. New cyber environments change at a speed that no longer matches point-in-time pentests, long cycles, and high dependence on manual execution. The AI-Native model allows companies to perform offensive testing more frequently, expand coverage, and reduce the time between vulnerability emergence and identification while keeping human experts responsible for monitoring and technical validation.
Related articles
Claude Encrypted Thinking Blocks Use Protobuf with Exposed Metadata and AES-GCM Ciphertext
A detailed reverse-engineering of Claude signatures shows that the encrypted reasoning blocks are not opaque containers but structured protobuf messages. The outer envelope contains a 312-byte inner message that holds a 135-byte header, fixed-length nonce and MAC fields, and the actual ciphertext. The header itself reveals the model name such as claude-opus-5, the block type as thinking, and the organizationUuid from the user's Anthropic account. Only the reasoning text is encrypted with AES-GCM, adding exactly 16 bytes for the authentication tag. The analysis covers four protocol versions and notes that organization binding was added in version 15, potentially allowing servers to reject cross-model or cross-organization reuse. The findings provide concrete implications for both the Opus-to-Haiku extraction attack and the leakage of account identifiers in public logs.
Researchers Extract Proprietary Reasoning Traces from Anthropic, OpenAI and Google LLMs, Revealing Hidden Secrets
A team of eight researchers from institutions including ELLIS Institute Tübingen, the Max Planck Institute for Intelligent Systems, Tübingen AI Center, MATS and Snyk published a preprint detailing a practical attack that recovers full reasoning traces from closed LLM APIs. The method requires only two API calls and works by feeding encrypted reasoning blocks from strong models such as Claude Opus 4.8 into weaker models from the same provider, such as Haiku 4.5, which then reproduce the hidden chain-of-thought verbatim. Analysis of 6,708 publicly shared agent logs from GitHub and Hugging Face yielded 315,320 recovered traces containing 704 unique secrets, including 62 API keys, 33 passwords and 24 access tokens that never appeared in visible session output. The attack also enables extraction of internal safety policies, system prompts and detailed harmful planning that providers normally filter from final answers. In addition, the same mechanism can be used in reverse to inject malicious instructions into shared logs that later get replayed by unsuspecting users. The authors recommend treating encrypted reasoning blocks as sensitive secrets and propose cryptographic binding of traces to sessions, users and models.
GhostSplice Technique Lets Malicious MCP Servers Trick AI Coding Agents into Exfiltrating Secrets
GhostSplice is a new technique that allows a malicious MCP server to induce an AI coding agent to leak SSH keys, environment secrets, and source code. The attack splits malicious instructions across tool metadata and responses so the agent reconstructs and executes the full exfiltration plan without detecting an overtly malicious command. Tests showed the method raised compliance rates from an average of 42 percent to 82 percent across eleven models, with some systems moving from zero to 100 percent success. The technique requires the developer to connect the attacker-controlled MCP server and for the agent to already possess read access to the targeted files. Defenses focus on strict allow-listing of MCP servers, least-privilege tool permissions, separation of tool output from instructions, and human approval for sensitive operations. The disclosure aligns with prior warnings about poisoned MCP tool descriptions and agentjacking attacks.
Anthropic Claude Code Auto Mode Launches August 14 with Local Classifier and Permission Rules
Starting August 14, Claude Code will run in auto mode on new sessions for Pro, Max, and Team plans, replacing the allow/deny dialog with a local classifier that evaluates every tool call. The classifier rules are stored locally and contain 103 categories across allow, soft_deny, hard_deny, and environment sections, with the single hard_deny rule focused on data exfiltration spanning over 5,000 characters. Enterprise, API, Bedrock, Vertex, and Foundry deployments remain on opt-in for another month. Auto mode pauses after three consecutive blocks or twenty blocks in a session, and broad allow rules such as python:* are disabled while narrow permissions continue to function. Administrators should populate the twenty environment fields, currently only one-third configured on clean machines, before the rollout date.