A tracker of publicly reported prompt-injection techniques, broken down by delivery method, encoding, and propagation behavior, with the models each was confirmed against, the reporting source, and (where known) the attack source. Fields are left blank when a report does not state them; techniques whose details are vague or unconfirmed are marked Not fully vetted. Nothing here is guessed.

• Prompt Injection in the Wild @ GitHub.

Open-source agent firewall for MCP and AI agent egress

Pipelock enforces MCP, HTTP, and WebSocket egress at the network boundary and produces verifiable audit evidence for every inspected action.

• Pipelock @ GitHub.

Related contents:

• Pipelock: Open-source AI agent firewall @ Help Net Security.

LLM Vulnerability Scanner.

Test large language models against 210+ adversarial attacks covering prompt injection, jailbreaks, encoding exploits, and data extraction.

Related contents:

• Digest #202: Terraform Claude Skills, FinOps FOCUS 1.2, AI Fatigue for Cloud Engineers, and MCP for Web Data Extraction @ DevOps Bulletin.

NOVA: The Prompt Pattern Matching. NOVA is a Python-based tool for detecting adversarial prompts - prompt injections, exfiltration, evasion and TTPs.

NOVA is an open-source prompt pattern matching system combining keyword detection, semantic similarity, and LLM-based evaluation to analyze and detect prompt content.

• Nova Framework @ GitHub.

Related contents:

• Introducing NOVA @ SecurityBreak's Medium.

⚡ Vigil ⚡ Detect prompt injections, jailbreaks, and other potentially risky Large Language Model (LLM) inputs.

Vigil is a Python library and REST API for assessing Large Language Model prompts and responses against a set of scanners to detect prompt injections, jailbreaks, and other potential risks.

• Vigil @ GitHub.
• Vigil: Open-source LLM security scanner.