Highflame is a security platform designed to provide identity, authorization, and governance for AI agents operating within enterprise environments. It addresses the challenges of managing agent authority, ensuring that every agent has a verifiable identity, and that each action taken by an agent is authorized, attributed, and auditable. The platform is positioned as an authority system for organizations deploying agentic systems, focusing on security, compliance, and operational control.
The platform functions by sitting between AI agents and the resources they access, granting each agent a real, standards-based identity and enforcing authorization policies before any action is executed. Highflame uses a signal detection engine to score agent actions against adaptive guardrails, making real-time inline decisions at every boundary an agent crosses. Every action is signed, creating a tamper-evident record that attributes the action to both the agent and its human owner. This enables audit-ready evidence mapped to common frameworks such as OWASP, NIST, MITRE, and the EU AI Act, supporting governance, risk, and compliance teams in their oversight responsibilities.
Highflame offers continuous discovery of agents across clouds, IDEs, and SaaS platforms, connecting existing identities or minting new ones as needed. It creates a live graph mapping agents, their owners, and the resources they can access. The platform supports integration with a range of agent frameworks and tools, including OpenAI, Anthropic, Claude, Code, LangGraph, CrewAI, AutoGen, LlamaIndex, GitHub Copilot, Cursor, Windsurf, Agentforce, Slack, Linear, Jira, Notion, and MCP. Features include inline redaction of sensitive information, short-lived per-tool credentials, and cascade revocation across agent fleets.
Highflame Identity, a component of the platform, is available as open source and is built on open standards. The platform is intended for engineering, security, IT, platform, and compliance teams seeking to discover, control, and govern AI agents at scale, ensuring that every action is authorized and attributable by design.
Highflame sits in PulseGate's Security & compliance platforms category. It focuses on managing and securing the identities and actions of AI agents within enterprise environments. Highflame is a B2B product aimed at enterprise IT and security teams. The project is open source (Apache-2.0). The product ships for the web and API.
Behind Highflame is Highflame, and the product first shipped in 2025. The project is developed in the open on GitHub with 92 stars and 100 commits in the last 90 days. Among its 5 catalogued features are agent identity, authorization control, and audit logs.
Latest indexed changes and source events
Show HN: Deconstructing Anthropic's Coding Agent Control Model verified by the PulseGate indexer
Other apps tracked under the same category.
