Customer guides

Learn the risk before buying the license.

Focused CapitalGuard guides for everyday AI users, freelancers, founders, agencies, and teams giving AI tools sensitive access.

Test an AI-agent kill switch

A five-step pre-flight for exact-session stop requests, connector closure, bounded shutdown, and evidence.

Open path

Respond to an AI-agent incident

A five-step containment and recovery workflow with a private 28-field incident register.

Open path

Secure MCP tools

A five-step review for server identity, tool definitions, least privilege, approvals, and untrusted results.

Open path

Audit AI-agent permissions

A five-step access audit and blank register for files, tools, secrets, networks, approvals, and permission drift.

Open path

Offboard people and agents

A five-step closure workflow and blank register for identities, sessions, repositories, tokens, connectors, and shared secrets.

Open path

Verify an AI-agent license

A five-step fail-closed workflow for authority, signature, expiry, revocation, and exact-session checks.

Open path

GitHub Actions AI-agent gate

A cited eight-control profile and SHA-pinned workflow for least privilege, trusted triggers, OIDC identity, and human approval.

Open path

Client code workflow

A cited pre-flight checklist and blank access register for freelancers and teams using AI with authorized client repositories.

Open path

Everyday AI users

A practical access check for people using AI with private files, client work, cloud apps, folders, or commands.

Open path

Scanner intent

Teams looking for AI-agent repo scanning, secret exposure checks, and sample reports.

Open path

Threat research

Customers comparing prompt injection, secret context, automation, and agent-tool bridge risk.

Open path

Policy intent

Founders who need AI-agent guardrail rules, policy files, and safer coding-agent defaults.

Open path

Agency distribution

Agencies packaging fixed-scope security checks before client onboarding.

Open path

Primary research

Open controls, test vectors, Guardprint methodology, verification, and the public evidence boundary.

Open path

Bottom-funnel library

16 focused pages for people already searching for the problem.

Compare Pricing

Cursor security

Cursor Security Checklist

A practical checklist for protecting secrets, workflows, commands, instructions, and sensitive repository paths before Cursor receives broad access.

Read guide

Claude Code security

Claude Code Security Risks

Review the file, command, tool, prompt-injection, and workflow risks that matter before Claude Code works across a production repository.

Read guide

Codex security

Codex Security Checklist

A defensive checklist for repository scope, terminal commands, tool access, secrets, deployment files, and human review when teams use Codex.

Read guide

Copilot security

Copilot Security Checklist

Review repository instructions, suggestions, secrets, pull requests, workflows, and protected paths before expanding Copilot across engineering teams.

Read guide

AI-agent risk

AI Coding Agent Risks

Understand the financial, operational, legal, reputational, and client-trust exposure created when coding agents gain broad repository and tool access.

Read guide

Prompt-injection defense

Prompt Injection in Repositories

How malicious or stale instructions in docs, issues, prompts, logs, comments, and fixtures can influence AI coding agents with repository access.

Read guide

Secret exposure

AI Agent Exposed Secrets

What to do when AI coding tools can read environment files, private keys, package credentials, service accounts, or sensitive production configuration.

Read guide

GitHub workflow

GitHub AI Security Scanner

What a GitHub-native AI-agent security scanner should cover across repositories, pull requests, status checks, workflows, policies, issues, and verification.

Read guide

Repository exposure

AI-Agent Repository Exposure

Map what AI agents can read, change, run, trust, summarize, or leak across code, configuration, docs, workflows, and client context.

Read guide

Founder security

AI Security License for Founders

Choose a one-time AI-agent security license before coding tools gain wider access to product code, credentials, customer context, and company workflows.

Read guide

Agency security

AI-Agent Security for Agencies

Package AI-agent exposure scans, client-ready reports, policy artifacts, exports, and verification into a repeatable agency service.

Read guide

Customer guide

One-Time AI Agent Security License

Why a fixed-scope license can be a cleaner first purchase than stacking monthly security tools before the AI-agent exposure surface is known.

Read guide

Prevention guide

How to Evaluate AI Firewall Claims for Coding Agents

How to distinguish policy guidance from real runtime interception before coding tools gain wider access.

Read guide

Founder guide

AI Security Check Before Investor Diligence

What founders should document before investors or enterprise customers ask how AI coding tools touch the product repository.

Read guide

Checklist guide

AI Coding Agent Security Checklist

A practical checklist for teams that need to review repository access, protected files, workflow edits, prompts, logs, and policy gaps before agents get more authority.

Read guide

Policy guide

AI Agent Security Policy File Template

What to include in an AI-agent policy file for protected paths, command boundaries, prompt-injection handling, redaction, and approval rules.

Read guide