The newest Secrets & Credential Leaks resources are on the Secrets & Credential Leaks page. For a reading order instead of a date order, see the Secrets & Credential Leaks learning path.
| Date Added | Link | Excerpt |
|---|---|---|
| 2026-04-16 2026 | Do Not Use Secrets in Environment Variables beginner 22 min read | Reference detailing why secrets should not be stored in environment variables, explaining risks like poor management, difficulty in rotation and auditing, lack of encryption, and leaks through frontend/backend blurring in frameworks like Next.js and Nuxt, as well as `.env` files and Docker images. It highlights the ease of access via `/proc/PID/environ` and the exposure in process lists, proposing better secrets management solutions. |
| 2026-04-16 2026 | Environment Variables Don't Keep Secrets beginner 5 min read | Library offering guidance on secure secrets management, highlighting risks of using environment variables for sensitive data like API keys and database credentials. It contrasts this with more secure alternatives such as memory-backed volumes and secrets files, referencing best practices from the CNCF Cloud Native Security Whitepaper and OWASP, including their Secrets Management and Kubernetes Cheat Sheets. The library also details methods for mitigating risks when environment variables are unavoidable, such as using secrets managers like CyberArk Conjur and injection tools like CyberArk Summon. |
| 2026-04-16 2026 | From .env to Leakage: Mishandling of Secrets by Coding Agents intermediate 4 min read | Writeup detailing secret leakage risks from AI coding assistants, specifically highlighting Claude Code's automatic loading of .env files and potential for API key exposure. The analysis includes a customer incident with Cursor uploading an API key and the CEO's Claude Code committing a Gemini API key to GitHub. Public incidents involving MCP-related data exfiltration via WhatsApp and supply chain attacks are also discussed, emphasizing the need to assume all data is accessible to these agents. |
| 2026-04-16 2026 | Secret Detection in Application Security beginner 7 min read | Library for detecting secrets in application security, addressing the high false positive rates of traditional regex-based scanners and the blind spots in their coverage. It emphasizes pairing secret detection with context, such as ownership, deployment location, and activity status, to enable effective prioritization and remediation. The library explores root causes of secret leaks, including cognitive load, complex pipelines, containerization, shadow IT, and the limitations of pattern matching, while highlighting the need for scalable prevention strategies and governance models. |
| 2026-04-16 2026 | 29 Million Leaked Secrets: How AI Coding Tools Are Making It Worse news 5 min read | Library for securing AI agents by treating them as governed non-human identities, focusing on credential management. It addresses the 28.6 million secrets leaked in 2025, a 34% increase, driven by AI-assisted commits and multi-provider integrations across services like OpenRouter and Hugging Face. The library advocates for scoped permissions, short-lived credentials (like OAuth 2.1), event-driven rotation, and robust revocation capabilities, contrasting with insecure practices seen in Model Context Protocol configurations. → helpnetsecurity.com |
| 2026-04-16 2026 | The State of Secrets Sprawl 2026 - GitGuardian Annual Report news 3 min read | Report analyzing the state of secrets sprawl in 2026, highlighting the acceleration of credential leaks driven by mainstream AI adoption. It reveals secrets are leaking 1.6x faster than developer growth, with AI-assisted commits showing significantly higher leak rates. The report details exposures beyond code repositories, impacting CI/CD runners and collaboration tools, and identifies exposed AI service secrets as a rapidly growing threat category. It also notes the prevalence of long-lived, unvalidated secrets and the challenges in remediation. |
| 2026-04-11 2026 | Terraform Secrets Management Best Practices beginner 9 min read | Library for managing secrets within Terraform configurations. It highlights best practices such as avoiding hard-coded secrets, leveraging secrets managers like AWS Secrets Manager, and utilizing the `sensitive` flag for outputs. The library also details how Terraform states and plans can expose sensitive data and introduces ephemeral resources as a mechanism to fetch secrets without persisting them in the state file. → blog.gitguardian.com |
| 2026-04-11 2026 | AWS IAM Roles Anywhere Workload Identities intermediate 1 min read | Guide to AWS IAM Roles Anywhere workload identity mapping, explaining how certificates with Subject (CN, FQDN, hostname) or SPIFFE identities are used to obtain temporary credentials. It details how certificate fields like Subject, Issuer, and Subject Alternative Name (SAN) are extracted and mapped to PrincipalTags (e.g., `aws:PrincipalTag/x509Subject/CN`, `aws:PrincipalTag/x509Issuer/O`, `aws:PrincipalTag/x509SAN/DNS`) for use in resource policies, allowing direct identity association with compute instances. |
| 2026-04-11 2026 | External Secrets Operator: Introduction beginner | External Secrets Operator: Introduction |
| 2026-04-11 2026 | Google Cloud SIEM Service Account Token Leak news 2 min read | Writeup of a Google Security Operations vulnerability (TRA-2025-52) allowing SIEM tenant service account token leakage. Manipulation of the API root parameter in custom integrations forced job execution through a code path that exposed the highly privileged cloud-platform scoped service account access token to an attacker-controlled server. This enabled privilege escalation and unauthorized actions within the Google Cloud project. → tenable.com |
| 2026-04-11 2026 | Secret Rotation: How It Works beginner 10 min read | Library detailing automatic secret rotation, explaining its importance in reducing vulnerability to compromised credentials and insider threats. It covers rotation conditions, revocation, new secret generation and validation, application across systems like Kubernetes, and types of secrets such as passwords, API keys, SSH keys, and encryption keys. The library also discusses challenges in large-scale implementation and highlights secret management solutions like HashiCorp Vault and StrongDM for automated rotation. |
| 2026-04-11 2026 | Secret Auto Rotation with Secrets Store CSI Driver intermediate 3 min read | Library for automatic secret rotation within Kubernetes pods, enhancing security by periodically updating mounted secrets and Kubernetes Secrets when changes occur in external stores. This feature leverages the CSI RequiresRepublish mechanism and the `--enable-secret-rotation=true` flag for the Secrets Store CSI Driver, removing the need for privileged RBAC permissions. The driver supports applications reading secrets from mounted volumes or container filesystems, with pods needing restarts or tools like Reloader for environment variable updates. This alpha feature, configurable via Helm or command-line flags, relies on providers implementing a gRPC server for communication. |
| 2026-04-11 2026 | Secretless GitHub Actions to AWS via OIDC intermediate 7 min read | Reference detailing secretless connections from GitHub Actions to AWS via OpenID Connect (OIDC). This guide explains how to configure GitHub Actions to generate OIDC JWTs, which can then be used to establish trust relationships with AWS IAM roles. It covers creating an OIDC provider in AWS, setting up trust policies with conditions like `StringEquals` on `token.actions.githubusercontent.com:sub` and `token.actions.githubusercontent.com:aud`, and granting necessary permissions. The article provides examples using AWS CDK and Terraform for infrastructure setup, alongside a sample GitHub workflow leveraging the `aws-actions/configure-aws-credentials` action. |
| 2026-04-11 2026 | OIDC Security Hardening for GitHub Actions intermediate 5 min read | Reference for hardening GitHub Actions security using OpenID Connect (OIDC). This guide details how workflows can leverage OIDC to obtain short-lived access tokens from cloud providers like AWS, Azure, GCP, and HashiCorp Vault, eliminating the need for long-lived secrets. It explains configuring OIDC trust, requesting tokens, and using custom repository properties as claims for attribute-based access control, enhancing authentication and authorization management. Dependabot is also covered, demonstrating its ability to use OIDC for private registry authentication. |
| 2026-04-11 2026 | Hardening HashiCorp Vault Best Practices intermediate 9 min read | Reference for hardening HashiCorp Vault in production, detailing techniques like TLS and mTLS configuration, firewall rules, and operational security practices such as using dedicated service accounts, minimal write privileges, disabling core dumps and swap, and implementing least privilege access control. It also covers additional considerations like avoiding root tokens, utilizing short-lived TTLs, robust audit logging with redundancy via file and syslog, and secure handling of credentials, emphasizing regular updates and patches for vulnerability mitigation. |
| 2026-04-11 2026 | HashiCorp Vault Production Hardening Guide intermediate 8 min read | Library providing best practices for hardening HashiCorp Vault production deployments, emphasizing defense in depth. Recommendations include running Vault as an unprivileged user, allowing minimal write privileges, enabling end-to-end TLS with HSTS, disabling swap and core dumps, using single tenancy, firewalling traffic, avoiding root tokens, configuring user lockout, enabling audit device logs, managing log files, disabling shell command history, maintaining frequent upgrades, synchronizing clocks, restricting storage access, avoiding clear-text credentials with platform-specific identity solutions, using the safest TLS algorithms, following plugin best practices, ensuring consistent configuration file merging, applying correct filesystem permissions, and using standard input for Vault secrets. |
| 2026-04-11 2026 | Leaked Env Variables Allow Large-Scale Cloud Extortion news 21 min read | Writeup of a cloud extortion campaign that successfully compromised and ransomed data by leveraging exposed environment variable files (.env). The campaign exploited credentials found in .env files, coupled with long-lived credentials and a lack of least privilege architecture, to gain initial access to victim AWS environments. Attackers utilized Tor, VPNs, and VPS endpoints for reconnaissance, lateral movement, and data exfiltration, targeting services like IAM, STS, S3, and SES. → unit42.paloaltonetworks.com |
| 2026-04-11 2026 | CVE-2025-68429: Storybook .env Secrets Exposure news 1 min read | Library CVE-2025-68429 details a .env secrets exposure vulnerability in Storybook. Versions prior to 7.6.21, 8.6.15, 9.1.17, and 10.1.10 can inadvertently bundle environment variables from .env files into build artifacts. If a built Storybook is published online, these variables become publicly viewable. Vulnerable projects typically build Storybook in a directory containing a .env file and publish the output. Users should upgrade to a patched version and audit for sensitive secrets in .env files, rotating affected keys and adjusting variable handling by prefixing with STORYBOOK_ or using the env property if necessary. |
| 2026-04-11 2026 | 10K Docker Images Spray Live Cloud Creds news 3 min read | Analysis of 10,456 Docker images reveals over 100 companies, including a Fortune 500 firm and a major bank, inadvertently exposing live cloud credentials and API keys for AI services. Flare's research indicates that secrets are often baked into build contexts and published via informal "shadow IT" accounts, bypassing enterprise monitoring. Even after removal from images, exposed secrets frequently remain active, necessitating the use of secrets management tools and pre-push scanning. → theregister.com |
| 2026-04-11 2026 | 10,000+ Docker Hub Images Leaking Credentials news 2 min read | Analysis of 10,000+ Docker Hub images reveals widespread credential and authentication key leaks affecting over 100 organizations, including a Fortune 500 company and a national bank. Researchers at Flare identified exposed secrets like OpenAI, HuggingFace, and Gemini API keys, GitHub tokens, and database credentials within container images and manifests. Common vulnerabilities stem from `.env` files and hardcoded secrets in application files, impacting cloud environments, CI/CD systems, and Git repositories. While some developers revoked keys within 48 hours, 75% of exposed secrets remained unrevoked, posing ongoing risks. → bleepingcomputer.com |
| 2026-04-11 2026 | Thousands of Secrets Exposed on Docker Hub news 19 min read | Analysis of over 10,000 Docker Hub images revealed leaked production credentials, including AI LLM model keys and cloud environment access tokens, impacting over 100 organizations. This research highlights a shift to "authenticate-in" attacks rather than traditional hacking, exacerbated by shadow IT accounts and insufficient credential rotation after exposure, as seen in incidents like the Shai-Hulud NPM worm and the tj-actions/changed-files GitHub Action compromise. |
| 2026-04-11 2026 | What Happens When You Leak AWS API Keys? beginner | What Happens When You Leak AWS API Keys? |
| 2026-04-11 2026 | CloudKeys in the Air: Exposed IAM Keys Cryptojacking news 16 min read | Analysis of the EleKtra-Leak campaign details automated targeting of exposed AWS IAM keys on GitHub for cryptojacking. Threat actors quickly leverage compromised credentials to launch Amazon EC2 instances for Monero mining. Researchers used a Prisma Cloud HoneyCloud project to monitor this activity, observing hundreds of EC2 instances linked to the operation. The campaign employed automated tools to scan repositories and block identified AWS accounts, with researchers countering by creating randomized, non-attributable AWS accounts with overly-permissive IAM credentials to track actor movements. → unit42.paloaltonetworks.com |
| 2026-04-11 2026 | AWS Customer Security Incidents Repository news | AWS Customer Security Incidents Repository |
| 2026-04-11 2026 | 2,622 Valid Certificates Exposed: Google-GitGuardian Study news 5 min read | Survey of X.509 certificate leaks mapping over 40,000 private keys to TLS certificates, revealing 2,600 valid certificates at risk. The GitGuardian and Google study highlights a critical misunderstanding of private key risks, with many organizations failing to revoke exposed certificates after disclosure campaigns. Techniques used included Certificate Transparency log analysis and OSINT for attribution, uncovering systemic failures in key management and revocation processes across Fortune 500 companies, healthcare providers, and government agencies. → blog.gitguardian.com |
| 2026-04-11 2026 | Secret Scanning in CI Pipelines using Gitleaks intermediate 6 min read | Tool for integrating Gitleaks into GitLab CI pipelines for secret scanning. It details setting up a CI job to run Gitleaks against the OWASP Juice Shop application, demonstrating how to detect and prevent secret leaks like API keys, usernames, and secret keys from being committed. The entry also mentions the use of Git hooks as a pre-commit security measure. |
| 2026-04-11 2026 | Add a Local Gitleaks Pre-Commit Hook intermediate 4 min read | Library for a local Gitleaks pre-commit hook that prevents accidental secrets from entering Git history. It utilizes files within `.git/` for a fast, easily removable safety net, supporting custom configurations via `.gitleaks.toml` and optional team-wide enforcement with Lefthook or GitHub Actions. |
| 2026-04-11 2026 | GitHub Comments Leak Live API Keys news 13 min read | Tool update. TruffleHog now scans GitHub issues, pull requests, and comments, revealing thousands of live API keys and passwords accidentally posted by human users. Many leakers have no association with the repository, and edited comments can still expose secrets in their history. Secrets are primarily found in plain text, not code blocks, and repositories with commented secrets often also have secrets within their git history. → trufflesecurity.com |
| 2026-04-11 2026 | Secret Scanning Encoded and Archived Data intermediate 4 min read | Library for automated secret scanning, TruffleHog detects sensitive data within various encoded string formats like Base64, UTF-8, UTF-16, and Escaped Unicode. It also supports scanning archived files, including Unix archives, Debian packages, RPM, CPIO, and common formats such as .zip, .tar, and .gz, efficiently uncovering secrets hidden within compressed data. → trufflesecurity.com |
| 2026-04-11 2026 | How TruffleHog Verifies Secrets intermediate 4 min read | Library detailing how TruffleHog verifies secrets, moving beyond simple entropy and regex checks. It explains the challenges in programmatically confirming API key validity by testing various endpoints like Doppler's `/v3/me`, handling diverse HTTP responses (including rate limits and error codes), and adapting to API changes and new key types. The library also covers complex verification for database credentials and emphasizes the community's role in maintaining TruffleHog's accuracy and low false-positive rates. → trufflesecurity.com |
| 2026-04-10 2026 | Secret Scanner Comparison: Finding Your Best Tool beginner | Secret Scanner Comparison: Finding Your Best Tool |
| 2026-04-10 2026 | 6 Effective Secret Scanning Tools beginner 8 min read | Library for automated secret scanning that identifies and protects sensitive information like API keys and database credentials exposed in codebases, logs, or configuration files. It supports detection of various secret types, integrates with development workflows and CI/CD pipelines for early vulnerability detection, and provides actionable remediation insights. Specific tools mentioned include Legit Security, GitGuardian, AWS Secrets Manager, TruffleHog, Doppler, and GitLeaks. |
| 2026-04-10 2026 | 8 Best Secret Scanning Tools (2026) beginner 10 min read | Library for detecting hardcoded credentials, API keys, and tokens. It highlights tools like Gitleaks for pre-commit blocking, TruffleHog for live credential verification, and detect-secrets for legacy codebases. GitGuardian is noted as a leading managed platform, offering real-time monitoring and collaboration tool scanning. The library emphasizes the importance of early detection to prevent data breaches and account takeovers, contrasting the cost of pre-commit remediation with post-commit incident response. → appsecsanta.com |
| 2026-04-10 2026 | Best Secret Scanning Tools in 2025 beginner 13 min read | Library for detecting hard-coded secrets in code, configurations, and cloud infrastructure. It utilizes pattern recognition, entropy checks, and AI to identify sensitive data like API keys and passwords, aiming to prevent data breaches by automating detection and remediation. Notable features include context-aware detection correlating secrets with other vulnerabilities, one-click remediation for many secret types, and integration into developer workflows via IDE extensions and pre-commit hooks. The library supports broad scanning across Git repositories, container images, and cloud environments, offering a free tier for basic use. → aikido.dev |
| 2026-04-10 2026 | GitHub Leaked API Keys and Secrets Reference news 3 min read | Reference list of GitHub search queries designed to locate hardcoded API keys and secrets. It includes specific patterns for OpenAI API keys, GitHub OAuth/App/Personal/Refresh Access Tokens, Slack Tokens, Google API keys, Square OAuth/access tokens, and Shopify secrets. The queries utilize file path extensions and keyword combinations to filter results effectively. |
| 2026-04-10 2026 | 23.8 Million Secrets Leaked on GitHub: The Case for Expiring Credentials news 7 min read | Library for managing secrets; addresses the persistence problem where leaked credentials remain active due to lack of revocation, as seen in the 23.8 million secrets found on GitHub, impacting collaboration tools like Slack and Jira, private repositories, and even increasing with AI assistants like GitHub Copilot. This library focuses on built-in expiration for shared credentials, offering a solution beyond traditional secrets managers like HashiCorp Vault or AWS Secrets Manager. |
| 2026-04-10 2026 | GitHub is Awash with Leaked AI Company Secrets news 4 min read | Survey of leaked AI company secrets on GitHub by Wiz reveals that 65% of examined companies exposed sensitive data like API keys and tokens. Leaks were found in deleted forks, gists, and developer repositories, potentially exposing organizational structures, training data, or private models. Companies like Anthropic and ElevenLabs were affected, with plaintext API keys and HuggingFace tokens for private models discovered. This research highlights significant security risks, with Palo Alto Networks data showing AI data security incidents doubling. Wiz recommends proactive secret scanning and establishing clear disclosure channels. |
| 2026-04-10 2026 | The State of Secrets Sprawl 2026: AI-Service Leaks Surge 81% news 5 min read | Report detailing the surge in hardcoded secrets on public GitHub, with AI service leaks increasing 81% and LLM infrastructure leaking 5x faster than core model providers. The analysis highlights insecure patterns in MCP configuration guides, the significant risk of secrets in internal repositories and collaboration tools like Slack and Jira, and the expanding attack surface of developer workstations and CI/CD runners due to AI agents. It also emphasizes the persistent issue of un-remediated valid secrets and the critical need for robust Non-Human Identity (NHI) governance. → blog.gitguardian.com |
| 2026-04-10 2026 | State of Secrets Sprawl Report 2025 news 2 min read | Report analyzing secrets sprawl in 2024, revealing 4.61% of public repositories contained secrets and 31% of breaches involved stolen credentials. GitGuardian's AI identifies more generic secrets than GitHub's Push Protection, which struggles with keys like MySQL and MongoDB credentials. Private repositories are nine times more likely to leak secrets, with widespread exposures found in Slack, Jira, and Confluence. Leaked secrets include AWS IAM, Slack webhooks, and Azure AD API keys, highlighting risks from misconfigurations, non-human identities, and AI tools like GitHub Copilot. Artifactory token leaks and AWS S3 ransomware attacks demonstrate the severe supply chain and operational impact of exposed credentials. |
| 2026-04-10 2026 | AI Frenzy Feeds Credential Chaos news 3 min read | Report from GitGuardian on secrets sprawl reveals 28.65 million hardcoded secrets in public GitHub commits in 2025, with exposure extending to internal repositories, collaboration platforms like Slack and Jira, and self-hosted infrastructure. AI development workflows introduce new credentials for model providers and agent frameworks, contributing to the rise in leaked secrets across code and configuration files. Many exposed credentials remain valid for years, complicating remediation efforts and widening the attack surface. → helpnetsecurity.com |
| 2026-04-10 2026 | GitHub Secret Leaks: 13 Million API Credentials in Public Repos news | GitHub Secret Leaks: 13 Million API Credentials in Public Repos |
| 2026-04-10 2026 | Best Secret Scanning Tools For 2026 beginner 14 min read | Library from SentinelOne scans code repositories for over 750 types of hardcoded secrets, including API keys and cloud tokens, preventing leakage and unauthorized access. It integrates into CI/CD pipelines, supports GitHub, GitLab, and BitBucket, and offers agentless vulnerability scanning with over 1,000 rules. Its CNAPP capabilities include Cloud Security Posture Management, Cloud Workload Security with AI-powered runtime protection, and Cloud Infrastructure Entitlement Management for tightening permissions and reducing alert fatigue. → sentinelone.com |
| 2026-04-10 2026 | 29 Million Secrets Leaked: AI Coding Tools Making It Worse news 7 min read | Library for detecting hardcoded secrets in code, especially concerning AI coding assistants like Copilot and Claude Code. The GitGuardian 2026 State of Secrets Sprawl report highlights a 34% increase in leaked secrets, with AI-generated code exhibiting a 3.2% secrets leak rate, double that of human-only commits. The library integrates with tools like gitleaks and detect-secrets for pre-commit hooks and CI/CD scanning, and emphasizes .gitignore configurations, secrets vault integration with AWS Secrets Manager and HashiCorp Vault, AI tool configuration audits, and secret rotation policies. |
| 2026-04-10 2026 | The State of Secrets Sprawl 2026: 9 Takeaways for CISOs news 4 min read | Library of insights from GitGuardian's State of Secrets Sprawl 2026 report detailing the accelerating growth of hardcoded secrets, with 29 million uncovered in 2025. The analysis highlights AI's impact, the 6x higher leak rate in internal repositories, and secrets found outside code in tools like Slack and Jira. It notes the continued validity of leaked secrets and the emergence of developer endpoints as credential aggregation layers, emphasizing the need for non-human identity governance over simple detection. → thehackernews.com |
| 2026-04-10 2026 | The State of Secrets Sprawl 2025 news 3 min read | Report on the State of Secrets Sprawl 2025 details a significant increase in leaked secrets, with 23.8 million found on public GitHub in 2024, a 25% rise. Generic secrets comprise 58% of leaks, and a troubling 70% of secrets from 2022 remain active, expanding the attack surface. The report highlights that 35% of private repositories also contain secrets, including AWS IAM keys and hardcoded passwords, and secrets are prevalent across the SDLC in tools like Jira and Slack. The U.S. Treasury Department's BeyondTrust breach illustrates the real-world impact of exposed credentials. → blog.gitguardian.com |
| 2026-04-10 2026 | The Complete 2026 Secrets Management Guide beginner 4 min read | Reference for secrets management, detailing techniques like SOPS with age for encrypted Git commits and Infisical for team-scale self-hosting. It highlights strategies to combat "Secrets Sprawl," including least privilege access, automated rotation, and embedding secrets scanning with tools like Gitleaks in pre-commit hooks and CI/CD pipelines, while warning against plaintext storage, shared super accounts, and unexpiring tokens. |
| 2026-04-06 2026 | Secure Code Scanning: Basics & Best Practices beginner 8 min read | Library for secure code scanning, a practice vital for identifying and resolving security flaws, code smells, and data privacy risks across first-party code, third-party libraries, and container images. It leverages techniques like flow-based analysis and pattern matching to detect vulnerabilities such as SQL injection and RCE, citing real-world examples like the MOVEit Transfer and Ollama (CVE-2024-37032) incidents. The library aids in preventing costly data breaches by integrating security into the SDLC and offers Software Composition Analysis (SCA) to manage dependencies and licenses. → wiz.io |
| 2026-04-06 2026 | Zen AI Pentest GitHub Action intermediate 18 min read | Library for an AI-powered penetration testing framework that integrates with GitHub Actions. It executes real security tools like Nmap, Nuclei, and SQLMap within isolated Docker sandboxes, employing safety controls such as private IP blocking and resource limits. The framework features a ReAct pattern with a state machine, memory system, and multi-agent collaboration for automated tool selection, execution, and self-correction, generating comprehensive PDF, HTML, and DOCX reports with evidence collection and chain of custody. |
| 2026-04-06 2026 | Shift Left Security That Developers Actually Keep Enabled beginner 12 min read | Library for architecting security checks into CI/CD pipelines based on speed and false positive rates. This framework assigns checks to stages like pre-commit (under 30s), PR checks (under 2m), and merge gates (under 5m) to maintain developer velocity. It recommends secret scanning and targeted SAST for pre-commit, functional security tests and dependency scanning for PR checks, and broad SAST and container scanning for merge gates, with async tasks for remaining checks. |
| 2026-04-06 2026 | The Claude Code Security Checklist: What the Source Code Reveals intermediate 14 min read | Library for auditing AI coding assistants, particularly Claude Code, by analyzing its leaked source code. The library details specific security controls organizations should verify, covering aspects like discovering all AI deployments, auditing `.mcp.json` files for MCP server connections, bounding bash tool scope, testing path traversal defenses against various attack vectors, and assessing sub-agent spawning permissions. These controls, applicable to tools like Cursor and GitHub Copilot, aim to mitigate risks exposed by the source leak, such as undocumented deployments and misconfigured MCP servers granting access to sensitive data or production systems. |