secure-codinglisted

# Secure Coding ## Purpose `threat-modeling` decides what to defend and `client-server-trust` decides where authority lives. This skill is the implementation layer: the coding practices that stop a correct design from being undermined by the code that implements it. Most exploited vulnerabilities are not exotic. They are a string concatenated into a query, a secret in a config file, an error message that reveals internal structure, or a dependency nobody updated. ## When to use - Writing or reviewing code that reads external input: network, files, user fields, configuration, environment. - Building any command, query, path, URL, or markup from values. - Handling credentials, tokens, keys, or personal data. - Deserialising data from anywhere you do not fully control. - Implementing permission checks. - Adding or upgrading a dependency. ## When NOT to use - Deciding *which* threats matter. Use `threat-modeling` first, or you will harden uniformly and misallocate effort. - Designing authority placement. Use `client-server-trust`. - Designing cryptographic protocols or primitives. Use vetted libraries; if the design genuinely requires new cryptography, it requires a specialist. ## Required context | Fact | Why it matters | |---|---| | Which inputs are attacker-controlled | Everything crossing a trust boundary is | | Which interpreter each value reaches | SQL, shell, HTML, path, and template each need different handling | | Where secrets currently live | Config file