mcpwn API
    Preparing search index...

    Interface HttpArgs

    Flags shared by the stdio proxy and the http subcommand.

    interface HttpArgs {
        baselineCapture?: string;
        baselineFile?: string;
        detectorPlugins: string[];
        findingsOut?: string;
        guard: boolean;
        interceptRequested: boolean;
        listenHost: string;
        listenPort: number;
        logFile?: string;
        mode: HttpMode;
        onlyTampered: boolean;
        pinFile?: string;
        policyFile?: string;
        redact: boolean;
        redactFile?: string;
        ruleActionPlugins: string[];
        ruleFiles: string[];
        scopeAffectsLog: boolean;
        scopeDirection?: Direction;
        scopeMethods: string[];
        scopeTools: string[];
        taint: boolean;
        templateIds: string[];
        upstream: string;
        verbose: boolean;
    }

    Hierarchy (View Summary)

    Index
    baselineCapture?: string
    baselineFile?: string
    detectorPlugins: string[]

    Custom passive detectors to load (--detector <file.mjs>, repeatable).

    findingsOut?: string
    guard: boolean

    Guard mode (--guard, §3.14): run as a defensive firewall instead of a tamper proxy — in-scope messages are sanitised/blocked per the policy.

    interceptRequested: boolean

    Set when -i was passed (ignored for HTTP, but warned about).

    listenHost: string
    listenPort: number
    logFile?: string
    mode: HttpMode
    onlyTampered: boolean
    pinFile?: string

    Rug-pull PIN lockfile (--pin <file>, §3.14/A3): diff the live tool/ resource/prompt surface against a committed mcpwn.lock.

    policyFile?: string

    Policy file for guard mode (--policy <file>); when absent and guard is on, ./mcpwn.policy.json is auto-loaded, else the baked-in default policy.

    redact: boolean
    redactFile?: string
    ruleActionPlugins: string[]

    Custom tamper-action plugins to load (--rule-plugin <file.mjs>, repeatable).

    ruleFiles: string[]
    scopeAffectsLog: boolean
    scopeDirection?: Direction
    scopeMethods: string[]
    scopeTools: string[]
    taint: boolean

    Provenance tainting (--taint, §3.14/A4): flag untrusted server output that is later reused inside a tools/call argument (injection→action chain).

    templateIds: string[]

    Community templates to load by id (resolved against the bundled gallery).

    upstream: string
    verbose: boolean