@cortexkit/aft-bridge 0.57.1 → 0.58.0

package/dist/bash-timeout.js

@@ -1,26 +0,0 @@
-// Shared resolution of the hard-kill timeout sent to the bridge for a bash
-// command. Used by both harness plugins so the semantics stay identical.
-/**
- * Resolve the hard-kill timeout to forward to the bridge for a foreground bash
- * command.
- *
- * A model-supplied `timeout` shorter than the foreground wait window is
- * incoherent: the task would be killed before (or exactly at) the moment we
- * promote it to background, which is never what the caller wants. This is the
- * #102 bug, where a model passed `timeout: 100` and the command was killed at
- * 100ms while `foreground_wait_window_ms` was silently overridden to 100ms by a
- * `Math.min(timeout, window)`.
- *
- * Treat such sub-window values as unset (return `undefined`) so the bridge
- * applies its default 30-minute kill cap and the foreground poll runs the full
- * wait window. A `timeout` at or above the window is a coherent cap and is
- * honored as-is. Callers use the returned value for the bridge payload, the
- * promotion message, and the subagent inline-wait, so all three agree.
- */
-export function resolveBashKillTimeout(modelTimeout, foregroundWaitMs) {
-    if (modelTimeout !== undefined && modelTimeout >= foregroundWaitMs) {
-        return modelTimeout;
-    }
-    return undefined;
-}
-//# sourceMappingURL=bash-timeout.js.map

package/dist/bash-timeout.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"bash-timeout.js","sourceRoot":"","sources":["../src/bash-timeout.ts"],"names":[],"mappings":"AAAA,2EAA2E;AAC3E,yEAAyE;AAEzE;;;;;;;;;;;;;;;;GAgBG;AACH,MAAM,UAAU,sBAAsB,CACpC,YAAgC,EAChC,gBAAwB;IAExB,IAAI,YAAY,KAAK,SAAS,IAAI,YAAY,IAAI,gBAAgB,EAAE,CAAC;QACnE,OAAO,YAAY,CAAC;IACtB,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC"}