[abmmhasan]

Pathwise's development setup and release gates depended on an unavailable Foundation 3.0.1 pin. Remove that framework dependency and its unused packages. The comparison, HTTP load and lifecycle fixtures now call Pathwise directly; optional Runwire 2.1.1 context forwarding remains host-owned.

Preserve all existing performance/error/resource budgets, source-isolated comparisons, streamed small/large/range responses, concurrency profiles and request-isolation checks. Production-only CI exercises direct downloads with Foundation, Webrick and Runwire absent. PHP support fixtures, regression references and documentation now describe framework-neutral ownership. An owner with no Runwire binding returns before inspecting the current Fiber; bound-context checks remain active.

Final revision: 954aa9a. Exact-revision CI passed all 19 executed jobs, including Windows, lowest/stable dependency lanes, security/analyzers, adapters, production-only installation, optional Runwire forwarding, release stress and matched performance on both PHP versions.

Validation:

• PHPForge processing, detailed checks and all 13 aggregate checks passed without Foundation/Webrick: 462 tests / 1,613 assertions.
• Production-only direct fixture: 200 valid downloads with Foundation and Runwire absent.
• Runwire forwarding, cancellation, isolated storage and 1,500 request lifecycles passed.
• 87 documented PHP examples passed syntax validation; Sphinx passed with warnings treated as errors.
• Production dependency audit: no advisories or abandoned production packages. Development-only doctrine/annotations remains upstream PHPForge/PHPBench tooling debt.
• Downloaded final artifacts verified 124 candidate source fingerprints and 50 HTTP phases against the checkout and unchanged resource/error budgets.

Hosted PHP	Synthetic RPM change	HTTP RPM change at c1 / c8 / c32 / c64	Bound soak successes
8.4.26	−0.43%	+0.15% / +0.25% / −0.75% / −0.05%	15,193
8.5.11	−0.09%	−0.21% / −0.66% / +0.39% / −0.01%	15,899

All hosted profiles met the unchanged 2% median RPM regression limit. Every HTTP phase had zero response errors and met latency/resource budgets. Both 30-second bound soaks had zero descriptor growth and less than 300 KiB RSS growth.

Local PHP 8.5.4 measurements initially exceeded the c64 limit by about 3%; the guard's targeted repeat met it, but large throughput variance prevents attributing that difference to the guard alone. The final independent hosted matrix supplies acceptance evidence for the defined fixture. These results do not guarantee throughput on other hardware, PHP patches or application workloads; no budget was relaxed.