Compress 6 discovery sections into one Discovery & Retrieval section, remove three-way echo of "drive the artifact" between Goal, Manual QA Gate, and Forbidden stops (now once + cross-ref), split persona prose into Tone vs Autonomy & Collaboration per OpenAI's GPT-5.5 prompting guide, trim Output formatting from 13 prior-restating bullets to 5 environment-specific ones, and defer the Oracle wait policy to the dynamic oracleSection so it lives in a single source of truth.
Add buildFrontendGuidanceSection: conditional injection that only emits the anti-slop frontend block when `visual-engineering` category is unavailable, since the category-skills guide already forces visual work to delegate when the category exists.
Static prose drops 277 -> ~175 lines (-37%); every Hephaestus identity item preserved (Manual QA Gate surface map, Intent table + commitment line, three-attempt failure protocol, Hard invariants, no defensive code, AGENTS.md and dirty worktree handling, four dynamic placeholders).
Address cubic and oracle review feedback to make the Node/Electron
fallback faithful to Bun.spawn behavior so cross-runtime callers
behave identically.
Changes:
- resolveStdio() default stdio is now ["ignore", "pipe", "inherit"]
to match Bun.spawn defaults (was ["pipe", "pipe", "pipe"]).
Prevents hangs in dispatcher and on-complete-hook callers that
did not explicitly set stdin and would otherwise wait forever
for input on Node.
- spawn-with-windows-hide.ts uses the same defaults so the Windows
Node helper aligns with the rest of the shim.
- wrapNodeProcess now rejects proc.exited with the original error
on "error" events (previously swallowed the error and resolved
to exit code 1, hiding ENOENT and friends).
- spawnSync result returns the real result.pid instead of -1 and
exposes stdout/stderr as Buffer | undefined to match Node's
spawnSync result shape when those streams are not piped.
Tests cover the new defaults, real pid surfacing, and missing
executable error propagation.
Refs cubic review and oracle audit on #3798.