issues: file #33 (release flag docs)

diff --git a/.forge/issues/1651beb0-8cd8-4c1f-a7dd-a1f09e089104/issue.md b/.forge/issues/1651beb0-8cd8-4c1f-a7dd-a1f09e089104/issue.mdnew file mode 100644index 0000000..f9bc42b--- /dev/null+++ b/.forge/issues/1651beb0-8cd8-4c1f-a7dd-a1f09e089104/issue.md@@ -0,0 +1,30 @@1+---2+id: 1651beb0-8cd8-4c1f-a7dd-a1f09e0891043+title: 'spec §Release: enumerate flags (--notes/--dry-run); -h may gain one line'4+state: open5+labels: []6+assignees: []7+milestone: null8+created: 2026-09-08T08:35:09.121963230Z9+updated: null10+---11+SOURCE: orchestrator friction minting v0.2.0 (2026-09-08): took 312+attempts to create a release. `--body` -> unknown flag; bare `f0rge13+release v0.2.0` -> "not found"; only the usage ERROR taught the real14+flag (`--notes <file|->`). The doc-truth wave (#30) added per-verb15+-h -> spec pointers everywhere, but spec [Release] says "Body =16+notes" WITHOUT naming the flag, so the pointer lands on a section17+that still hides the surface. Same bug class the swarm flagged in18+#27, blown on my own mint.19+20+FIX (boring, one section):21+- spec [Release] enumerates the exact flags:22+  create <ver> --notes <file|->  (--title S optional)23+  install [ver|main] [--dry-run]24+  (list|view unchanged)25+- release -h may gain ONE flag-summary line IF it fits the 250B/8ln26+  budget (byte-count; if it doesn't fit, spec-only is fine — the -h27+  already points at [Release]).28+- golden: spec/fixtures/release.md still parses; budgets hold.29+DONE: a fresh agent reading `f0rge release -h` + spec [Release] can30+create a release + dry-run an install first-try, zero trial-error.