[Louie0806]

(ai: deepseek/deepseek-v41-flash)

Why

lifecycle-ledger-guard is a required check and it is red on master itself, so it is red on every branch that merges master:

FAIL: 2 BILI_* env var(s) read from src/ but NOT in LIFECYCLE.md:
  - BILI_SESSION_CONTENT_STORE_PIN
  - BILI_SPAWN_FINGERPRINT

Both reads are legitimate and already documented at their definition sites; only the ledger rows are missing. This PR adds them. Filed as #5313.

Because the gate is red on master, reviewers currently cannot use it to distinguish "this branch broke something" from "master was already broken" — and it is concretely blocking unrelated fixes: #4785 (the draftTailKind one-word fix that closes #4764, the <summary> draft-envelope root cause) fails this job on BILI_SESSION_CONTENT_STORE_PIN alone (job 124424975017).

Changes

LIFECYCLE.md, two table rows, no code:

row	section	read at
BILI_SPAWN_FINGERPRINT	§2.2 internal spawn channel, after BILI_LAUNCHER_LANE	src/instance.ts:308 SPAWN_FINGERPRINT_ENV
BILI_SESSION_CONTENT_STORE_PIN	§2.4 session pool / GC / persistence, after BILI_SESSION_RESIDENT_HARD_MAX_BYTES	src/knobs.ts:444 contentStorePinned()

• BILI_SPAWN_FINGERPRINT ([Bug] Plugin bootstrap spawns a new instance while the old one still holds the port and drops proxying (goes direct): la | [Bug] 插件 bootstrap 在旧实例仍占端口时新起实例并放弃代理（goes direct）：大会话随即以未压缩历史撞上游上限（400 Range of input length / 413） #4125, merged as [PR] Two installations differing by one version no longer deadlock on the same port; traffic is no longer sent in the clear ( | 两份安装差一个版本不再死锁在同一端口上,流量不再裸奔(#4125) #5303): the spawner passes its entry-script sha256 to the child so a rotate/attach gate never judges one side's digest against the other side's twin; a malformed or absent value falls back to a local re-hash.
• BILI_SESSION_CONTENT_STORE_PIN ([Performance] v0.1.208 long sessions exceed 2 GiB memory and spend 14–22 seconds in local processing #4788 phase 2): =1/true keeps a session's CCR content store ({byRef} / {byHash} originals) resident in memory instead of folding it away after the save lands; re-materialises from the unified ccr mirror or the .content-store.json sidecar on first touch (payload-first write order, CCR content-store companion write is non-atomic — a crash mid-write silently destroys all retrievable originals #2674). Unset / 0 / false = fold (default).

Verification

gate	result
node scripts/lifecycle-ledger-guard.mjs	OK: every BILI_* env var read from src/ is registered in LIFECYCLE.md (171 read / 174 ledger / 16 §3 exempt)
negative control: check out upstream/master -- LIFECYCLE.md	same command re-reports both names — the assertion is not loosened
node tools/gen-config-docs.mjs generate + git diff --exit-code	exit 0 (docs-sync clean)
node tools/gen-config-docs.mjs check-coverage	coverage ok: 193 env names + 59 top-level keys
node scripts/config-surface-guard.mjs	config-surface coverage OK
node scripts/check-unused-exports.mjs	clean (6 waived)
npx tsc --noEmit	clean

No existing open PR fixes this. The only other open PR touching LIFECYCLE.md is #5309 (the BILI_DSH_MAX_TESTED_VERSION row) — different section, different line, no conflict.

Same shape as two prior docs-only incidents: 6932ab48c (#2222 → BILI_CODEX_STATE_MODE) and afd6287de (#4308 follow-up).

Closes #5313.

[ranxianglei]

关单说明:本 PR 的内容(LIFECYCLE.md 补 BILI_SPAWN_FINGERPRINT / BILI_SESSION_CONTENT_STORE_PIN 两行登记)已由 #5327 逐字并入 master(f9cec9f),lifecycle-ledger-guard 随之转绿。留它只会冲突白烧 CI。—— 由 #5327 交付方代为关闭。