diff --git a/docs/BACKLOG.md b/docs/BACKLOG.md
index ae204751..dc6207e1 100644
--- a/docs/BACKLOG.md
+++ b/docs/BACKLOG.md
@@ -2027,7 +2027,7 @@ lane; demand-gated on a first enterprise Windows/AD deployment.
 
 ## 156. Alert hysteresis (separate fire/clear thresholds)
 
-> 🔢 **Re-scored 2026-08-03 → DEMAND-GATE.** Value **2/10** · Difficulty **3/10** · _fill-in_. Anti-flap refinement the shipped `realert_seconds` / per-rule `cooldown_seconds` throttle already damps (`messagefoundry/config/settings.py:2678`, `:2823`), with single-sided `min_depth`/`min_oldest_seconds` matching confirmed at `messagefoundry/pipeline/alert_sinks.py:617-623`; two new AlertRule fields plus clear-edge state in the sink, no store or migration.
+> 🔢 **Re-scored 2026-08-03 → DEMAND-GATE.** Value **2/10** · Difficulty **3/10** · _fill-in_. Anti-flap refinement the shipped `realert_seconds` / per-rule `cooldown_seconds` throttle already damps (`messagefoundry/config/settings.py:2678`, `:2823`), with single-sided `min_depth`/`min_oldest_seconds` matching confirmed at `messagefoundry/pipeline/alert_sinks.py:617-623`; two new AlertRule fields plus clear-edge state in the sink, no store or migration. **Reviewed 2026-08-10 -- deferral stands, and this is the cleanest deferral in the set.** Its trigger names an observation nobody can make yet: *operators reporting flapping the flat realert/cooldown throttle does not damp*. That cannot fire from analysis, only from watching real alerts misbehave. The covering mechanism (`realert_seconds` + per-rule `cooldown_seconds`) is genuine rather than a fig leaf, and building a deadband now means choosing two threshold constants with no traffic to choose them from -- the shape of change that gets tuned once in anger and never revisited.
 
 > **On-trigger / demand-gate.** Numbered for tracking only — build when the trigger below fires (“demand-gate, don’t schedule”).
 
@@ -2204,8 +2204,8 @@ lane; demand-gated on a first enterprise Windows/AD deployment.
 
 ## 172. Gzip/zip compression codec + file-connector option
 
-> 🔢 **Re-scored 2026-08-03 → DEMAND-GATE.** Value **5/10** · Difficulty **3/10** · _fill-in_. File-feed parity breadth with a clean code-first workaround: the reusable codec shipped including `zip_compress`/`zip_decompress` (`messagefoundry/parsing/compression.py:40-48`), so a zip-delivering partner is served by a Handler call today. What remains is connector-level — widening `_SUPPORTED_COMPRESSION` (`messagefoundry/transports/file.py:88`), which forces an archive-member-to-message decision, plus REMOTEFILE, which has zero compression to extend.
-> **On-trigger / demand-gate.** Numbered for tracking only — build when the trigger below fires (“demand-gate, don’t schedule”).
+> ⛔ **DECLINED 2026-08-10 -- the remainder is a DECISION, not a gap. The codec SHIPPED; what is declined is auto-extraction on the CONNECTOR.** All six functions (`gzip_compress`/`_decompress`, `zip_compress`/`_decompress`, `deflate_*`) are exported from the TOP-LEVEL `messagefoundry` package, so a Handler compresses or extracts in one line, and binary rides the store via the ADR 0028 `mfb64:v1:` carriage. The connector's `gzip`-only restriction is defended at THREE layers -- the wiring `Literal['gzip'] | None`, `_SUPPORTED_COMPRESSION`, and a validator that raises on `'zip'` -- and those are deliberate: **a gzip is one stream, a zip is an ARCHIVE.** Auto-extracting would make the ingest path decide which member is the message, on a partner-supplied file this project treats as untrusted, with traversal entries and decompression bombs to adjudicate. Two details show the split was designed rather than unfinished: the decompress functions take a `max_output_bytes` BOMB CEILING, and `zip_compress` takes a mapping of NAME to bytes, so an outbound zip needs an entry name only the feed author can choose. **Carried open, this invites someone to remove three guards to 'finish' the item -- and what they would switch on is automatic extraction of partner archives at the front door.**
+> **No longer demand-gated -- declined by design. Zip is a Handler job, deliberately.**
 
 > **AMENDED 2026-07-28 — the codec is BUILT; the connector covers gzip only.** Adversarial verification refuted a full close. **BUILT:** the pure three-algorithm compression codec (`messagefoundry/parsing/compression.py`, Handler-callable) and the File connector's gzip/gunzip option ([ADR 0123](adr/0123-compression-codec-gzip-zip-deflate-file-connector-compress-decompress-option.md)).
 >