This log records dataset releases, corrections, evidence-status changes, and whether numeric values changed. Product features and editorial rewrites are excluded unless they alter how value data is interpreted.
September 4, 2026 · Crop mutation evidence release
gag1-mutation-stack-source-audit-2026-09-04-r1
Added an eleven-claim, seven-source audit with five unresolved multiplier conflicts. The additive bracket is documented separately from the dated community values; Wet, Moonlit, Frozen, Shocked and Cyclonic disagreements remain visible.
The calculator now supports reusable scenarios and editable reference shortcuts. It does not infer mutation compatibility, resolve conflicting values or provide a live trade price.
September 2, 2026 · Crimson Thorn evidence release
crimson-thorn-source-audit-2026-09-02-r1
Replaced a stale single-price guide with a fourteen-claim source audit that keeps the 10B, 10T and dated 8B Daily Deal figures separate. Three unresolved conflicts remain visible instead of being averaged into one unsupported answer.
Added an editable break-even calculator and reusable scenario links. Results use the player's current cost, sell value, daily produce and already recovered Sheckles; they are estimates, not promises of profit or live stock information.
Added a live side-by-side comparison of the original Grow a Garden and Grow a Garden 2. Both records use matching Roblox Games and votes fields fetched together, with a dated checksummed fallback containing two official universe records.
Current CCU remains separate from lifetime visits and favorites; no DAU, peak, revenue, retention, launch date or progress-transfer rule is inferred.
July 30, 2026 · first-party Roblox statistics release
gag2-roblox-stats-snapshot-2026-07-30-r1
Added a live Roblox-backed view of concurrent players, cumulative visits, favorites, votes, approval rate and server capacity. The response carries its fetch time and falls back to a checksummed first-party snapshot when Roblox is unavailable.
The release deliberately does not infer DAU, unique visitors, revenue, retention, rank or historical peak from these public fields.
Compared ten defender-focused pages and retained sixteen first-party, repeated, limited, conflicting, and boundary claims. The release separates physical walls and one-entrance advice from active presence, recovery gear, pets, crops, trusted players, and private-server exposure control.
Added a local player-input budget scenario and eight-layer coverage checklist. The Roleplay Crate Wood Wall conflict remains visible at 10.53% in the dated multi-source audit versus 15.53% in one current defense article; no layer is called theft-proof.
Added a fourteen-claim source audit and local-only invite-link checker. Roblox's first-party setup, access, pricing-control, and credential rules are kept separate from limited community claims about free access, disabled stealing, server resets, and public-server event rates.
The checker parses HTTPS, the exact Roblox hostname, and common server markers without following or uploading the URL. It does not certify the sender, redirects, or account security, and this site does not distribute public private-server invitations.
Merged the 43-row dated crop-value release with 25 exact-name Seed Shop records. The new crop database exposes rarity, seed price, listed chance, Sell Stand reference, weight and harvest type while preserving ten dated value conflicts.
Five current crop SERP pages publish scopes from 30 to 73 rows. The widest rosters mix ordinary crops with variants, testing entries, secrets and TBA names. This release imports none of those competitor-only rows and labels missing shop coverage as outside the snapshot, not unobtainable.
Added the July 25 Pet Safari guild competition as the twelfth update window after four current pages repeated the round, rarity-point rows or variant multipliers. This supersedes the archive's earlier statement that July 19 was the latest repeated community window; it does not claim an official patch version or a live active-round feed.
Compared three scope models and eighteen claims across ten pages. The release preserves the on-vine versus world-fruit conflict, uses the repeated 80% loss as an editable scenario, and calculates whether Watering Can recovery beats selling decayed immediately.
Compared three capacity routes and eighteen claims across ten pages. The release preserves the launch-era versus paid-upgrade conflict, labels 325 as limited, and calculates sell trips without hiding Turtle's reported movement penalty.
Compared two duration models and eighteen claims across ten pages. Seven sources publish a two-minute Night, while two publish around four minutes. The timer requires a live HUD observation and never labels its browser projection a server feed.
Compared three upgrade targets and sixteen claims across ten pages. The repeated three-to-five-slot path is separated from the disputed sixth slot, while the planner protects an editable Sheckle reserve and estimates sessions to close any shortfall.
Compared five sprinkler tiers and eighteen claims across eleven pages. Common and Uncommon prices align, while Rare, Legendary and Super prices conflict. The editable planner uses observed before-and-after harvest values instead of fabricated mutation odds.
Compared Vote, Gold Carrot, Lucky Swipe, Carrot Contest and account-connection claims across eight pages. Added an editable target planner, primary-policy identity checks and a privacy checklist without inventing reward or contest odds.
Compared four similarly named systems and thirteen claims across eight pages. Standard Daily Quests remain conflicting, the weekly Guild competition is better supported, and single-source reset, milestone and event claims stay explicitly limited.
Published three CORS-enabled, read-only JSON resources exposing 115 acquisition, crop-value and code records. Responses preserve release IDs, evidence metadata, source dataset paths and checksum manifests; OpenAPI 3.1 documents four GET operations.
Target confidence without calling expected rolls a guarantee
Added 11 pack records, ten probability presets and five conflict checks across ten pages. The tool calculates hit and miss chance, separates paid, free, TBA and unavailable claims, and keeps rebalanced community tables labeled unverified.
All 23 official records, with progress saved locally
Added the complete Roblox API roster, dated award totals and four conflict checks across three ranking guides. The checklist fixes MEGA-versus-HUGE, Carrot-versus-First-Seed and watering-window errors without turning editorial tactics into requirements.
Added nine findable NPC or place records and 24 claims across ten sources. The finder preserves stand-label conflicts and exposes unsupported secret-merchant, quest-giver and crafting claims.
Added 25 claim-level records across 11 sources. The planner uses player-entered cash, cycle values, reserve and server exposure to choose an ordered next step while preserving published conflicts.
Added eleven content windows and 29 archive claims across Roblox metadata and nine current update pages. The searchable archive keeps adjacent-day reports, incompatible version labels and the limited July 19 Explorer window visible instead of normalizing them into a false official changelog.
Event synchronization without an invented official time
Added 23 claim-level records across official Roblox metadata and eight current pages. The calculator accepts an in-game remaining time or exact trusted local timestamp, converts seven regions and does not collapse Explorer's Fall Harvest countdown into an unverified Admin Party schedule.
User-synchronized restock windows without a fake live feed
Added 19 claim-level records across eight current pages. The planner keeps the five-minute Seed Shop default editable, exposes future local check windows and summarizes a player-entered observation sample with a Wilson 95% interval. It does not show current stock or predict future items.
Added 18 claim-level records across official Roblox metadata and five community pages. The calculator accepts low, typical and high observations, exposes worst and best cases, and does not turn a mailbox gift or public drop into protected escrow.
Added 20 claim-level records and four progression stages across nine money and beginner guides. The calculator measures risk-adjusted net per cycle, profit per minute, reserve, cycles and time to one user-entered target.
Status: archived · Evidence grade D · 103 records · 0 raw trade observations · 0 current verified values. File integrity is checksum-verifiable; current-market accuracy is not established.
Added 18 claim-level records and five conflict groups across eight community pages. The calculator compares Daily Deal with Sell Stock and models Double or Nothing risk without normalizing the unresolved 2.2× and 5× claims.
Status: editorial consensus · 14 shared pet records · 3 dated source rankings · 4 unanimous S-tier records · 1 disputed record · 0 publisher in-game tests. Trade value is out of scope.
Status: reported active · Confidence: moderate · 3 code records · 3 named secondary sources · 0 publisher in-game redemptions. This is a dated source-consensus snapshot, not a live official feed.
Added 16 claim-level records across seven community pages, including repeated night, owner-away, carry-home, teleport and Shovel mechanics plus limited wall and guild guidance. The “locked” versus “deterred” wording remains explicitly conflicting.
Current Pet Safari rewards and historical competition drift
Added 16 setup, role, competition, reward and defense claims, eleven latest-reviewed reward bands, and six weekly competition snapshots. Older Biggest Plant/Ice Serpent rules remain historical, while the one-source 1,001–5,000 Pet Safari band is explicitly marked.
Dutch-auction mechanics, item coverage and timing conflicts
Added 18 claim-level records, 12 non-exhaustive item mentions and four system comparisons across seven community pages. The dominant 30-minute reset remains separate from one 20-minute claim, and Robux and sniping-threshold conflicts stay visible.
Stopped applying Eclipsed unless selected. Corrected combination expressions so a single ingredient does not activate a multi-ingredient mutation. Historical base prices and multipliers remain unchanged; calculated results can change. Added observed XP-rate input and community-source Dragonfly/Raccoon guides, keeping current quotes separate from the 2025 archive.
release: crop-input-rules-2026-09-16
codes-snapshotNumeric values unchanged
Published a GAG2 code-status snapshot covering three reported-active codes, three named secondary sources, source dates, explicit verification limits, and a SHA-256 file checksum.
release: gag2-codes-2026-07-30-r1
metadataNumeric values unchanged
Published the first machine-readable release manifest, checksums, coverage counts, field definitions, and evidence grade.
release: gag1-pets-2025-08-20-r1
correctionNumeric values unchanged
Normalized relative update strings to the known 2025-08-20 observation timestamp and stopped presenting demand as sample count.
release: gag1-pets-2025-08-20-r1
snapshotNumeric values changed
The three bundled GAG1 pet batch files were recorded in repository commit 9243eec.
release: gag1-pets-2025-08-20-r1
What triggers a data entry?
A new or replaced source file.
A numeric value, currency, or item attribute changes.
An evidence grade or verification status changes.
A correction changes how a field must be interpreted.
A release becomes current, superseded, or archived.
What every future entry must show
An absolute date and stable release ID.
Whether numeric values changed.
The evidence or methodology impact.
A machine-readable manifest and file hashes.
Patch context when it is actually known.
Corrections are evidence-led
A complaint or screenshot is a review candidate, not an automatic price update. A current range requires comparable item attributes, currency, observation time, source type, and enough corroboration to support the published statistics.