SorcX Patchnotes & System Updates
Transparent engineering log covering DLMM on-chain data normalization, quantitative feed upgrades, security hardening, and UI improvements.
Autonomous Squad Architecture, Semantic Vector Memory RAG Expansion, Multi-Enclave Wallet Management, and Anti-Churn Telemetry
Rolled out Milestone Phase 1-4 full-stack architectural upgrades: 8-agent async deliberation council, semantic vector memory RAG store with expanded 1,000-doc disk capacity and dynamic UI limit selector (50/100/200/Semua), Privy TEE embedded enclave wallet switching, primary bot trading vault keypair restoration, Sentinel due diligence audit badges with predatory-dump/sybil veto, and disk space optimization on production VPS.
MAX_VECTOR_DOCS = 1000) and raised API default limit to 100. Added an interactive Capacity Selector (Limit: 50, Limit: 100, Limit: 200, Limit: Semua) to the Virtual Office Semantic Memory Viewer with dynamic count tracking ({displayed} / {total} Memories). Atj7rBTDrWdxbPY5poENprCzhuttWVLM6fPFfr16d9B) across local and VPS environments, maintaining full continuity for on-chain SOL deposits, token sweep operations, and owner-directed withdrawals to the whitelisted cold wallet. POST /api/virtual_office/wallet_mode/update) allowing users to dynamically switch between encrypted local keypair storage and hardware-isolated TEE enclaves without risking secret key leaks to disk. Dev Score, Predatory Dump, Sybil Veto) across pool discovery tables and the AI Council feed, enforcing a strict 20% Sybil cluster threshold and zero-tolerance veto for serial rug-pull deployers. Float Launchpad Screener, Discover Careless Strat Engine Fix, and Universal Table & Yield Guards
Launched dedicated Float ecosystem screener (/float) and integrated Float filtering into the Discover terminal. Fixed Discover Careless Strat candidate enrichment by dynamically resolving pool token age against major quote mints and DexScreener pair timestamps. Standardized platform-wide yield metrics with a universal 999% APR safety cap and token average parity, and implemented auto-update freeze guards across all launchpad accordion tables.
/float), 25-row default pagination view, official branding badges, and instant launchpad filtering in the Discover module with automatic token registry detection. > 3 Days filter before DexScreener enrichment. Added dynamic token age resolution verifying true on-chain age via DexScreener pairCreatedAt and major quote mint anchors (SOL, USDC, USDT, WIF, BONK). > 99,999,999%). Synchronized screener token APR with expanded pool calculations and strictly excluded low-liquidity dust pools (< $1,000 TVL) from token averages. Float, Stonk, Paid, Sunrise, xStocks, Ember, Hooked) now automatically pause whenever rows are hovered or when any expandable DLMM accordion is open, preserving active inspection states without resetting into loading spinners. Reversion of Artificial Annual Multipliers to Align with Native Meteora DLMM Fee/TVL Ratios
Reverted artificial annualization multipliers (x365, x8760, x17520) across all backend models and frontend client modules. Aligned yield metrics directly with Meteora DLMM DatAPI native fee_tvl_ratio percentage conventions (e.g. 48.35% for 24H, 2.68% for 1H, 1.86% for 30M), eliminating distorted multi-thousand percent figures on volatile meme pools.
Harmonization of 24H APR & Daily Fee Yield Across Backend Services and Frontend Views
Comprehensive mathematical audit and remediation of 24H APR and daily fee yields. Clarified Meteora DLMM unannualized fee_tvl_ratio (daily fee/TVL) vs True Annualized APR ((fee_24h/tvl) * 365), implemented dual-metric presentation with daily yield subtext across all 12 frontend views, enforced strict $1,000 TVL floor, and corrected multi-day return projections in calculator.js.
Root-Level volume_1h & Multi-Timeframe Volume Telemetry for Grok Bot Slow Cook Hype Detection
Exposed volume_1h and volume_1h_usd directly on root pool objects in GET /api/v2/pools/{address} and batch /api/v2/pools without requiring include=volume, added volume_1h alongside volume_1h_usd on /api/v2/pools/search, supported volume_1h aliases in /api/v2/snapshots, and standardized token hype telemetry to enable real-time hype spike calculations.
API v2.0.3 `order_by` Field Alias & `/snapshots` `meta.missing` Completeness
Added smart disambiguation for `order_by=<field>` / `sort_key=<field>` on `/api/v2/pools/search` and populated `meta.missing` for `null` fields across `/api/v2/snapshots` and JSON `/api/v2/export`.
order_by=<field> (e.g. order_by=tvl&order=asc) and sort_key=<field> seamlessly alongside sort=<field>:asc|desc and order_by=asc|desc. null column names (such as fee_24h_pct and fee_apr_24h_pct on < $1,000 TVL pools) inside meta.missing on /api/v2/snapshots and JSON /api/v2/export. API v2.0.2 Search Ordering, `max_tvl` Filter & Canonical/Causal `age_source` Validation
Added standalone `order=asc|desc` and `max_tvl` / `max_mcap` support on `/api/v2/pools/search` (explaining and exposing `< $1,000` TVL pools directly on Page 1) and hardened `age_source="mint"` with canonical Solana genesis timestamps (WSOL, USDC, USDT, LSTs) plus on-chain pool causality checks.
order=asc|desc (and order_by=asc|desc) alongside sort=field:asc|desc, plus max_tvl, max_mcap, sort=age_hours, and sort=mcap_usd. Confirmed zero hidden $3k server TVL floor: low-TVL pools (tvl_usd < $1,000) previously sat on Page 2 (rows 51..76) because fee_apr_24h_pct = null sorts last on default fee_apr_24h:desc and standalone order=asc was ignored. CANONICAL_MINT_BIRTHS for native/system/blue-chip Solana mints (WSOL 2020-03-16T14:29:00Z, USDC, USDT, mSOL, JitoSOL, JLP, JUP, etc.) and enforced an on-chain causality check rejecting late RugCheck detectedAt re-index timestamps whenever a DEX pair or DLMM pool existed earlier. API v2.0.1 Data Precision Hotfix — Dust-TVL Yield Guard, `age_source` Provenance, Rolling 7d Fees & Hype Labels
Resolved three live data edge cases surfaced during Grok Dev Team v2 adapter integration (`1.69e9` dust-TVL yield ratios, `age_hours` mint vs pool provenance, and `fees.7d_usd < fees.24h_usd` UTC rollover lag) and added native `hype.label` / `hype_label` classification.
$1,000 minimum TVL floor across /api/v2/pools, /api/v2/pools/search, /api/v2/market/aggregate, and v2_pool_snapshots. Pools with tvl_usd < $1,000 (such as drained pools with $0.00003 dust TVL that rounded to 0 while dividing residual 24h fees into 1.69e9%) now return null for fee_24h_pct (fees.fee_tvl_24h_pct) and fee_apr_24h_pct (fees.fee_apr_24h_pct) and record them in meta.missing. age_source ("mint" | "earliest_pair" | "pool" | null) alongside age_hours on /api/v2/tokens and /api/v2/pools/search, backed by a persistent on-disk mint birth dictionary (sorcx_v2_mint_births.json) continuously warmed by the 5-minute snapshot daemon and parallel RugCheck/StonkFun lookups. /volume/history?num_of_days=7 summation (where today's UTC bucket reports 0 until midnight rollover and returns 10 rows) with a true rolling 7-day calculation combining live rolling fees.24h ([now-24h..now]) with non-overlapping UTC daily buckets (d-1..d-7) and lifetime cumulative_metrics.fees bounds, guaranteeing fees.7d_usd >= fees.24h_usd. hype.label on /api/v2/tokens and hype_label on /api/v2/pools/search aligned to Slow Cook thresholds: >= 80 BREAKOUT, >= 60 SURGING, >= 35 STEADY, < 35 COOLING. Quantitative Feed API v2.0 — Slow Cook v1.1 Suite, On-Chain BinArray Open-Cost & Historical Snapshots
Launched the complete `/api/v2/*` endpoint suite engineered for Slow Cook v1.1 autonomous DLMM operations: 11 standardized UTC ISO-8601 endpoints, live Solana PDA BinArray rent verification, anti-rug security audits, 7-day 5m/1h OHLCV, 240 req/min rate limits, and 5-minute append-only snapshots.
BinArray PDAs (["bin_array", lbPair, index]) in pure PHP (RFC 8032 Ed25519 curve verification) and queried Solana RPC live to separate non-refundable BinArray initialization costs (0.07143744 SOL) from refundable PositionV2 rent (0.05740608 SOL per 70-bin position) with rpc_slot proof. transferFee, transferHook, permanentDelegate), LP burn/lock percentages, top-10 non-pool/non-CEX holder concentration, creator wallet & rug history, insider wallet clusters, and verdict_hint (pass | warn | fail). /pools/search (filtering pumpfun/stonkfun launchpad, age_hours, mcap_usd, fee_24h_pct, plus batched -80% SOL range refundability), batch endpoints (?mints= & ?addresses= up to 50 items = 1 request), /pools/{address}/ohlcv (5m/1h up to 7d with mcap_close_usd), /prices/sol, and /market/aggregate. v2_pool_snapshots without survivorship bias, supporting cursor-paginated JSON and text/csv exports. X-Data-Updated-At, X-Api-Version: 2.0.0, and Retry-After headers while keeping /api/v1/grok/* (v1.5.0) online seamlessly. Multi-Timeframe Volume (`volume_1h`) & Hype Surge Telemetry (Meteora + DexScreener)
Added native `volume_1h`, granular rolling timeframe metrics (5m to 24h), orderflow buy-pressure, and hype surge scoring by fusing Meteora DatAPI with batched DexScreener pair telemetry.
volume_5m, volume_30m, volume_1h, fees_30m, fees_1h, and full timeframes (5m, 30m, 1h, 2h, 4h, 6h, 12h, 24h) across /pools, /pool/:address/state, /market-metrics, and /token/:address. volume_1h_vs_24h_avg (1h pace vs 24h hourly average), volume_30m_vs_1h_pace (intraday acceleration), composite hype_score (0..100), and hype_label (BREAKOUT, SURGING, STEADY, COOLING). curl_multi DexScreener pair enrichment (45s cache) providing txns_5m, txns_1h (buys/sells), buy_pressure_1h_pct, price_change_pct (m5, h1, h6, h24), plus token holders & market_cap_usd. sort=volume_1h, sort=fees_1h, and sort=hype_score support on GET /api/v1/grok/pools. Live Data Normalization, On-Chain Max Fee Decoder & Yield Sanity Guards
Resolved upstream Meteora DatAPI anomalies affecting apy_24h integer overflows, missing max_fee_pct values, apr_24h scaling, and non-ASCII spoofed pool symbols.
apy_capped boolean indicator. max_fee_pct = 0 omission in Meteora DatAPI by decoding raw Solana LbPair StaticParameters (bytes 8..35: base_factor, variable_fee_control, max_volatility_accumulator) via batched getMultipleAccounts RPC with persistent caching and 10.0% (MAX_FEE_RATE) protocol fallback. apr_24h calculation to true 365-day annualization ((fees_24h / tvl) * 365 * 100) across /pools, /market-metrics, /token/:address, and /pool/:address/state, correcting the 3.65x underestimation caused by unannualized upstream daily ratios. dlmm_pools and hardened Dlmm_model and Orca_model ingestion pipelines. /patchnote changelog page (with ?format=json feed support) to track all platform updates and engineering fixes transparently. Agent-Accessible API Specification & Sliding-Window Quota Telemetry
Enabled direct machine-readable specification access for automated AI agents while keeping key management strictly gated behind authenticated developer sessions.
X-RateLimit-Limit (60 req/min), X-RateLimit-Remaining, and X-RateLimit-Reset headers alongside structured HTTP 429 schemas. Developer API Key Portal & Google Firebase Authentication
Introduced self-service personal API key management powered by Google Firebase Authentication and RS256 JWT verification.
/login verified server-side against Google SecureToken X.509 certificates and HMAC-SHA256 HttpOnly session cookies. sx_live_... API key issuance, instant rotation, revocation, SHA-256 hashed storage, hash_equals() timing-safe verification, and per-key request counters. Explorer Navigation Streamlining & Quantitative Feed Architecture
Focused the public SorcX experience on core DLMM pool discovery and launchpad trackers while establishing dedicated v1 quantitative feed endpoints.
/api/v1/grok/* endpoints delivering pool scout scores, multi-timeframe ecosystem volatility regimes, token liquidity depth audits, and discrete LB Pair bin ladders.