solana-traderclaw 1.0.133 → 1.0.135

package/config/gateway-v1-upgraded.json5

@@ -1,6 +1,6 @@
 // OpenClaw Gateway Configuration — TraderClaw V1-Upgraded (Single Agent)
-// Single "main" agent with 5-minute heartbeat + 10 consolidated cron jobs.
-// Consolidated from 13 → 10: portfolio-health (dead-money+whale+risk-audit),
+// Single "main" agent with 5-minute heartbeat + 11 preset TraderClaw cron jobs.
+// alpha-stream-resubscribe (every 20m WS keepalive); portfolio-health (dead-money+whale+risk-audit);
 // trust-refresh (source-reputation+deployer-trust). Shortened prompts.
 {
   agents: {
@@ -22,23 +22,25 @@
       keepLines: 2000
     },
     jobs: [
-      // ── Strategy & Learning ───────────────────────────────────────
+      // ── Alpha stream keepalive ────────────────────────────────────────
       {
-        id: "strategy-evolution",
-        schedule: "0 */4 * * *",
+        id: "alpha-stream-resubscribe",
+        schedule: "*/20 * * * *",
         agentId: "main",
         sessionTarget: "isolated",
         delivery: { mode: "none" },
-        message: "CRON_JOB: strategy_evolution\n\nStep 1: Call solana_journal_summary to get aggregate performance stats (win rate, avg PnL, trade count). If fewer than 10 closed trades since the last strategy evolution, skip weight updates but still run pattern detection.\n\nStep 2: Call solana_memory_search for 'strategy_evolution' — find last 3 evolution cycle results. Call solana_memory_search for 'strategy_drift_warning' — find drift warnings since last evolution. Call solana_memory_search for 'pre_trade_rationale' — recent decision patterns.\n\nStep 3: Run Recurring Pattern Detection — search for learning_entry tags, group by area, check linked chains (3+ = confirmed pattern), investigate drift warnings.\n\nStep 4: Call solana_strategy_state to read current feature weights.\n\nStep 5: Call solana_trades to get recent closed trades. Apply ADL checks (direction consistency, weight velocity, reversion check).\n\nStep 6: Compute proposed weight changes. Score each with VFM (Frequency + Failure Reduction + Self-Cost). Only apply changes scoring >= 3/5.\n\nStep 7: Verify guardrails: maxDeltaOk, sumWeightsOk, minTradesOk, floorCapOk. If all pass, call solana_strategy_update with incremented version.\n\nStep 8: Run Named Pattern Recognition — search for winning trade clusters, catalog new patterns, evolve existing ones.\n\nStep 9: Evaluate discovery filter performance. Log all results via solana_memory_write with tags: strategy_evolution, vfm_scorecard, pattern_detection, named_pattern.\n\nFORMATTING RULES:\n- Every token reference MUST use SYMBOL (full_CA) format.\n- Do not execute trades. Do not ask questions.",
+        message: "CRON_JOB: alpha_stream_resubscribe\n\nKeep alpha WebSocket subscription healthy. solana_runtime_status — if subscribed is false: solana_alpha_subscribe. If subscribed but ingestionStale: solana_alpha_subscribe force:true (zombie heal). Else: reply ALPHA_KEEPALIVE_OK.",
         enabled: true
       },
+
+      // ── Strategy & Learning ───────────────────────────────────────
       {
-        id: "alpha-scan",
-        schedule: "0 */3 * * *",
+        id: "strategy-evolution",
+        schedule: "0 */4 * * *",
         agentId: "main",
         sessionTarget: "isolated",
-        delivery: { mode: "announce", channel: "last", bestEffort: true },
-        message: "CRON_JOB: alpha_scan\n\nScan new launches, filter, score, log alpha. Tools: solana_scan_launches → filter (vol>30K, mcap>10K, liq>5K) → solana_token_snapshot for survivors → quality filter (top10 <50%, deployer <3 abandoned, has social) → score 0-100 → solana_alpha_log for 65+. Summarize results.",
+        delivery: { mode: "none" },
+        message: "CRON_JOB: strategy_evolution\n\nStep 1: Call solana_journal_summary to get aggregate performance stats (win rate, avg PnL, trade count). If fewer than 10 closed trades since the last strategy evolution, skip weight updates but still run pattern detection.\n\nStep 2: Call solana_memory_search for 'strategy_evolution' — find last 3 evolution cycle results. Call solana_memory_search for 'strategy_drift_warning' — find drift warnings since last evolution. Call solana_memory_search for 'pre_trade_rationale' — recent decision patterns.\n\nStep 3: Run Recurring Pattern Detection — search for learning_entry tags, group by area, check linked chains (3+ = confirmed pattern), investigate drift warnings.\n\nStep 4: Call solana_strategy_state to read current feature weights.\n\nStep 5: Call solana_trades to get recent closed trades. Apply ADL checks (direction consistency, weight velocity, reversion check).\n\nStep 6: Compute proposed weight changes. Score each with VFM (Frequency + Failure Reduction + Self-Cost). Only apply changes scoring >= 3/5.\n\nStep 7: Verify guardrails: maxDeltaOk, sumWeightsOk, minTradesOk, floorCapOk. If all pass, call solana_strategy_update with incremented version.\n\nStep 8: Run Named Pattern Recognition — search for winning trade clusters, catalog new patterns, evolve existing ones.\n\nStep 9: Evaluate discovery filter performance. Log all results via solana_memory_write with tags: strategy_evolution, vfm_scorecard, pattern_detection, named_pattern.\n\nFORMATTING RULES:\n- Every token reference MUST use SYMBOL (full_CA) format.\n- Do not execute trades. Do not ask questions.",
         enabled: true
       },
 

package/config/gateway-v1.json5

@@ -1,6 +1,6 @@
 // OpenClaw Gateway Configuration — TraderClaw V1 (Single Agent)
-// Single "main" agent with 5-minute heartbeat + 10 consolidated cron jobs.
-// Consolidated from 13 → 10: portfolio-health (dead-money+whale+risk-audit),
+// Single "main" agent with 5-minute heartbeat + 11 preset TraderClaw cron jobs.
+// alpha-stream-resubscribe (every 20m WS keepalive); portfolio-health (dead-money+whale+risk-audit);
 // trust-refresh (source-reputation+deployer-trust). Shortened prompts.
 {
   agents: {
@@ -22,14 +22,14 @@
       keepLines: 2000
     },
     jobs: [
-      // ── Alpha Scanning ──────────────────────────────────────────────
+      // ── Alpha stream keepalive ────────────────────────────────────────
       {
-        id: "alpha-scan",
-        schedule: "0 */3 * * *",
+        id: "alpha-stream-resubscribe",
+        schedule: "*/20 * * * *",
         agentId: "main",
         sessionTarget: "isolated",
-        delivery: { mode: "announce", channel: "last", bestEffort: true },
-        message: "CRON_JOB: alpha_scan\n\nScan new launches, filter, score, log alpha. Tools: solana_scan_launches → filter (vol>30K, mcap>10K, liq>5K) → solana_token_snapshot for survivors → quality filter (top10 <50%, deployer <3 abandoned, has social) → score 0-100 → solana_alpha_log for 65+. Summarize results.",
+        delivery: { mode: "none" },
+        message: "CRON_JOB: alpha_stream_resubscribe\n\nKeep alpha WebSocket subscription healthy. solana_runtime_status — if subscribed is false: solana_alpha_subscribe. If subscribed but ingestionStale: solana_alpha_subscribe force:true (zombie heal). Else: reply ALPHA_KEEPALIVE_OK.",
         enabled: true
       },
 

package/dist/index.js

@@ -2831,12 +2831,12 @@ ${notes}
     });
     api.registerTool({
       name: "solana_alpha_submit",
-      description: "Submit a candidate token to the alpha buffer for evaluation in the next heartbeat cycle. Used by cron alpha_scan to queue discovered tokens with thesis data.",
+      description: "Submit a candidate token to the alpha buffer for evaluation in the next heartbeat cycle. Use from isolated sessions or manual workflows to queue a token with thesis data (default install: discovery runs in heartbeat STEP 1, not a separate cron).",
       parameters: Type.Object({
         tokenAddress: Type.String({ description: "Solana token mint address" }),
         symbol: Type.Optional(Type.String({ description: "Token symbol" })),
         thesis: Type.Optional(Type.String({ description: "Thesis summary for why this token qualifies (volume, holders, risk flags, narrative)" })),
-        source: Type.Optional(Type.String({ description: "Signal source (e.g., cron_alpha_scan, manual)" })),
+        source: Type.Optional(Type.String({ description: "Signal source label (e.g., alpha_submit, manual, isolated_session)" })),
         confidence: Type.Optional(Type.Number({ description: "Confidence score 0-100" }))
       }),
       execute: wrapExecute(

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "solana-traderclaw",
-  "version": "1.0.133",
+  "version": "1.0.135",
   "description": "TraderClaw V1-Upgraded — Solana trading for OpenClaw with intelligence lab, tool envelopes, prompt scrubbing, read-only X social intel, and split skill docs",
   "type": "module",
   "main": "./dist/index.js",

package/skills/solana-trader/HEARTBEAT.md

@@ -20,7 +20,7 @@ If MEMORY.md contains a **User Preferences** section, those values override defa
 
 | Preference key | What it overrides |
 |---|---|
-| `volumeMinUsd` | Min 24h volume filter in STEP 1 SCAN and alpha_scan cron (default: 50000) |
+| `volumeMinUsd` | Min 24h volume filter in STEP 1 SCAN (default: 50000) |
 | `marketCapMinUsd` | Min market cap filter (default: 10000) |
 | `maxPositionSizeSol` | Max position size in SOL (overrides entitlement cap if lower) |
 | `scanMode` | `"conservative"` / `"standard"` / `"aggressive"` — adjusts confidence thresholds |
@@ -316,7 +316,7 @@ Do NOT skip these. They feed the bootstrap digest that loads into your next sess
 HEARTBEAT REPORT — [timestamp UTC]
 
 Capital: X.XXX SOL | Positions: N open | Holdings verified: [yes/no via solana_wallet_token_balance]
-Scanned: N launches, N hot pairs | Alpha: N signals (top score: XX)
+Alpha: N signals (top score: XX)
 
 DEEP ANALYSIS:
 Bitquery: [N templates run on N tokens | "none — no FRESH tokens"]
@@ -365,5 +365,5 @@ NEXT CYCLE: [1 sentence — what you're watching for]
 | API endpoint reference | refs/api-reference.md |
 | Wallet proof vs signup | SKILL.md § Wallet proof vs signup |
 | Strategy evolution details | refs/strategy-evolution.md |
-| Cron job definitions | refs/cron-jobs.md (10 consolidated jobs, ~39 sessions/day) |
+| Cron job definitions | refs/cron-jobs.md (11 preset jobs; see Schedule Summary for runs/day) |
 | Position management details | refs/position-management.md |

package/skills/solana-trader/refs/alpha-signals.md

@@ -22,6 +22,8 @@ Subsequent heartbeats:
   → Returns only new signals since last check
 ```
 
+Installer / Gateway preset **`alpha_stream_resubscribe`** (every 20 minutes) calls `solana_runtime_status` and re-invokes `solana_alpha_subscribe` when the subscription dropped or ingestion is stale. See refs/cron-jobs.md.
+
 If buffered signals stay empty for multiple cycles:
 ```
 solana_gateway_forward_probe({ agentId: "main", source: "heartbeat_recovery" })
@@ -158,7 +160,7 @@ Use `solana_alpha_history` for:
 
 ## Alpha Submission (Cron → Heartbeat Pipeline)
 
-Use `solana_alpha_submit` to queue a candidate token into the alpha buffer after cron-based scanning (alpha_scan job). The heartbeat cycle then evaluates buffered candidates. Include: tokenAddress, symbol, thesis (volume, holders, risk, narrative), source, confidence.
+Use `solana_alpha_submit` to queue a candidate token into the alpha buffer for evaluation on the next heartbeat (e.g. after deep research outside the fast loop). Launch and pair discovery runs in **STEP 1: SCAN** on each heartbeat — there is no separate alpha_scan cron in the default presets. Include: tokenAddress, symbol, thesis (volume, holders, risk, narrative), source, confidence.
 
 ## Firehose Configuration
 

package/skills/solana-trader/refs/cron-jobs.md

@@ -4,7 +4,7 @@ Cron jobs run in **isolated sessions** separate from the trading loop. Each job
 
 If a cron job fails, the fast loop continues unaffected — failures are retried on the next scheduled run.
 
-When you receive a `CRON_JOB:` message, execute ONLY the specified job. Do not run the trading loop.
+When you receive a `CRON_JOB:` message, execute ONLY the specified job. Do not run the trading loop. **Launch and pair scanning** is heartbeat **STEP 1: SCAN**, not a cron — there is no default `alpha_scan` job.
 
 ## Memory Context Load (mandatory for every cron job)
 
@@ -19,21 +19,21 @@ At start of every cron job, check whether sufficient new data exists since last
 
 ---
 
-## Job: `alpha_scan`
+## Job: `alpha_stream_resubscribe`
 
-**Schedule:** Every 3 hours (`0 */3 * * *`) — 8 runs/day
+**Schedule:** Every 20 minutes (`*/20 * * * *`) — 72 runs/day
 
-**Purpose:** Scan new token launches, filter candidates, score quality, log alpha signals.
+**Purpose:** Recover from dropped or zombie alpha WebSocket subscriptions (signals stop buffering). Complements plugin watchdog; explicit cron ensures reconnect even across process edge cases.
 
-**Tools:** `solana_scan_launches`, `solana_token_snapshot`, `solana_token_holders`, `solana_token_risk`, `solana_alpha_log`, `solana_memory_write`
+**Tools:** `solana_runtime_status`, `solana_alpha_subscribe`
 
-**Workflow:** Scan launches → filter (vol>30K, mcap>10K, liq>5K) → snapshot survivors → quality filter (top10 <50%, deployer <3 abandoned, has social) → score 0-100 → log 65+ via alpha_log.
+**Workflow:** Call `solana_runtime_status`. If `alphaStream.subscribed` is false, call `solana_alpha_subscribe`. If subscribed but `alphaStream.ingestionStale`, call `solana_alpha_subscribe` with `force: true`. Otherwise reply `ALPHA_KEEPALIVE_OK`.
 
 **Configuration:**
-- Model: Sonnet (judgment — scoring candidates, filtering quality signals)
+- Model: Haiku (branch on flags, single tool calls)
 - Thinking: off
 - lightContext: on
-- Delivery: announce/last/bestEffort
+- Delivery: none
 
 ---
 
@@ -222,8 +222,8 @@ At start of every cron job, check whether sufficient new data exists since last
 ## Schedule Summary
 
 | # | Job ID | Cron Expression | Runs/Day | Model | Thinking | lightContext | Delivery |
-|---|--------|----------------|----------|-------|----------|-------------|----------|
-| 1 | `alpha-scan` | `0 */3 * * *` | 8 | Sonnet | off | on | announce/last |
+|---|--------|-----------------|----------|-------|----------|-------------|----------|
+| 1 | `alpha-stream-resubscribe` | `*/20 * * * *` | 72 | Haiku | off | on | none |
 | 2 | `portfolio-health` | `0 */4 * * *` | 6 | Sonnet | off | on | announce/last |
 | 3 | `trust-refresh` | `0 */8 * * *` | 3 | Haiku | off | on | none |
 | 4 | `meta-rotation` | `30 */8 * * *` | 3 | Sonnet | off | on | announce/last |
@@ -233,4 +233,4 @@ At start of every cron job, check whether sufficient new data exists since last
 | 8 | `intelligence-lab-eval` | `0 16 * * *` | 1 | Sonnet | **on** | **off** | none |
 | 9 | `memory-trim` | `0 3 * * *` | 1 | Haiku | off | on | none |
 | 10 | `balance-watchdog` | `0 */2 * * *` | 12 | Haiku | off | on | announce/telegram |
-| | **Total** | | **39** | | | | |
+| | **Total** | | **103** | | | | |

package/skills/solana-trader/workspace/TOOLS.md

@@ -101,7 +101,7 @@ Every tool has a mandatory trigger — when the trigger condition is met, you MU
 ### Alpha Signals (5)
 | Tool | Purpose | When to Call |
 |---|---|---|
-| `solana_alpha_subscribe` | Subscribe to alpha stream (`force: true` forces reconnect) | Startup sequence; when stream disconnects or ingestion stalls |
+| `solana_alpha_subscribe` | Subscribe to alpha stream (`force: true` forces reconnect) | Startup sequence; `alpha_stream_resubscribe` cron (every 20m); when stream disconnects or ingestion stalls |
 | `solana_alpha_signals` | Poll buffered signals | Step 1.5 every heartbeat |
 | `solana_alpha_history` | Historical signal data | Step 1.5 to check prior calls on a token; Step 7 to check source accuracy after exit |
 | `solana_alpha_sources` | Per-source performance stats | Step 1.5 to check source win rates; `source_reputation_recalc` cron |
@@ -191,7 +191,7 @@ Every tool has a mandatory trigger — when the trigger condition is met, you MU
 ### Runtime (3)
 | Tool | Purpose | When to Call |
 |---|---|---|
-| `solana_runtime_status` | Plugin runtime health | Diagnostics; when tools behave unexpectedly |
+| `solana_runtime_status` | Plugin runtime health (startup gate cache, alpha stream subscription + ingestionStale) | `alpha_stream_resubscribe` cron; diagnostics when tools behave unexpectedly |
 | `solana_agent_sessions` | List agent sessions | Diagnostics; when checking session state |
 | `solana_classify_deployer_risk` | Deployer risk (alias) | Same as `solana_compute_deployer_risk` — use either |