July 15, 2026

Generated lab notebook from daily_summary, plan_journal, plan_delivery_log, and setpoint audit data. It is intentionally chronological and may include in-progress cycles before validation.

Planner Execution Ledger

Planner checkpoints can acknowledge that the active plan is still suitable without writing a new public plan ID. Those no-change decisions are part of the audit trail.

🌅 Morning Cycle (12:18 AM) — iris-20260715-0016

Reflection

Validating previous cycle: iris-20260714-0545

Previous hypothesis: SUNRISE plan for Tuesday 2026-07-14. Yesterday’s daily scorecard endpoint is degraded by db_statement_timeout, and mandatory semantic retrieval is degraded because knowledge_search/lessons_search require OPENAI_API_KEY. I evaluated iris-20260713-0544 at 5/10, but plan_evaluate returned deterministic anchor 2 due 324 VPD-high guardrail events and 256 held-by-guardrail events; I accept the anchor as the stronger materialization warning. The forecast repeats an extreme clear hot-dry sequence: today 97F, RH 12-13%, solar 923 W/m2, raw VPD near 5.15 kPa, with corrected VPD still far above the Vanda/food corridor from late morning through evening. Current house is safe and in-band at about 69.3F, VPD 0.76 kPa, RH 69%, dew margin 10.6F, IDLE, all relays off… Result: MIDNIGHT reset governed the late-night/pre-dawn window before the SUNRISE plan replaced it. Window score was 66.4, with temp compliance 52.3%, VPD compliance 37.4%, heat stress 1.86h, VPD-high stress 2.67h, no cold stress, no VPD-low stress, and dew safety preserved (dp_margin_min 9.0F, dp_risk_hours 0.0). It correctly kept the house safe and preserved moisture assist for the coming dry ramp, but VPD remained the bottleneck and materialization remained guardrail-limited; this does not prove the full daylight hypothesis because the plan was superseded at sunrise. Score: 6/10

New finding: The late-night reset was safe but did not materially improve dry-edge VPD before sunrise; on repeated July hot-dry sequences, overnight plans should keep measured-recovery wet assist available while avoiding claims that pre-dawn posture alone can solve the next daylight VPD capacity limit. → Added to Lessons Learned

Hypothesis

Testing: Test whether a directly guardrail-compatible midnight-to-daylight wet-assist posture can reduce guardrail reshaping and preserve dew safety through another severe hot-dry ramp. Expected outcome: By sunrise, dew margin remains above 8F with no cold stress; through Wednesday, VPD-high remains the bottleneck but wet/fog assist stays materialized with fewer guardrail-held rows than 2026-07-14, no dp-risk hours, and no water-budget clipping of live VPD recovery.

Conditions

another clear severe hot-dry July ramp after a guardrail-heavy VPD-high day; current average VPD is served but east zone remains dry-edge and dew margin is safe

Expected stress windows

Parameter rationale

Setpoints

Wednesday July 15

Primary crop-band changes:

Tactical tunable changes:

Thursday July 16

Primary crop-band changes:

Tactical tunable changes:

Friday July 17

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

🌅 Morning Cycle (5:47 AM) — iris-20260715-0545

Reflection

Validating previous cycle: iris-20260714-2030

Previous hypothesis: SUNSET 2026-07-14 overnight plan. Prior MIDNIGHT reset iris-20260714-0018 was evaluated 4/10 and matched deterministic anchor 4; 291 VPD-high guardrail events mean the plan was safe but only partially materialized, so this plan requests guardrail-compatible band-relative wet assist instead of relying on correction. Today scored 71.4 versus a 7-day average of 61.6, but VPD remained the limiting axis: temp compliance 51.1%, VPD compliance 38.3%, heat stress 8.43h, VPD-high stress 11.14h, no cold stress, and no dew risk (minimum dew margin 8.8F). Live house is safe and mostly served at about 79.7F, VPD 1.05 kPa, RH 70%, dew margin 10.7F, IDLE, but east remains dry-edge at 1.34 kPa and zone spread is large. Overnight challenge is not heating; it is finishing dry-edge recovery without wasting water or adding relay churn, then handing off to another severe hot-dry Wednesday ramp… Result: SUNRISE plan governed 61.5% of 2026-07-15 and correctly predicted the stress class: severe heat and VPD-high occurred, with actual outdoor temp near 95.4F and solar peaking 1053 W/m2 versus the 901 W/m2 plan prior. Safety goals were met: no cold stress, no VPD-low stress, dp_margin_min stayed 9.0F, and dp_risk_hours were 0. However the measurable compliance outcome was only partial: window score 55.2, temp compliance 41.2%, VPD compliance 29.5%, heat stress 5.28h, and VPD-high stress 7.23h while daily water remained high at 350 gal total / 143 gal mister water. Rationale grading: band-coupled mister/fog settings kept wet assist available and safe but did not produce VPD recovery into the corridor; fog escalation 0.15 was directionally correct, but capacity and guardrail/materialization limits dominated; extending evening wet recovery was safe but did not erase the dry-edge deficit. Overall: safe and correctly diagnosed, but optimistic about achievable VPD compliance under >1000 W/m2 solar and mid-90s heat. Score: 5/10

New finding: On repeated July hot-dry days where solar exceeds about 1000 W/m2 and outdoor highs reach the mid-90s, even guardrail-compatible band-coupled wet/fog assist can preserve dew safety but still leave VPD compliance near 30%. Frame plans around severity reduction and measured recovery, not full corridor recovery, and do not unwind wet assist until observed VPD is back inside band. → Added to Lessons Learned

Hypothesis

Testing: Test whether a fully guardrail-compatible sunrise plan, with aggressive daylight wet assist and 22:00 measured-recovery window, reduces held-by-guardrail events and improves VPD compliance versus 2026-07-14 without adding dew-risk hours. Expected outcome: Dew-risk hours remain 0, VPD-high stress improves from 13.96h toward under 12h despite 95F/900 W/m2 conditions, and guardrail held events fall relative to the 434-event sunset-plan anchor.

Conditions

clear severe hot-dry July day after another guardrail-heavy VPD-high day; forecast VPD has known positive bias but corrected VPD remains above crop corridor from morning through evening

Expected stress windows

Parameter rationale

Setpoints

Wednesday July 15

Primary crop-band changes:

Tactical tunable changes:

Thursday July 16

Primary crop-band changes:

Tactical tunable changes:

Friday July 17

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

🌆 Evening Cycle (8:32 PM) — iris-20260715-2030

Reflection

Validating previous cycle: iris-20260715-0545

Previous hypothesis: SUNRISE plan for Wednesday 2026-07-15. Yesterday scored 69.9, above the 7-day average 61.6, but the apparent score masks a persistent dual-axis hot-dry limitation: temp compliance was 48.2%, VPD compliance only 36.6%, total stress 25.02h, with VPD-high dominant at 13.96h and heat stress 11.05h. Water use was high at 366 gal total and 171 gal mister water, well above the 167 gal 7-day average, while reliable cost/electric/gas terms were unavailable. I evaluated the prior SUNSET plan iris-20260714-2030 at 4/10; deterministic anchor was 2 because 473 VPD-high guardrail events and 434 held-by-guardrail events show materialization remained strongly reshaped. Current house is safe and in band at about 68.9F, VPD 0.84 kPa, RH 65%, dew margin 12F, IDLE, all relays off… Result: SUNSET overnight plan governed the late evening / pre-dawn handoff window into 2026-07-16. It was safe and directionally correct but only partially effective: scoped window score 60.3, temp compliance 42.6%, VPD compliance 30.7%, heat stress 0.48h, VPD-high stress 0.58h, no cold stress, no VPD-low stress, dp_margin_min stayed safe at 9.0F with 0.0 dp_risk_hours. The hypothesis that overnight challenge was dry-edge recovery and relay/water churn was right; the plan preserved wet assist while dew margin was safe and avoided disease risk. However VPD remained the bottleneck at sunrise and the explicit direct_wet_stress_latest_hour=22 posture did not stick in the current active setpoints, so the plan did not materially improve the next-day dry-edge baseline. Daily water was high at 350 gal total / 143 gal mister water versus 234 gal 7-day average total water; reliable cost/electric/gas terms were unavailable. Score: 5/10

New finding: On hot-dry July sequences, sunset/overnight recovery plans can safely preserve wet assist and dew margin, but they should be graded as partial success if VPD remains the sunrise bottleneck; do not claim overnight posture solved the next daylight VPD capacity limit unless measured VPD is back inside band by dawn. → Added to Lessons Learned

Hypothesis

Testing: Test whether a measured-recovery sunset taper can reduce overnight water/relay churn after 22:30 without adding VPD-low or dew-risk hours, while still handing off aggressive wet assist for Thursday morning. Expected outcome: By sunrise, dew-risk hours remain 0, VPD-low stress remains 0, VPD-high stress after 22:30 stays under 3h, and the controller reports fewer guardrail-held wet-assist rows than the prior night.

Conditions

warm dry evening after a hot-dry VPD-high day; current house VPD is above band with safe dew margin around 13F, and Thursday ramps to severe heat/VPD again after sunrise

Expected stress windows

Parameter rationale

Setpoints

Wednesday July 15

Primary crop-band changes:

Tactical tunable changes:

Thursday July 16

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

End-of-Day Summary

Climate

Interior Light Evidence

• Crop DLI: unavailable
• Reason: source_reading_missing
• Provenance: legacy_invalid_exterior_proxy_plus_fixture_estimate
• Validity: dli-validity-v1; 2024-01-01T00:00:00+00:00 → open
• Qualified-light minutes and photoperiod remain independent of unavailable DLI.

Stress Hours

• Heat stress (>85°F): 4.2h
• VPD stress (>2.0 kPa): 9.6h
• Cold stress (<55°F): 8.8h

Economics

Equipment Runtimes

Water

• Total: 350 gal
• Mister: 143 gal

Hourly Pattern

7-Day Stress Context