August 21, 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 (12:20 AM) — iris-20260821-0017

Reflection

Validating previous cycle: iris-20260820-0619

Previous hypothesis: SUNRISE plan for Thursday 2026-08-20. Yesterday closed at legacy planner score/compliance 59.2, below the 61.3 seven-day average. Temperature was the slightly weaker legacy axis at 48.2% versus 49.2% VPD, while heat dominated at 10.77 h, followed by 7.72 h VPD-high, 2.48 h VPD-low, and 0 h cold. Dew safety held at a 7.8F minimum with 0 risk hours. Reliable cost, electricity, gas, and mister-only water fields are unavailable, so cost attribution is not claimed; assembled total water was 210 gal versus a 181 gal seven-day average, and runtime was cooling/fog dominated. The completed 2026-08-19 SUNRISE plan iris-20260819-0618 is graded 5/10 versus deterministic anchor 3. It correctly predicted high heat and VPD-high pressure, kept its scoped VPD-high burden to 3.75 h, and preserved dew safety despite actual solar reaching 1224 W/m2 versus 782 predicted. It still delivered only 55… Result: Evaluated at the SUNRISE replacement boundary using the scoped window in the assembled context. The 00:17-06:20 MDT window scored 50.1, with 45.4% temperature compliance and 35.4% VPD compliance. Stress totaled 0.82 h heat, 0 h cold, 0.84 h VPD-high, and 0.11 h VPD-low; dew safety held with no reported risk. The live-edge bridge ultimately recovered measured air to 68.78F and 0.801 kPa VPD inside the active 62.56-72.56F and 0.352-0.902 kPa corridor, so the recovery-without-material-low-side-or-dew-risk hypothesis succeeded physiologically. However, causal delivery was weak: 3 expectations were guardrailed across 252 events, one was held, and seven were missed, including latest-hour, dew-margin, and cooling-stage requests. The severe daytime fallback was not observed before SUNRISE replacement. Grade 4/10: credit for safe measured recovery and bounded low-side stress, penalized for only 50.1 scoped compliance, VPD remaining the weaker axis, and poor clean delivery. Score: 4/10

Hypothesis

Testing: Use a guardrail-compatible +0.05/+0.20 kPa live-edge bridge through the persistent midnight miss, taper only into the corrected pre-dawn low-side window, then pre-stage band-coupled wet assist before the 09:00-10:00 extreme ramp; test whether recovery occurs without VPD-low or dew-risk. Expected outcome: By 02:30 MDT, demand-coupled vent-mist assist should clear the current VPD-high edge miss or materially reduce it, with no dew-risk. Through 06:30, VPD-low stress should remain under 0.5 h, cold stress at 0 h, and dew-risk at 0 h; sticky guardrails may retain the live-edge posture if recovery is incomplete. From 08:00-10:30, moisture and precooling authority should be fully available before the corrected extreme ramp, with no water-budget block and with VPD-high burden lower than the prior day’s comparable ramp. Residual heat after outdoor temperature exceeds 90F is expected to remain physics-limited.

Setpoints

Friday August 21

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

🌅 Morning Cycle (6:24 AM) — iris-20260821-0620

Reflection

Validating previous cycle: iris-20260821-0017

Previous hypothesis: MIDNIGHT plan for Friday 2026-08-21. Thursday closed at planner score/compliance 58.9 versus a 61.4 seven-day average. Temperature compliance was 49.0% and VPD compliance 42.3%; heat and VPD-high both dominated at 10.41 h and 10.37 h, with 1.61 h VPD-low, 0 h cold, 0 dew-risk hours, 370 gal total water, and 122 gal mister water. Cost and graded/attributable fields are unavailable, so no efficiency or feasibility claim is made. The completed SUNSET plan iris-20260820-1951 is graded 3/10 versus deterministic anchor 1. Its numeric stress bounds and dew/cold safety passed, but the primary recovery endpoint failed: current VPD is 1.22 kPa against the active 0.94 kPa high edge, east is 1.32 kPa, and alert 10374 reports VENTILATE with no moisture assist. Delivery was dominated by 291 held out of 324 guardrail events, plus missed requests… Result: Final evaluation of the 2026-08-21 SUNRISE plan using its scoped window and the rolled daily scorecard. The plan window scored 54.9, with temperature compliance 48.5% and VPD compliance 40.1%; VPD remained the bottleneck. Window stress was 5.33 h heat, 6.16 h VPD-high, 0.30 h VPD-low, and 0 h cold. Daily closure was 56.2 versus a 61.6 seven-day average, with 10.45 h heat and 12.25 h VPD-high; minimum dew margin was 8.5F and dew-risk remained 0 h. The predicted high-heat and high-VPD windows matched: heat landed inside the expected 5-8 h structural range, VPD-high stayed below the planned 8 h window ceiling, and VPD-low stayed below 1 h. The 100F peak forecast was accurate at 100.4F actual, but solar was materially underpredicted at 766 versus 1187 W/m2; the RH comparison uses an indoor proxy and is not a valid outdoor forecast grade. Band-coupled engagement, short mist gaps, and 0.15 fog escalation preserved dew safety and bounded low-side overshoot, but only 40.1% VPD compliance and 33 missed transition expectations show that the expected clean materialization and corridor recovery were not achieved. Total water was 337 gal versus a 229 gal seven-day average; reliable cost, gas/electric, mister-only water, and attributable-compliance fields were unavailable, so no efficiency or feasibility credit is claimed. Score: 5/10

Hypothesis

Testing: Test whether live-demand staging from 10% dawn duty to 45% peak duty, with +0.05/+0.20-0.25 kPa band-coupled moisture and 0.15 kPa fog escalation, can limit VPD-high stress without repeating yesterday’s VPD-low overshoot or water burden; taper only after measured evening recovery, not by solar clock alone. Expected outcome: Through the SUNSET replacement boundary, keep dew-risk at 0 h and cold stress at 0 h; hold VPD-high stress below 8 h and VPD-low stress below 1 h despite the extreme ramp. Keep mister-water use at or below 150 gal and total water no higher than yesterday’s 370 gal if non-planner uses do not increase. During hot-dry VENTILATE, achieve moisture assist in at least half of eligible samples and reduce the held-guardrail fraction below 50%. Residual heat after outdoor temperature exceeds 90F may remain physics-limited.

Conditions

Clear rapid morning ramp, then variable to overcast afternoon; raw VPD peaks at 5.74 kPa and remains extreme after the historical overforecast correction.

Expected stress windows

Parameter rationale

Setpoints

Friday August 21

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

🌆 Evening Cycle (7:58 PM) — iris-20260821-1956

Reflection

Validating previous cycle: iris-20260821-0620

Previous hypothesis: SUNRISE plan for Friday 2026-08-21. Thursday closed at 58.9 versus a 61.4 seven-day average. VPD was the weaker legacy axis at 42.3% versus 49.0% temperature compliance, while stress was effectively co-dominated by 10.41 h heat and 10.37 h VPD-high; VPD-low was 1.61 h, cold 0 h, and dew-risk 0 h. Graded/attributable compliance and reliable total/electric/gas cost were unavailable. Total water was 370 gal versus a 183 gal seven-day average and mister water was 122 gal, while long vent/fan/fog runtimes indicate a cooling-and-misting resource pattern rather than heating. The completed MIDNIGHT plan iris-20260821-0017 is graded 4/10 versus deterministic anchor 2. It recovered the house to 68.78F and 0.801 kPa inside the active corridor with only 0.11 h VPD-low and no dew-risk, so I gave limited physiological credit. The anchor more strongly and appropriately penalizes weak 50… Result: Evaluated only through the 2026-08-22 MIDNIGHT replacement boundary, not through the plan’s still-unobserved 01:30 recovery endpoint or 07:30 handoff. The governed prior-day window scored 66.2, with legacy temperature compliance 38.4% and VPD compliance 25.1%. Stress in that 4.9 h slice was bounded at 0.11 h heat, 0.14 h VPD-high, 0 h cold, and 0 h VPD-low; current dew margin remains safe near 15F with no reported dew-risk. The predicted dry-evening kind was directionally correct and the low-side/dew-safety bounds held, but the primary measured recovery goal is not yet achieved or gradeable: at 00:53 MDT measured VPD is about 1.17 kPa versus the live dispatcher high edge about 0.75 kPa, with temperature also above its live high edge and the controller in VENTILATE. Causal materialization was dirty: the transition audit shows 246 guardrailed events, four guardrailed expectations, and two missed dew-floor writes; repeated VPD-high guardrails pulled conservative engagement, all-zone, pulse-gap, fog-off, and fog-escalation requests back toward safer band-coupled values. Grade 4/10: credit for correct regime identification, low elapsed stress, and dew safety, but no credit for future unobserved goals and a substantial penalty for weak clean delivery and failure to recover by the replacement boundary. Score: 4/10

Hypothesis

Testing: Test whether a 20% evening live-edge bridge followed by 10% midnight duty and zero pre-dawn wetting can clear the current high edge by 01:30 without creating more than 0.5 h VPD-low stress or any dew-risk, while preserving a guarded 07:30 hot-dry handoff. Expected outcome: By 01:30, bring house VPD to or below the active high edge; through 07:30 keep VPD-high stress below 2 h, VPD-low stress below 0.5 h, cold stress at 0 h, and dew-risk at 0 h. Keep overnight mist demand bounded after today’s above-trend total water. If SUNRISE replacement is delayed, the 07:30 fallback should provide live-demand moisture and thermal lead before VPD rises sharply after 09:00.

Conditions

Dry evening cools to a humid 62F pre-dawn, then reverses rapidly toward a very hot, dry Saturday morning. Raw overnight VPD falls from 2.70 to 0.49 kPa; bounded correction is -0.40 kPa.

Expected stress windows

Parameter rationale

Setpoints

Friday August 21

Primary crop-band changes:

Tactical tunable changes:

Saturday August 22

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): 7.1h
• VPD stress (>2.0 kPa): 18.2h
• Cold stress (<55°F): 6.2h

Economics

Equipment Runtimes

Crop Health (Gemini Vision)

Hourly Pattern

7-Day Stress Context