Each /boucle subagent now ends its output with a Difficultés section;
at the end of a session the orchestrator calls the releveur once, which
locates the instruction or file at fault and records it in
docs/ameliorations-processus.md, without fixing anything.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The agent now serves one precise case, reads the project's displacement
knowledge and the whole case base, starts from the detectors' events
without judging them, and no longer has conclusions imposed on it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Plan points 18 to 26, 30, 39, 40 and the point 37 housekeeping:
- greyed zones per target, vertical scale without the zones to ignore, Fit all,
mm by default with a px switch, Support view, Expand (one row per category);
- cleaned curves: cut in the zones to ignore, raw data kept there in thin grey;
- target02 to target05 count as targets; smoothing questions answered;
- categories.py sorts in slices with 3 days of history by default;
- the pumping script points to ~/monitoring-dashboard and skips diff images;
- CLAUDE.md back under 200 lines; a case page is named after its case.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Visibility: v0 kept after a measured trial of a simpler version (archived); two levels, Reduced and No visibility
- Thermal: a 24-hour sinusoid confirms the daily wave, split into thermal expansion and contraction events
- Jumps tested from the first measurement; turbulence never kept over a shift to check (CLARIDEN)
- Data gaps say whether camera images exist; Displacements trends become slow drift events
- Threshold registry moved to vues-deplacements/contexte/seuils-du-journal.md; plan points 11 to 17, 38 and 41 validated
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Event map regrouped into six families, 33 events including hypothetical ones, with the journal categories (section 4)
- categories.py turns detector episodes into categorised events: absorption of brief jumps, merging, catégorique phrases
- Journal tab: one "What happened" row coloured by family, cards, and an Images button playing the event's timelapse
- Plan: user-validated points marked, points 39 and 40 added; anomaly definition clarified in D-056
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The jumps and trends now sit in a Displacements tab after Diagnosis
(#tab=displacements), no longer in a separate sidebar page; a click on
a row opens the Diagnosis tab on that target.
Documented: what the client looks at (jumps and trends, not errors,
handled upstream) in docs/displacement/domaine/ce-que-regarde-le-client.md;
the expert's reading of NORDEND in the case journey; the experimental
way of working in vues-deplacements/CLAUDE.md; D-055, the spec, the
note, the README and the project map.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Jumps: a sudden change within 2 h (8 times the target's usual 2-hour
change) that lasts 24 h (at least 1.4 times its daily swing); every
jump of the range, returns included, so PR17's return on 12/10 now
shows. Trends: Theil-Sen slope and Kendall tau of the daily levels,
jumps set aside, kept when the drift over the period reaches the daily
swing and goes steadily one way. Provisional thresholds; method and
results in vues-deplacements/contexte/detection-d-anomalies.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Mechanical anomaly detection on the hourly displacements in mm, data
unchanged: for each target and axis, the largest gap between the median
of the next 24 h and the previous 24 h, kept above 3 times the target's
daily swing (provisional). Jumps within 6 h form one event, so a common
jump is not several anomalies. Grey curves for all targets, colour for
those that jumped; a click opens Explore, Diagnosis.
Note on the idea, first trial and leads in
vues-deplacements/contexte/detection-d-anomalies.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Copy the station's displacement export and images into the case folder,
with coded sensor names; the station takes the case number instead of
S01, and graphique.py / vignettes.py drop --alias (D-050).
Rewrite C-001 as the expert's diagnosis of this event only: summary,
symptom, verdict, signature with commands, diagnosis, evidence. Drop
exclusions, expected output, unknowns and test history; matching no
longer uses exclusions (D-051). The corr_x jumps at the reference
switches are judged negligible.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add the anonymiser-cas skill: a growing registry gives stable codes
(S01, R1, C1), a stdlib script rewrites a case and renames its files,
and graphique.py / vignettes.py take --alias to plot with codes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Reports become a folder with rapport.md and its figures: a verdict, one
short section per anomaly (series plot with shaded window, full-resolution
image crops around the sensors), constant behaviour in one line each.
graphique.py gains --zone and --sortie; new vignettes.py crops images at
the sensors' raw x/y. Old reports removed; the blind rescan of PS des
Coudres finds the single anomaly flagged by the specialist (REF_G/REF_D).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Images extracted by hand, one per full hour from 2026-10-13 12:00 to
2026-10-14 11:00 UTC, renamed <UTC time>_<original number>.jpg. UTC is
confirmed by the user and matches sunset and sunrise. They cover the
second night episode on REF_G and REF_D.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Exports go to data/<STATION>/displacement/, out of reach of the system
metrics loader. The 15-min "advanced" export (UTC) is the analysis source;
the hourly export is kept for the corr_x/corr_y conversion factors.
Document both formats and record the open questions in the framing doc.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The user's preferred name. The module becomes src/energy/power_source.py
and the sheet docs/energy/analyses/daily_power_source.md; the
calculation and the results are unchanged (D-043).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A station is on mains when its BusVoltage mean over the 3 days before
the day is >= 14.0 V (solar never exceeds 13.50 V on VERMONT; mains
stays >= 14.08 V). On mains, the mains loss decides the day color
instead of the state x trend matrix: red as soon as an hour falls under
13.4 V (top of the OCV plateau; mains-on hours stay >= 13.65 V, mains-off
hours are 13.1-13.3 V). On MISSISSIPPI the day turns red 1-2 h after the
loss instead of 29-40 h with the matrix; WEISSHORN and NORDEND (mains)
raise no false alert, VERMONT (solar) is unchanged (D-043).
The matrix drawio gets the mains case, the site figure a third panel,
and the support notification and the client Energy block header show
"On mains" / "Not on mains" / "Mains unknown".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Temperature joins the mockups' list of domains (D-041): a Red
notification on the support Home, and a one-line block on the client
view ("Hottest (24 h) 43 °C · limit 97.5 °C") for the stations that
measure the GPU. src/temperature/final.py writes the site figure; the
comparison page reuses its panel. Short two-layer guide in
docs/temperature/guide/.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
DAILY_GPU_TEMPERATURE_LIMIT: red when an hourly GPU mean reached 97.5 °C
(T.gpu, Jetson Nano Thermal Design Guide) in the last 24 hours, green
otherwise, gray without measurement. Working sets WEISSHORN and NORDEND;
all measured days are green. D-040 replaces D-036; the exploration note
moves to docs/temperature/domaine/point-de-depart.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The mockups now walk a list of domains (D-039): the support Home loads
every station in data/ and notifies each station x domain that is Orange
or Red; the client view becomes maquettes/user/summary.py, one System
Summary page per station with one block per measured domain. Storage
shows a free-space scale and the estimated full date; Cellular shows the
day's sends in words and a four-class signal scale.
HEADLINE moves into each day_color.py, with a Green sentence for the
client view. src/storage/final.py and src/cellular/final.py write the
site figures in English; the comparison pages reuse their panels. Each
domain gets a short two-layer guide in docs/<domaine>/guide/.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Docker Engine in WSL (mirrored networking), Python 3.14 and plotly,
--network=host to reach the camera over Ethernet (D-038).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
DAILY_STORAGE_STATE classifies the end-of-day free space of /data and
DAYS_UNTIL_STORAGE_FULL divides it by the median daily fill over 7
days. The day color reads a state x horizon matrix drawn in
docs/storage/matrice-etat-horizon.drawio. Thresholds are arbitrary for
now (D-034, D-035); expert questions stay open. On WEISSHORN the disk
fills at 28.2 MiB/day and would be full around 06/02/2027.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
DAILY_LINK_STATE falls back on last_sync for exports without link,
treats the start of history as unknown, and records D-032. Document
the hourly counting flaws, the observed 15-minute cadence, and the
open questions for the specialist.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Explore WEISSHORN for the cellular domain, count lost sends and
messages, grade the daily RSRP and RSRQ medians on the published
Quectel grid, and record D-031.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
New data will be added per station without mixing with the working set.
All scripts now default to data/VERMONT/; docs and D-029 follow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
D-028: CLAUDE.md names every folder and document with its size in
lines, says how to read without filling the context (outline of a long
file then one section, index of the decision log, CSV by script,
folders not to read), and gives the current state instead of a dated
history. It and D-026 now name final.py for the delivered script.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Record decision D-027: a server-side LLM agent layer next to the
unchanged deterministic rules, aimed at unknown anomalies and rule
edge cases. The agent annotates and never rewrites the day color.
Home-made harness (Messages API + Tool Runner), staged start with a
manual Energy pilot on VERMONT history. D-003 is partly replaced.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Energy code moves to src/energy/ and its docs to docs/energy/; shared code
(CSV loading, venv switch, browser, zoom script) goes to src/common/, and the
AIMsight originals to docs/aimsight/. Analyses are no longer numbered: two
Energy scripts remain, daily.py (delivered state and trend) and
all_analyses.py (comparison of the four). Calculations and pages unchanged (D-026).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
D-025: the support side gets the site's home page (one tile per
station: name, last voltage, curve) with a Status section on top. It
holds one notification per station whose last day of the window is
Orange or Red (src/day_color.py), the most serious first; Green
notifies nothing. Each notification, like the station tile, links to
the station's client view (out/maquette_user_energy.html).
maquettes/support/home.py writes out/maquette_support_home.html with
the standard library only (inline SVG, no plotly). Only VERMONT has
data; the other tiles keep the site's names and stay empty.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CLAUDE.md records D-022/D-023 in the project state and the next mockup
steps (client Energy step by step, then the support side). D-022 lists
the open points of the Energy mockup: the guide is still in French,
the battery scale has no IR compensation, the energy save button is
for the site team to build, and a day in progress is not yet flagged
as provisional on the client side.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
D-024: to explain a day's color, show only that day and the day before,
the only days analyses 03 and 04 read for it. day_color.focus() selects
the data (standard library); src/focus_jour.py prints a traceable fact
summary and writes a 48 h BusVoltage figure (out/<STATION>_focus_<date>
.html and .json). Analysis 03 gains the night_end_h fact; the other
facts are unchanged. The client Energy mockup does not use this figure:
the user prefers numeric facts there.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>