Council #2: the empty GA 2/5/2 spawned missing_status + policy_taxonomy_outlier +
a lighting classification on top of its real defect. An empty name means the
classifier can't be trusted, so detect_missing_status and check_policy now skip
blank-name GAs; the single root cause is empty_name (check_naming). Demo 2/5/2:
3 findings -> 1. tests/test_root_cause.py; full suite green (10 tests).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Three independent reviewers (two external councils + a field integrator) asked for
the same thing: the enriched model mixes ETS facts, DPT structure and name
heuristics, and downstream tools treat it almost as fact. explain_ga makes the
reasoning auditable per GA — evidence per decision with a confidence tier
(authoritative ETS Function > structural DPT > heuristic name), the status-pairing
method, and CONFLICTS (name 'AC' vs DPT 'lighting' -> contested), the silent-
misclassification hotspot. Additive, read-only, no core-model change. Full suite green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
'Does the function have *a* status' missed a dimmer with on/off status but no
brightness status (demo planted error #3), silently inflating coverage/Matter.
detect_role_completeness flags a brightness/position command (5.001) whose device
has no matching value status (missing_value_status), with a device-identity match
so it doesn't borrow a sibling's status. Demo recall 4/5 -> 5/5. Raised by the
external expert review. Test + demo ground-truth updated. Full suite green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
External expert review (real run on the demo house) flagged: scenes (18.001)
wrongly marked missing_status + an unsafe DPT-1.011 status repair for them;
'All blinds down' warned while 'All lights off' was correctly INFO; and a Russian
'(статус)' suffix synthesised on an English project. Fixes: scene_no_status INFO
+ no synth status for scenes; broadened central-macro detection (generic 'all
blinds/shutters/...'); language-matched repair suffixes. Demo missing_status
warnings ~7 -> 1 genuine. tests/test_council_fixes.py; full suite green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Rank findings by significance: config_value (setpoint/hysteresis/time/threshold —
an odd device is usually a real mistake) vs config_flag (mode/type) vs label
(per-room text, often intentional). New focus list surfaces the config-value clear
outliers first. On a real 275-device project this turns 422 raw outliers into a
9-item focus that includes the thermostats whose init-setpoint differs from their
siblings. Multilingual name hints (EN/DE/RU). Test updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Validate a project against ITS OWN rules, not a universal standard. With a YAML
profile the declared main-group taxonomy / naming / pairing is authoritative;
with no profile the taxonomy is INFERRED from the project itself and GAs that
deviate from their main group's own majority domain are flagged — never against
an alien standard (a well-organised real 356-GA project drops from 329 false
mismatches vs the default to 31 genuine self-deviations). Answers the recurring
integrator critique that 'your best practices aren't universal'. Example profile
+ test_policy.py included.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
New module param_check.py + MCP tool check_device_parameters: reads per-device
ParameterInstanceRef values from the .knxproj project part (xknxproject does not
expose them), groups identical devices by application program, and flags the odd
one out — clear_outliers (a strong majority with a small minority, e.g. one
thermostat with a different setpoint/hysteresis) and split_configs (balanced
variants → review). Parameter names resolved from the app-program; encrypted
projects skipped honestly. Read-only, no ETS/bus. Validated on real 42-275-device
projects; synthetic test_param_check.py. Answers a community feature request.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
From validator to repairer: propose concrete fixes, not just flag problems.
- B1 repair-suggestion engine (repair.py) + suggest_repairs MCP tool: infer a
DPT for a missing-DPT GA, correct a suspect sub-DPT, synthesise a status GA in
a free slot, or add an absolute-brightness GA. Suggestions only; accepted new
GAs feed generate_ets_group_addresses; never writes to ETS or the bus.
- A2 relative-only-dimming detector (analyze.py): 3.007 relative dimmer with no
5.001 absolute-brightness GA in its zone -> HA cannot set a level.
- A3 cover invert/travel-time surfacing (generate_ha.py): verify_cover_invert
review note lists actuator-dependent flags absent from the .knxproj and warns
when position lacks a state address.
- A2/A3/B1 regression tests; CHANGELOG + version bump to 0.5.0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A1 — sub-DPT sanity linter: when a GA name implies a specific DPT
sub-type (temperature->9.001, power->14.056, brightness/position->5.001),
flag a wrong sub or wrong main. Multilingual, conservative; surfaced as
the subdpt_suspect finding via check_dpt / analyze_all.
A4 — KNX Data Secure posture: secure_posture() + new check_secure MCP
tool. Report-only summary (secured vs plaintext counts, mixed
secure/plaintext middle groups, keyring handover checklist). Reads only
the per-GA Security flag; no key material touched. Handover pack section
5 rewritten to this posture section.
Version bump to 0.4.0; CHANGELOG updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Turn the tool from a .knxproj validator into a design aid.
- device_library.py + decompose_device / list_device_recipes MCP tools:
expand a device (order number / type / alias) into its group-address
recipe — command/status/dimming/position/mode objects with DPTs —
across Zennio + ABB families. Generic vendor facts, typical-wired set.
- docs/spec-to-structure.md: the spec→structure methodology, with a
measured account of what a spec reproduces (~90%) vs the per-device
object count it cannot (2–9× per-project parameterisation).
- Ship alongside the Track B generate_handover_pack and the de-noise
refinements accumulated since 0.2.0.
- CHANGELOG 0.3.0; version bump 0.2.0 → 0.3.0; new device-library tests.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Three false-alarm downgrades derived from validating on a larger multi-vendor
villa, each turning a false positive into an INFO note without hiding any real
defect:
- intent: divider/separator names (punctuation-only or a marker wrapped in it,
e.g. "-----addition------", "---") classify as `scratch`, so a missing DPT on
them is INFO, not a red error.
- analyze: typed GAs wired into Zennio "[LF] ... Data Entry" type-agnostic
logic-function containers surface as INFO `dpt_on_logic_object` instead of a
false `dpt_mismatch_co` warning.
- analyze: all-groups broadcast/central-macro commands surface as INFO
`central_macro_no_status` instead of `missing_status_address`, since a fan-out
broadcast has no single state to read back.
Adds a regression block covering all three; real DPT mismatches and real
missing-status gaps still warn. Unreleased; no version bump.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bump version 0.1.2 -> 0.2.0 (pyproject + __init__). Promote the
Unreleased changelog section to [0.2.0] — 2026-06-30. Scrub a client
city label from two test comments (provenance note only, no PII/topology).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Colour control assembled into light entities (RGB 232.600, RGBW 251.600,
xyY 242.600, colour-temp 7.600) with statuses, matched by zone identity.
Climate entities generated from HVAC mode zones (temp 9.001 + target status
+ operation/controller mode 20.102/20.105 + valve command_value 5.001),
emitted only when HA-required keys are present, else routed to review.
B1 fix: a command no longer borrows a sibling's status (identity must nest,
status maps to exactly one entity) — removed all shared-status addresses.
New DPTs 232.600/251.600/242.600/7.600/20.105/1.100. HA keys verified vs docs.
Real files: signed demo 6 climate + 4 colour lights; Zennio 685-GA 7 climate;
zero shared-status; all 10 regression groups pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Classify every group address as functional/reserve/logic/scratch (intent.py)
and reclassify intentional non-functional GAs instead of crying wolf:
- reserve spare with no DPT -> INFO reserve_without_dpt (not a missing_dpt error)
- reserve name reused across DPTs -> no duplicate_name / inconsistent_dpt
- logic/virtual + scratch GAs -> excluded from missing-status warnings
dpt_mismatch_co and all real functional findings are untouched.
Driven by a real 685-GA Zennio project: false errors 29 -> 6,
missing-status noise 79 -> 45, all 12 real dpt_mismatch_co catches preserved.
Intent breakdown now exposed via list_group_addresses, report inventory and
the analyze_all summary. New regression test covers the reserve/logic/scratch
patterns and proves real DPT + status problems still surface.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Round-2 fixes found by running against more public ETS4/5/6 fixtures (yene
DemoCase, KnxProjParser, knxray, open-knxviewer):
- Lights assemble both on/off and brightness status via identity-based pairing
and fold the on/off command into one light entity (no duplicate switch),
pairing even when device identity is a single name token. Switches too.
- Cover up/down recognised on 1.008 or any 1.x named up/down; stop on
1.007/1.010/1.017 or a stop-name. Siblings attach only to the same shutter
(zone-identity guard) so multiple blinds don't cross-wire.
- Date/time/text DPTs (10/11/16/19) route to review (manual_datetime/manual_text).
- Add regression tests for all three.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Venetian slat GAs (Lamelle/slat/tilt) attach to their parent blind cover as
move_short_address instead of becoming standalone covers; unmatched slats are
flagged shutter_slat_unattached for manual attachment.
- 1-bit diagnostics GAs (wind/frost/smoke/leak alarm, fault) become a read-only
binary_sensor instead of a phantom command switch.
- Add regression tests for both.
Backlog item #6 from VALIDATION-FINDINGS, surfaced by real ETS fixtures.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- generate_ha now lists every unmapped GA in `review` instead of dropping it
silently ("no silent caps"); YAML header reports the review count.
- Shutter category recognises German/directional names (Behang, Lamelle,
auf/ab, Raffstore, Markise, RU equivalents). Shutter-looking commands with an
incomplete DPT get an actionable `shutter_incomplete_dpt` review hint instead
of vanishing.
- Add regression tests for both.
Backlog items #4 and #5 from VALIDATION-FINDINGS, surfaced by real ETS fixtures.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Implement pairing.function_status_pairs(): pair command<->status GAs from ETS
Function roles (SwitchOnOff<->InfoOnOff etc.), used by check_missing_status and
the HA generator. Fixes false "missing status" on function-paired commands and
gives HA entities the correct state_address even when the feedback GA is named
only "Status". Add a regression test.
Found via real-ETS6 fixture testing: Functions were previously ignored despite
the README claiming they were the primary pairing signal.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The server tool function `load_project` shadowed the imported project loader,
so it called itself instead of the parser -> RecursionError on every real
.knxproj. Alias the import as `load_project_file` and call that.
Add a regression test (server.load_project on a bad path must not raise
RecursionError) and a CHANGELOG entry. Found via end-to-end testing against
real ETS5/ETS6 fixtures from XKNX/xknxproject.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Design-time MCP server that reads .knxproj (read-only), validates naming/DPT/status,
and generates Home Assistant KNX YAML + ETS-importable group addresses (XML/CSV).
No live bus access — confined-workspace writes only.
Includes: 12 MCP tools, end-to-end smoke test, MIT license, English-first README
(+ Russian), CONTRIBUTING with a real-project test call, SECURITY policy, CHANGELOG,
GitHub Actions CI (Python 3.10–3.12), and issue/PR templates.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
