feat(competences): extract batch candidates

2026-05-29 11:25:00 +02:00
parent 4ba426c205
commit e8a0fb0e42
60 changed files with 18176 additions and 0 deletions
--- a/data/competences/candidate/saisir_texte_word.yaml
+++ b/data/competences/candidate/saisir_texte_word.yaml
@@ -0,0 +1,128 @@
+schema_version: 1
+id: saisir_texte_word
+name: Saisir du texte dans Word
+version: 1
+learning_state: candidate
+
+intent:
+  fr: saisir du texte dans un document Word actif
+
+parameters:
+  text: "Ceci est un test word !"
+
+preconditions:
+  - id: word_document_active
+    kind: active_window
+    any_of:
+      - title_in: ["Document2 - Word"]
+      - process_active: WINWORD.EXE
+
+methods:
+  - id: text_input_word_concat
+    kind: text_input
+    primitive_ref: text_input_focused
+    parameters:
+      text: "Ceci est un test word !"
+      concat_rule: concat_in_order
+    description: "Saisie texte par fragments dans un document Word deja focus"
+    observed: true
+    trace_source: live_events.jsonl
+    concat_rule: "join(selected text_input events in segment)"
+    reconstructed_text: "Ceci est un test word !"
+
+success_marker:
+  mode: all_of
+  timeout_ms: 5000
+  markers:
+    - kind: active_window_title_in
+      values: ["Document2 - Word"]
+    - kind: active_process_name_is
+      value: WINWORD.EXE
+    - kind: text_input_reconstructed_equals
+      value: "Ceci est un test word !"
+      evidence_source: trace_text_input_concat
+  supervised_requires:
+    - kind: ocr_contains
+      text: "Ceci est un test word !"
+      region_hint: document_body
+      evidence_state: hypothesis_offline
+      required_for: supervised_or_replay_verified
+
+failure_message_template:
+  intention: saisir du texte dans un document Word actif
+  attendu: voir le texte attendu apparaitre dans le corps du document Word
+  vu: "{observed_human_state}"
+  demande: placer le curseur dans le document Word puis saisir le texte attendu
+
+chain_refs:
+  source_session: sess_20260330T175739_6e190b
+  machine_id: DESKTOP-58D5CAC_windows
+  streaming_session_path: data/training/live_sessions/streaming_sessions/sess_20260330T175739_6e190b.json
+  live_events_path: data/training/live_sessions/DESKTOP-58D5CAC_windows/sess_20260330T175739_6e190b/live_events.jsonl
+  cleaned_segment:
+    status: documented_offline
+    keep_event_indices: [34, 35, 36, 37, 38, 39, 40]
+    method_event_indices: [34, 35, 37, 38, 39]
+    success_event_indices: [40]
+    excluded_event_indices: [36]
+    stop_before_event_index: 41
+    stop_before:
+      - extra_newline_after_text_entry
+      - date_and_email_text_input_later_in_session
+      - word_window_clicks_and_document_switching
+      - systray_interaction
+      - python_focus
+    ignored_between_method_and_success:
+      - heartbeat_without_window_metadata
+    ignored_after_success: []
+    notes:
+      - "Le segment demarre apres l'ouverture/focus de Document2 - Word, qui n'est pas revendiquee par cette competence."
+      - "Event #36 est un heartbeat sans metadonnees fenetre et ne fait pas partie de la saisie."
+      - "Events #34/#35/#37/#38/#39 reconstruisent exactement 'Ceci est un test word !'."
+      - "Event #40 est un text_input newline post-methode, utilise comme preuve que Word reste la fenetre active juste apres la saisie."
+      - "Le texte visible n'est pas prouve par OCR offline; l'OCR est reserve au replay/supervised."
+  workflow_pipeline_id: null
+  graph_node_id: null
+  faiss_state_signatures: []
+  target_memory_keys: []
+  dashboard_knowledge_visible: false
+
+promotion:
+  history:
+    - at: "2026-05-28T11:05:00+02:00"
+      from: observed
+      to: candidate
+      by: Dom
+      reason: "GO explicite apres ACK Claude/Qwen du P2 observed."
+  candidate_requires:
+    - cleaned_segment_validated
+    - method_trace_present
+    - success_marker_defined
+    - failure_message_template_valid
+    - primitive_ref_satisfied
+  supervised_requires:
+    - replay_verified_once
+    - success_marker_matched_after_action
+    - ocr_or_replay_verified_text
+    - human_validation
+  stable_requires:
+    min_successes: 3
+    distinct_contexts: 3
+    max_unexplained_failures: 0
+  t2_known_gaps:
+    - id: marker_continuation_human
+      description: "success_event #40 est un text_input humain post-methode."
+      impact: "T2 non satisfaisable tel quel: Lea ne produit pas de text_input newline supplementaire apres la methode."
+      proposed_resolution: "Ajouter wait_state apres saisie ou verifier le texte par OCR/runtime avant promotion supervised."
+      acted_by: Dom
+      acted_at: "2026-05-28T11:50:00+02:00"
+
+generalisation:
+  seen_contexts: []
+  method_success_rate: {}
+  variance_log: []
+
+failure_log: []
+
+created_at: "2026-05-28T10:55:00+02:00"
+last_updated_at: "2026-05-28T11:05:00+02:00"