Previous value: -"How much of the record to return. Handled by the MCP server, not the JobMojito API — it only narrows the response, never the query.\n\n- `summary`: scores, the overall AI analysis, and each question with the candidate's answer — no per-answer AI commentary, recording paths or raw assessment data. Use this to review or compare candidates.\n- `standard`: everything a human reviewer reads: the full transcript with per-answer analysis, scores and recordings, minus the raw machine assessment blobs. This is the default.\n- `full`: the API response verbatim, including the raw per-answer pronunciation/sentiment data. Large — a long interview can exceed the result limit and fail. Only ask for this if you need those raw fields."New value: +"How much of the record to return. Handled by the MCP server, not the JobMojito API — it only narrows the response, never the query.\n\n- `summary`: scores, the overall AI analysis, and each question with the candidate's answer — no per-answer AI commentary, recording paths or raw assessment data. Use this to review or compare candidates.\n- `standard`: everything a human reviewer reads: the full transcript with per-answer analysis, scores and recordings, minus the raw machine assessment blobs. This is the default.\n- `full`: every field, including the per-answer raw assessment data — condensed: pronunciation and sentiment scores per recognised segment with the lowest-scoring words, reading/gaze statistics, STT reliability, proctor status and seconds per detected speaker. Word-, character- and frame-level arrays are left out (listed in `_omitted`). Only ask for this if you need those assessment details."

Previous value: -"Echo the conversation_id from the server's previous response. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it."New value: +"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request."

Added value: +{
+  "description": "Echo the conversation_id from the server's previous response. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it.",
+  "type": "string"
+}

Added value: +{
+  "description": "Server-issued metadata for this conversation.",
+  "properties": {
+    "conversation_id": {
+      "description": "The server-issued conversation identifier.",
+      "type": "string"
+    }
+  },
+  "type": "object"
+}

Previous value: -"Session-level AI analyses keyed by analysis name (jsonb). `hiring_reasons`: { reasons_to_hire: string[], reasons_not_to_hire: string[] } (0-4 items each, recruiter-facing). `takeaways`: { what_went_well: string[], next_steps: { text, priority: impact | quick | longterm, guidance: { kind: say | structure | length | exercise | perspective, text }, proof? }[], fixed?: string[] } (coaching / persona sessions, learner-facing; `proof` entries are { kind: quote | ai_summary | transcript_ref | document, text, ref?: { result_question_id } }; `fixed` lists the steps from the previous attempt that the learner did this time). `multi_department`: the multi-department-selection analysis. `score_rationale`: the rubric bracket of the session score and evidenced counts. `key_statement`: { quote, result_question_id }, the one verbatim quotation a recruiter would remember the candidate by (interview / persona_interview, absent when nothing distinctive was said). Further keys may be added."New value: +"Session-level AI analyses keyed by analysis name (jsonb). `strengths_concerns`: { strengths: string[], concerns: string[] } (0-4 items each, recruiter-facing; called `hiring_reasons` with `reasons_to_hire` / `reasons_not_to_hire` inside before 2026-09-15, and rows written before then may still carry that shape). `takeaways`: { what_went_well: string[], next_steps: { text, priority: impact | quick | longterm, guidance: { kind: say | structure | length | exercise | perspective, text }, proof? }[], fixed?: string[] } (coaching / persona sessions, learner-facing; `proof` entries are { kind: quote | ai_summary | transcript_ref | document, text, ref?: { result_question_id } }; `fixed` lists the steps from the previous attempt that the learner did this time). `multi_department`: the multi-department-selection analysis. `score_rationale`: the rubric bracket of the session score and evidenced counts. `key_statement`: { quote, result_question_id }, the one verbatim quotation a recruiter would remember the candidate by (interview / persona_interview, absent when nothing distinctive was said). Further keys may be added."

Previous value: -"Reasons to hire (list). Deprecated: a copy of ai_analysis_other.hiring_reasons.reasons_to_hire, kept for existing integrations."New value: +"Reasons to hire (list). Deprecated: a copy of ai_analysis_other.strengths_concerns.strengths, kept for existing integrations."

Previous value: -"Reasons not to hire (list). Deprecated: a copy of ai_analysis_other.hiring_reasons.reasons_not_to_hire, kept for existing integrations."New value: +"Reasons not to hire (list). Deprecated: a copy of ai_analysis_other.strengths_concerns.concerns, kept for existing integrations."

Added value: +{
+  "description": "Session-level AI analyses keyed by analysis name (jsonb). `hiring_reasons`: { reasons_to_hire: string[], reasons_not_to_hire: string[] } (0-4 items each, recruiter-facing). `takeaways`: { what_went_well: string[], next_steps: { text, priority: impact | quick | longterm, guidance: { kind: say | structure | length | exercise | perspective, text }, proof? }[], fixed?: string[] } (coaching / persona sessions, learner-facing; `proof` entries are { kind: quote | ai_summary | transcript_ref | document, text, ref?: { result_question_id } }; `fixed` lists the steps from the previous attempt that the learner did this time). `multi_department`: the multi-department-selection analysis. `score_rationale`: the rubric bracket of the session score and evidenced counts. `key_statement`: { quote, result_question_id }, the one verbatim quotation a recruiter would remember the candidate by (interview / persona_interview, absent when nothing distinctive was said). Further keys may be added."
+}

Previous value: -"Reasons to hire (list/text)."New value: +"Reasons to hire (list). Deprecated: a copy of ai_analysis_other.hiring_reasons.reasons_to_hire, kept for existing integrations."

Previous value: -"Reasons not to hire (list/text)."New value: +"Reasons not to hire (list). Deprecated: a copy of ai_analysis_other.hiring_reasons.reasons_not_to_hire, kept for existing integrations."

Added value: +{
+  "default": "standard",
+  "description": "How much of the record to return. Handled by the MCP server, not the JobMojito API — it only narrows the response, never the query.\n\n- `summary`: scores, the overall AI analysis, and each question with the candidate's answer — no per-answer AI commentary, recording paths or raw assessment data. Use this to review or compare candidates.\n- `standard`: everything a human reviewer reads: the full transcript with per-answer analysis, scores and recordings, minus the raw machine assessment blobs. This is the default.\n- `full`: the API response verbatim, including the raw per-answer pronunciation/sentiment data. Large — a long interview can exceed the result limit and fail. Only ask for this if you need those raw fields.",
+  "enum": [
+    "summary",
+    "standard",
+    "full"
+  ],
+  "type": "string"
+}