@they-juanreina/compost-cli 0.1.0-rc.1

package/dist/lib/schemas.generated.js

@@ -0,0 +1,975 @@
+/**
+ * GENERATED — DO NOT EDIT. Run `pnpm --filter compost-cli run build` to regenerate.
+ *
+ * Embeds every schema/*.json file as a typed constant so validate.ts can use
+ * them without filesystem IO. The CLI can be published standalone without
+ * the schema/ tree because the schemas are now part of the compiled bundle.
+ */
+/* eslint-disable */
+export const ANSWER_SCHEMA = {
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "$id": "https://compost.dev/schema/answer/1.0.json",
+    "title": "Compost Grounded Answer",
+    "description": "A RAG answer whose every claim carries a citation. Used by the deterministic citation validator for non-Anthropic providers (#46).",
+    "type": "object",
+    "required": [
+        "answer",
+        "claims"
+    ],
+    "additionalProperties": false,
+    "properties": {
+        "answer": {
+            "type": "string",
+            "minLength": 1
+        },
+        "claims": {
+            "type": "array",
+            "items": {
+                "$ref": "#/$defs/claim"
+            }
+        },
+        "insufficient_evidence": {
+            "type": "boolean",
+            "description": "True when retrieval came up short; answer should say so and claims may be empty."
+        }
+    },
+    "$defs": {
+        "claim": {
+            "type": "object",
+            "required": [
+                "quote",
+                "utterance_id",
+                "session_id",
+                "confidence"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "quote": {
+                    "type": "string",
+                    "minLength": 1
+                },
+                "utterance_id": {
+                    "type": "string",
+                    "pattern": "^U-[0-9]{4,}$"
+                },
+                "session_id": {
+                    "type": "string"
+                },
+                "confidence": {
+                    "type": "number",
+                    "minimum": 0,
+                    "maximum": 1
+                }
+            }
+        }
+    }
+};
+export const CUES_TAXONOMY = {
+    "version": "1.0",
+    "title": "Compost Audio Cue Taxonomy",
+    "description": "Canonical catalog of audio cue kinds. The transcript schema's cue.kind enum (schema/transcript.schema.json#/$defs/cue/properties/kind) is kept in sync with this list. Compost cues are audio-only by design — no pose or gesture classification.",
+    "kinds": [
+        {
+            "kind": "laughter",
+            "description": "Voiced laughter from a speaker — short bursts (\"ha\") or sustained chuckles. Detected by Whisper-AT / Calm-Whisper event-tag tokens.",
+            "confidence_floor": 0.7,
+            "source": "audio"
+        },
+        {
+            "kind": "sigh",
+            "description": "Audible voiced exhale, typically signaling resignation, relief, or thinking-out-loud. Detected by event-tag tokens.",
+            "confidence_floor": 0.6,
+            "source": "audio"
+        },
+        {
+            "kind": "cough",
+            "description": "Single or chained cough. High-energy short burst; distinctive spectral signature.",
+            "confidence_floor": 0.75,
+            "source": "audio"
+        },
+        {
+            "kind": "throat-clear",
+            "description": "Brief vocal-fry clear of the throat, often before a new utterance.",
+            "confidence_floor": 0.65,
+            "source": "audio"
+        },
+        {
+            "kind": "unintelligible",
+            "description": "Speech the ASR could not transcribe with acceptable confidence. Boundary inherited from the failed Whisper segment.",
+            "confidence_floor": 0.5,
+            "source": "audio"
+        },
+        {
+            "kind": "code-switching",
+            "description": "Speaker switched languages mid-utterance. Whisper-large-v3 emits language tokens that mark the transition.",
+            "confidence_floor": 0.7,
+            "source": "audio"
+        },
+        {
+            "kind": "overlap",
+            "description": "Two or more speakers vocalizing simultaneously. Detected by pyannote-audio overlap detection.",
+            "confidence_floor": 0.7,
+            "source": "audio"
+        },
+        {
+            "kind": "interruption",
+            "description": "Turn-stealing overlap where speaker B begins while speaker A is still speaking and A yields. Heuristic over overlap + turn-change.",
+            "confidence_floor": 0.65,
+            "source": "audio"
+        }
+    ]
+};
+export const EVENTS_SCHEMA = {
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "$id": "https://compost.dev/schema/event/1.0.json",
+    "title": "Compost Provenance Event",
+    "description": "Append-only event recording a change to a compost artifact. Mirrors the .compost/events.sqlite row shape. Three-actor model: researcher | agent | ai.",
+    "type": "object",
+    "required": [
+        "id",
+        "ts",
+        "artifact_kind",
+        "artifact_id",
+        "action",
+        "actor_type",
+        "actor_id",
+        "payload"
+    ],
+    "additionalProperties": false,
+    "properties": {
+        "id": {
+            "type": "string",
+            "pattern": "^[0-9A-HJKMNP-TV-Z]{26}$",
+            "description": "ULID (Crockford base32, 26 chars). Sortable and globally unique."
+        },
+        "ts": {
+            "type": "string",
+            "format": "date-time",
+            "description": "ISO 8601 UTC timestamp the event was recorded."
+        },
+        "artifact_kind": {
+            "type": "string",
+            "minLength": 1,
+            "description": "Kind of artifact this event affects: highlight, code, theme, term, insight, comment, frame, frame_annotation, utterance_annotation, code_theme_link, term_utterance_link, etc. Free-form string to allow extension without schema migration."
+        },
+        "artifact_id": {
+            "type": "string",
+            "pattern": "^[a-f0-9]{64}$",
+            "description": "SHA256 of the artifact's initial state. Stable across the artifact's lifetime."
+        },
+        "action": {
+            "enum": [
+                "create",
+                "update",
+                "endorse",
+                "reject",
+                "link",
+                "unlink"
+            ],
+            "description": "Lifecycle action. AI-authored artifacts surface as un-endorsed (human_approved=false) until an `endorse` event from a researcher promotes them."
+        },
+        "actor_type": {
+            "enum": [
+                "researcher",
+                "agent",
+                "ai"
+            ],
+            "description": "researcher = human, accountable. agent = deterministic-ish software actor (skill, loop, slash command). ai = raw LLM output, untrusted by default."
+        },
+        "actor_id": {
+            "type": "string",
+            "minLength": 1,
+            "description": "For researchers: handle or email. For agents: name@semver. For ai: model identifier (e.g. anthropic:claude-opus-4-7)."
+        },
+        "agent_name": {
+            "type": "string",
+            "description": "Required when actor_type=agent. Identifies the software actor (e.g. cross-session-similarity-scanner)."
+        },
+        "agent_version": {
+            "type": "string",
+            "description": "Required when actor_type=agent. Semver."
+        },
+        "prompt_hash": {
+            "type": "string",
+            "pattern": "^[a-f0-9]{64}$",
+            "description": "Required when actor_type=ai or actor_type=agent and the agent invoked an LLM. SHA256(prompt + model + temperature + context_window)."
+        },
+        "model": {
+            "type": "string",
+            "description": "Required when actor_type=ai. The model that produced the suggestion (e.g. anthropic:claude-opus-4-7, ollama:llama3.1:8b)."
+        },
+        "payload": {
+            "description": "Action-specific JSON payload. The structure varies by (artifact_kind, action) pair and is validated separately by the CLI when serializing/deserializing.",
+            "oneOf": [
+                {
+                    "type": "object"
+                },
+                {
+                    "type": "array"
+                },
+                {
+                    "type": "null"
+                }
+            ]
+        },
+        "parent_event": {
+            "type": [
+                "string",
+                "null"
+            ],
+            "pattern": "^[0-9A-HJKMNP-TV-Z]{26}$",
+            "description": "ULID of the event this one continues from. null for root events. Endorse/reject/update/unlink events MUST reference the event being acted on; link events may have null parent_event for the first link of a relationship."
+        },
+        "batch_id": {
+            "type": [
+                "string",
+                "null"
+            ],
+            "minLength": 1,
+            "description": "Optional opaque grouping for events emitted by one loop run, for transactional rollback and audit."
+        }
+    },
+    "allOf": [
+        {
+            "if": {
+                "properties": {
+                    "actor_type": {
+                        "const": "agent"
+                    }
+                },
+                "required": [
+                    "actor_type"
+                ]
+            },
+            "then": {
+                "required": [
+                    "agent_name",
+                    "agent_version"
+                ]
+            }
+        },
+        {
+            "if": {
+                "properties": {
+                    "actor_type": {
+                        "const": "ai"
+                    }
+                },
+                "required": [
+                    "actor_type"
+                ]
+            },
+            "then": {
+                "required": [
+                    "model",
+                    "prompt_hash"
+                ]
+            }
+        },
+        {
+            "if": {
+                "properties": {
+                    "action": {
+                        "enum": [
+                            "endorse",
+                            "reject",
+                            "update",
+                            "unlink"
+                        ]
+                    }
+                },
+                "required": [
+                    "action"
+                ]
+            },
+            "then": {
+                "required": [
+                    "parent_event"
+                ]
+            }
+        }
+    ]
+};
+export const FRAMES_TAXONOMY = {
+    "version": "1.0",
+    "title": "Compost Frame Trigger Catalog",
+    "description": "Canonical catalog of frame triggers — the reasons a screenshot may be captured from the video stream during transcription. The transcript schema's frame.trigger enum (schema/transcript.schema.json#/$defs/frame/properties/trigger) is kept in sync with this list. default_enabled reflects the 'balanced' default profile (decision #73).",
+    "triggers": [
+        {
+            "trigger": "silence_after_question",
+            "description": "Frame captured during a silence the silence-typer classified as after_question. Pauses after moderator questions are high-signal.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "silence_mid_utterance",
+            "description": "Frame captured during a mid-utterance silence — speaker paused inside their own turn, often searching for a word.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "silence_thinking",
+            "description": "Frame captured during a thinking silence — speaker hasn't committed to answering yet.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "silence_interruption",
+            "description": "Frame captured at a silence inside an interruption — turn-stealing dynamics often visible.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "audio_cue",
+            "description": "Frame captured when an audio cue fires (laughter, sigh, cough, throat-clear). Linked to the cue id.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "shot_change",
+            "description": "Frame captured when the perceptual-hash distance between consecutive sampled frames crosses the shot-change threshold. Catches camera/screen-share switches.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "highlight",
+            "description": "Frame captured at the moment a researcher creates a highlight. The contemporaneous frame is auto-linked to the highlight.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "manual",
+            "description": "Frame captured on explicit researcher request via `compost snap` or a one-click control in the web player.",
+            "default_enabled": true
+        },
+        {
+            "trigger": "sampling",
+            "description": "Fallback frame captured at fixed intervals (60s default) when none of the other triggers have fired recently. Ensures every minute of every session has at least one frame.",
+            "default_enabled": true
+        }
+    ]
+};
+export const STATUS_SCHEMA = {
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "$id": "https://compost.dev/schema/status/1.0.json",
+    "title": "Compost Status Snapshot",
+    "description": "Structured snapshot of every seed in the working tree. Stable contract for agents calling `compost status --json`.",
+    "type": "object",
+    "required": [
+        "schema_version",
+        "generated_at",
+        "root",
+        "seeds"
+    ],
+    "additionalProperties": false,
+    "properties": {
+        "schema_version": {
+            "const": "1.0"
+        },
+        "generated_at": {
+            "type": "string",
+            "format": "date-time"
+        },
+        "root": {
+            "type": "string",
+            "description": "Absolute path of the Seeds/ root that was scanned."
+        },
+        "seeds": {
+            "type": "array",
+            "items": {
+                "$ref": "#/$defs/seed"
+            }
+        }
+    },
+    "$defs": {
+        "seed": {
+            "type": "object",
+            "required": [
+                "name",
+                "path",
+                "counts"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "name": {
+                    "type": "string"
+                },
+                "path": {
+                    "type": "string"
+                },
+                "status": {
+                    "type": [
+                        "string",
+                        "null"
+                    ],
+                    "description": "From seed.md frontmatter. Common values: planning | active | synthesis | done."
+                },
+                "owners": {
+                    "type": "array",
+                    "items": {
+                        "type": "string"
+                    }
+                },
+                "created_at": {
+                    "type": [
+                        "string",
+                        "null"
+                    ]
+                },
+                "counts": {
+                    "$ref": "#/$defs/counts"
+                }
+            }
+        },
+        "counts": {
+            "type": "object",
+            "required": [
+                "sessions",
+                "highlights",
+                "codes",
+                "themes",
+                "frames",
+                "insights",
+                "legacy_assets"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "sessions": {
+                    "$ref": "#/$defs/sessionCounts"
+                },
+                "highlights": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "codes": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "themes": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "frames": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "insights": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "legacy_assets": {
+                    "type": "integer",
+                    "minimum": 0
+                }
+            }
+        },
+        "sessionCounts": {
+            "type": "object",
+            "required": [
+                "total",
+                "transcribed",
+                "queued",
+                "inbox"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "total": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "transcribed": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "queued": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "inbox": {
+                    "type": "integer",
+                    "minimum": 0
+                }
+            }
+        }
+    }
+};
+export const TRANSCRIPT_SCHEMA = {
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "$id": "https://compost.dev/schema/transcript/1.0.json",
+    "title": "Compost Rich Transcript",
+    "description": "Per-session rich transcript with utterances, typed silences, audio cues, frames, prosody hints, and provenance pointer.",
+    "type": "object",
+    "required": [
+        "schema_version",
+        "session_id",
+        "source",
+        "language",
+        "duration_ms",
+        "modality",
+        "speakers",
+        "utterances",
+        "provenance"
+    ],
+    "additionalProperties": false,
+    "properties": {
+        "schema_version": {
+            "const": "1.0",
+            "description": "Locked at 1.0 for this contract. Breaking changes bump the major."
+        },
+        "kind": {
+            "enum": [
+                "session",
+                "document"
+            ],
+            "default": "session",
+            "description": "session = recorded interview; document = legacy PDF/DOCX/PPTX/CSV normalized into a transcript-shaped doc."
+        },
+        "session_id": {
+            "type": "string",
+            "pattern": "^[A-Za-z0-9_-]+$",
+            "description": "Stable, filesystem-safe session identifier (e.g. \"S023\")."
+        },
+        "source": {
+            "type": "string",
+            "description": "Path to the source media file, relative to the seed root."
+        },
+        "language": {
+            "type": "string",
+            "description": "BCP-47 language tag (e.g. \"es-CO\", \"en-US\")."
+        },
+        "duration_ms": {
+            "type": "integer",
+            "minimum": 0
+        },
+        "modality": {
+            "type": "array",
+            "items": {
+                "enum": [
+                    "audio",
+                    "video",
+                    "document"
+                ]
+            },
+            "minItems": 1,
+            "uniqueItems": true
+        },
+        "speakers": {
+            "type": "array",
+            "items": {
+                "$ref": "#/$defs/speaker"
+            }
+        },
+        "utterances": {
+            "type": "array",
+            "items": {
+                "$ref": "#/$defs/utterance"
+            }
+        },
+        "silences": {
+            "type": "array",
+            "default": [],
+            "items": {
+                "$ref": "#/$defs/silence"
+            }
+        },
+        "cues": {
+            "type": "array",
+            "default": [],
+            "items": {
+                "$ref": "#/$defs/cue"
+            }
+        },
+        "frames": {
+            "type": "array",
+            "default": [],
+            "items": {
+                "$ref": "#/$defs/frame"
+            }
+        },
+        "glossary_refs": {
+            "type": "array",
+            "default": [],
+            "items": {
+                "$ref": "#/$defs/sessionGlossaryRef"
+            },
+            "description": "Optional session-level rollup of glossary term mentions; derivable from utterances[].glossary_refs."
+        },
+        "provenance": {
+            "$ref": "#/$defs/provenance"
+        }
+    },
+    "$defs": {
+        "speaker": {
+            "type": "object",
+            "required": [
+                "id",
+                "type"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "id": {
+                    "type": "string",
+                    "pattern": "^S[0-9]+$",
+                    "description": "Speaker handle, e.g. \"S1\", \"S2\"."
+                },
+                "name": {
+                    "type": "string"
+                },
+                "type": {
+                    "enum": [
+                        "moderator",
+                        "participant",
+                        "other"
+                    ]
+                }
+            }
+        },
+        "utterance": {
+            "type": "object",
+            "required": [
+                "id",
+                "speaker_id",
+                "turn",
+                "start_ms",
+                "end_ms",
+                "text"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "id": {
+                    "type": "string",
+                    "pattern": "^U-[0-9]{4,}$",
+                    "description": "Zero-padded ordinal, e.g. \"U-0001\"."
+                },
+                "speaker_id": {
+                    "type": "string",
+                    "pattern": "^S[0-9]+$"
+                },
+                "turn": {
+                    "type": "integer",
+                    "minimum": 1
+                },
+                "start_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "end_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "text": {
+                    "type": "string"
+                },
+                "source_page": {
+                    "type": "integer",
+                    "minimum": 1,
+                    "description": "For document-kind transcripts: the 1-based source page/slide/row the utterance came from."
+                },
+                "words": {
+                    "type": "array",
+                    "items": {
+                        "$ref": "#/$defs/word"
+                    }
+                },
+                "prosody": {
+                    "$ref": "#/$defs/prosody"
+                },
+                "annotation": {
+                    "type": "string",
+                    "description": "Free-text editorial layer. May be human-authored, agent-authored, or AI-suggested. Provenance lives in .compost/events.sqlite — this field is a snapshot of current text only."
+                },
+                "glossary_refs": {
+                    "type": "array",
+                    "items": {
+                        "$ref": "#/$defs/utteranceGlossaryRef"
+                    }
+                }
+            }
+        },
+        "word": {
+            "type": "object",
+            "required": [
+                "w",
+                "s",
+                "e"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "w": {
+                    "type": "string"
+                },
+                "s": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "e": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "conf": {
+                    "type": "number",
+                    "minimum": 0,
+                    "maximum": 1
+                }
+            }
+        },
+        "prosody": {
+            "type": "object",
+            "additionalProperties": false,
+            "description": "Deterministic prosody hints derived from word confidence + VAD energy + speech rate. Not ML predictions.",
+            "properties": {
+                "volume": {
+                    "enum": [
+                        "low",
+                        "normal",
+                        "high"
+                    ]
+                },
+                "pace": {
+                    "enum": [
+                        "slow",
+                        "normal",
+                        "fast"
+                    ]
+                },
+                "hesitations": {
+                    "type": "integer",
+                    "minimum": 0
+                }
+            }
+        },
+        "silence": {
+            "type": "object",
+            "required": [
+                "id",
+                "start_ms",
+                "end_ms",
+                "duration_ms",
+                "context"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "id": {
+                    "type": "string",
+                    "pattern": "^SIL-[0-9]{3,}$"
+                },
+                "start_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "end_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "duration_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "context": {
+                    "enum": [
+                        "after_question",
+                        "mid_utterance",
+                        "thinking",
+                        "interruption"
+                    ],
+                    "description": "Semantic typing from the silence-typer heuristic. Versioned rules; researcher can override."
+                },
+                "annotation": {
+                    "type": "string"
+                }
+            }
+        },
+        "cue": {
+            "type": "object",
+            "required": [
+                "id",
+                "kind",
+                "start_ms",
+                "end_ms",
+                "source"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "id": {
+                    "type": "string",
+                    "pattern": "^CUE-[0-9]{3,}$"
+                },
+                "kind": {
+                    "enum": [
+                        "laughter",
+                        "sigh",
+                        "cough",
+                        "throat-clear",
+                        "unintelligible",
+                        "code-switching",
+                        "overlap",
+                        "interruption"
+                    ],
+                    "description": "See cues.taxonomy.json for confidence floors per kind."
+                },
+                "start_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "end_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "source": {
+                    "const": "audio",
+                    "description": "Compost cues are audio-only by design. No pose/gesture classification."
+                },
+                "speaker_id": {
+                    "type": "string",
+                    "pattern": "^S[0-9]+$"
+                },
+                "confidence": {
+                    "type": "number",
+                    "minimum": 0,
+                    "maximum": 1
+                },
+                "annotation": {
+                    "type": "string"
+                }
+            }
+        },
+        "frame": {
+            "type": "object",
+            "required": [
+                "id",
+                "at_ms",
+                "path",
+                "trigger"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "id": {
+                    "type": "string",
+                    "pattern": "^FR-[0-9]{3,}$"
+                },
+                "at_ms": {
+                    "type": "integer",
+                    "minimum": 0
+                },
+                "path": {
+                    "type": "string",
+                    "description": "Path to the captured frame, relative to the seed root."
+                },
+                "trigger": {
+                    "enum": [
+                        "silence_after_question",
+                        "silence_mid_utterance",
+                        "silence_thinking",
+                        "silence_interruption",
+                        "audio_cue",
+                        "shot_change",
+                        "highlight",
+                        "manual",
+                        "sampling"
+                    ],
+                    "description": "See frames.taxonomy.json for trigger semantics. Silence triggers are split by silence context so a frame can carry the same semantic type as its linked silence."
+                },
+                "linked_utterance_id": {
+                    "type": "string",
+                    "pattern": "^U-[0-9]{4,}$"
+                },
+                "annotation": {
+                    "type": "string",
+                    "description": "Optional one-sentence description from a vision-capable LLM. AI-authored events surface as [draft] until endorsed."
+                }
+            }
+        },
+        "utteranceGlossaryRef": {
+            "type": "object",
+            "required": [
+                "term_id",
+                "span"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "term_id": {
+                    "type": "string",
+                    "pattern": "^T-[A-Za-z0-9_-]+$"
+                },
+                "span": {
+                    "type": "array",
+                    "items": {
+                        "type": "integer",
+                        "minimum": 0
+                    },
+                    "minItems": 2,
+                    "maxItems": 2,
+                    "description": "[start, end] character offsets into the parent utterance.text."
+                }
+            }
+        },
+        "sessionGlossaryRef": {
+            "type": "object",
+            "required": [
+                "term_id",
+                "utterance_id",
+                "span"
+            ],
+            "additionalProperties": false,
+            "properties": {
+                "term_id": {
+                    "type": "string",
+                    "pattern": "^T-[A-Za-z0-9_-]+$"
+                },
+                "utterance_id": {
+                    "type": "string",
+                    "pattern": "^U-[0-9]{4,}$"
+                },
+                "span": {
+                    "type": "array",
+                    "items": {
+                        "type": "integer",
+                        "minimum": 0
+                    },
+                    "minItems": 2,
+                    "maxItems": 2
+                }
+            }
+        },
+        "provenance": {
+            "type": "object",
+            "required": [
+                "transcriber"
+            ],
+            "additionalProperties": false,
+            "description": "Tool+version pointers describing how this transcript was produced. Per-artifact event provenance lives in .compost/events.sqlite.",
+            "properties": {
+                "transcriber": {
+                    "type": "string",
+                    "description": "Tool@version that produced this transcript (e.g. \"compost-transcriber@0.3.1\")."
+                },
+                "asr_model": {
+                    "type": "string"
+                },
+                "diarizer": {
+                    "type": "string"
+                },
+                "audio_cues": {
+                    "type": "string"
+                },
+                "frame_capture": {
+                    "type": "string"
+                },
+                "frame_annotation": {
+                    "type": "string"
+                },
+                "text_col_resolved": {
+                    "type": "string",
+                    "description": "For CSV/XLSX legacy ingest: which column was actually used as the text source (auto-detected or explicit). Optional; absent on audio/video transcripts."
+                },
+                "xlsx_rows_skipped_empty_text": {
+                    "type": "integer",
+                    "minimum": 0,
+                    "description": "For XLSX legacy ingest: count of rows where other columns had data but the text column was empty. Often signals un-evaluated formulas — open the file in Excel once or export to CSV. Absent when zero."
+                }
+            }
+        }
+    }
+};
+export const ALL_SCHEMAS = {
+    "answer.schema.json": ANSWER_SCHEMA,
+    "cues.taxonomy.json": CUES_TAXONOMY,
+    "events.schema.json": EVENTS_SCHEMA,
+    "frames.taxonomy.json": FRAMES_TAXONOMY,
+    "status.schema.json": STATUS_SCHEMA,
+    "transcript.schema.json": TRANSCRIPT_SCHEMA,
+};
+//# sourceMappingURL=schemas.generated.js.map