Skip to main content
Glama

kelpo_apply_operations

Apply one conversational turn atomically with the same operations and validations as the Kelpo studio: classify uploads (shared photos with per-photo positions), edit the scene, answer preparation steps, choose actors/outfits/voice, ask the user with a picker, etc. All operations succeed together or none is saved; read receipts (status rejected + code/message explains why). The exchange (message, attachments, your response) is recorded by this call. Retry with the identical request_id/body after a timeout. Never generates or charges: generation stays an explicit user approval (quote + generate). Arguments: {"path": {"id": ""}, "body": {"version": 3, "message": "<the person's exact words of this turn>", "operations": [{"op": "edit_scene", "args": {"fields": ["tone"], "values": {"tone": "posé"}}}]}}

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "path": "<path>",
      "body": "<body>"
    },
    "name": "kelpo_apply_operations"
  }
}

Input Schema

NameRequiredDescriptionDefault
bodyYes
pathYes

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed4 schema fields changed
    • changedInput schema / $defs / EditingLaterArgs / properties / subtitle_style / description
      Previous value: -"shorts (default): bold capitals, a box follows the spoken word. word: one big lowercase word at a time, keywords in yellow or blue. bold: two or three big capital words at a time, keywords in yellow or blue, light shadow. classic: white outlined lines."New value: +"shorts (default): bold capitals, a box follows the spoken word. word: one big lowercase word at a time, keywords in yellow or blue. bold: two or three big capital words at a time, keywords in yellow or blue, light shadow. stack: each sentence builds up word by word at the top in stacked capital lines, spoken word in a glowing box. classic: white outlined lines."
    • changedInput schema / $defs / EditingLaterArgs / properties / subtitle_style / enum
      Previous value: -[
      -  "shorts",
      -  "classic",
      -  "word",
      -  "bold"
      -]New value: +[
      +  "shorts",
      +  "classic",
      +  "word",
      +  "bold",
      +  "stack"
      +]
    • changedInput schema / $defs / EditingPlanArgs / properties / subtitle_style / description
      Previous value: -"shorts (default): one text at a time, bold capitals, keywords coloured, a box follows the spoken word; hooks show alone before the speech, a label replaces the subtitles during its insert. classic: white outlined lines. word: one big lowercase word at a time on the voice, keywords in yellow or blue. bold: two or three big capital words at a time, keywords in yellow or blue, light shadow. With subtitles=keep, a style given here restyles the existing subtitles."New value: +"shorts (default): one text at a time, bold capitals, keywords coloured, a box follows the spoken word; hooks show alone before the speech, a label replaces the subtitles during its insert. classic: white outlined lines. word: one big lowercase word at a time on the voice, keywords in yellow or blue. bold: two or three big capital words at a time, keywords in yellow or blue, light shadow. stack: phone-Reel look at the top, each sentence builds up word by word in stacked capital lines of equal width, spoken word in a glowing violet (or brand) box, keywords in yellow, a short opening phrase stays small above. With subtitles=keep, a style given here restyles the existing subtitles."
    • changedInput schema / $defs / EditingPlanArgs / properties / subtitle_style / enum
      Previous value: -[
      -  "shorts",
      -  "classic",
      -  "word",
      -  "bold"
      -]New value: +[
      +  "shorts",
      +  "classic",
      +  "word",
      +  "bold",
      +  "stack"
      +]
  2. Changed2 schema fields changed
    • changedInput schema / $defs / ChooseActorArgs / properties / action / description
      Previous value: -"select: use a library person (person_id). create: the main actor named by the user, for this scene only (name); never for a secondary actor. remove: take an actor out of this scene (person_id or name), without deleting anything."New value: +"select: use a library person (person_id; the person you are filming, no name to ask). create: the main actor named by the user, for this scene only (name); never for a secondary actor. remove: take an actor out of this scene (person_id or name), without deleting anything."
    • changedInput schema / $defs / Op_choose_wardrobe / properties / op / description
      Previous value: -"Choose what an on-screen actor wears in this video. Use the actor id and photo ids from the state; never pick a\nphoto the user did not designate (random only on explicit indifference). For a delegated request whose domain recipe\nimplies an outfit (UGC: simple everyday clothes matching the place), use mode=custom with that description and announce it."New value: +"Choose what an on-screen actor wears in this video. Use the actor id and photo ids from the state; never pick a\nphoto the user did not designate (random only on explicit indifference). \"the outfit of my photos\" without naming one\n= mode=photo with their first photo, never ask which one. For a delegated request whose domain recipe\nimplies an outfit (UGC: simple everyday clothes matching the place), use mode=custom with that description and announce it."
  3. Changed1 schema field changed
    • changedInput schema / $defs / Op_choose_voice / properties / op / description
      Previous value: -"Record the user's explicit voice decision for the spoken text. A generated voice is never a default: use\nmode=generated only if the user asked for it. Reusing an audio requires one of the offered conversation audios\n(at least 60 s for a new voice reference; a shorter one is refused with voice_too_short, whose data lists the\nalternatives to offer: complete it then assemble_audio, record at the microphone, generated voice, exact words if 2-15 s).\nAn audio assembled earlier in the same plan may be given as asset_id \"@N.asset_ids[0]\" (assigned directly to `person`)."New value: +"Record the user's explicit voice decision for the spoken text. A generated voice does not exist in Kelpo\n(mode=generated is refused with real_voice_required): a spoken video always has a real voice. Reusing an audio requires one of the offered conversation audios\n(at least 60 s for a new voice reference; a shorter one is refused with voice_too_short, whose data lists the\nalternatives to offer: complete it then assemble_audio, record at the microphone, a catalogue actor with a real voice, exact words if 2-15 s).\nAn audio assembled earlier in the same plan may be given as asset_id \"@N.asset_ids[0]\" (assigned directly to `person`)."
  4. Changed2 schema fields changed
    • changedInput schema / $defs / Op_set_location / properties / op / description
      Previous value: -"Set the filming location. Use whenever the user gives the place, even before the location question\nor together with other choices in the same message. Never invent a place, except the one a domain recipe implies\nfor a delegated request (UGC product video: a real everyday room linked to the product, mode=named), announced in\nyour reply; use mode=any only when the user explicitly leaves it free. A name matching several library places is refused: ask which one.\nWhen place photos of this conversation are classified (also earlier in the same plan), the place is those photos:\nuse mode=photos; mode=named then simply labels them (it never removes place photos)."New value: +"Set the filming location. Use whenever the user gives the place, even before the location question\nor together with other choices in the same message. Never invent a place, except the one a domain recipe implies\nfor a delegated request (UGC product video: a real everyday room linked to the product, mode=named), announced in\nyour reply; use mode=any only when the user explicitly leaves it free. A name matching several library places is refused: ask which one.\nWhen place photos of this conversation are classified (also earlier in the same plan), the place is those photos:\nuse mode=photos; mode=named then simply labels them (it never removes place photos).\nstart_frame=true (mode photos, one photo): the take opens exactly on that photo; otherwise it is only the place."
    • addedInput schema / $defs / SetLocationArgs / properties / start_frame
      Added value: +{
      +  "default": false,
      +  "description": "photos, ONE photo (a place photo, or an actor photo whose background is the place): the video OPENS exactly on it (its first frame: same framing, place, light, people and pose), e.g. when the user wants every shot to start on their image. False: the photo is only the place. Conditions: a single take (a video with a montage only uses it as the shoot's reference), the photo at the video's format (16:9 at least 1344×768, 9:16 at least 768×1344, ≤5 % cropped), Normal quality (in HD it only guides the opening).",
      +  "title": "Start Frame",
      +  "type": "boolean"
      +}
  5. Changed4 schema fields changed
    • addedInput schema / $defs / EditFooter
      Added value: +{
      +  "additionalProperties": false,
      +  "description": "Fixed link line (editing_footer.py): lengths checked by editing_footer.problems (translated messages).",
      +  "properties": {
      +    "accent": {
      +      "default": "",
      +      "description": "Colour #rrggbb of the *accented* part. Empty = the title accent colour (cyan).",
      +      "pattern": "^(#[0-9a-fA-F]{6})?$",
      +      "title": "Accent",
      +      "type": "string"
      +    },
      +    "caption": {
      +      "default": "",
      +      "description": "Optional small grey line above the link (max 40, video language). Empty = link alone.",
      +      "maxLength": 200,
      +      "title": "Caption",
      +      "type": "string"
      +    },
      +    "link": {
      +      "default": "",
      +      "description": "The link or short address in bold (max 40 characters without the *), e.g. \"monsite.com/*promo*\"; *word* = accent colour. Copied from the person, never invented.",
      +      "maxLength": 200,
      +      "title": "Link",
      +      "type": "string"
      +    }
      +  },
      +  "title": "EditFooter",
      +  "type": "object"
      +}
    • addedInput schema / $defs / EditingLaterArgs / properties / footer
      Added value: +{
      +  "$ref": "#/$defs/EditFooter",
      +  "description": "Fixed link line for the whole video, only when the person gives a link to show (site, promo or affiliate page): link in bold (max 40, e.g. \"monsite.com/*promo*\", *word* = accent colour), optional small grey caption above it (max 40, video language). The server places it: under the picture in a vertical frame, else above the TikTok / Reels / Shorts captions, clear of the right-hand buttons; subtitles move up. Copy their link exactly, never invent one. Omitted = none."
      +}
    • addedInput schema / $defs / EditingPlanArgs / properties / footer
      Added value: +{
      +  "$ref": "#/$defs/EditFooter",
      +  "description": "Fixed link line for the whole video, only when the person gives a link to show (site, promo or affiliate page): link in bold (max 40, e.g. \"monsite.com/*promo*\", *word* = accent colour), optional small grey caption above it (max 40, video language). The server places it: under the picture in a vertical frame, else above the TikTok / Reels / Shorts captions, clear of the right-hand buttons; subtitles move up. Copy their link exactly, never invent one. Omitted = keep."
      +}
    • addedInput schema / $defs / EditingPlanArgs / properties / remove_footer
      Added value: +{
      +  "default": false,
      +  "description": "True = no link line any more.",
      +  "title": "Remove Footer",
      +  "type": "boolean"
      +}
  6. Changed18 schema fields changed
    • changedInput schema / $defs / ChooseVoiceArgs / properties / mode / description
      Previous value: -"recording: the user has/will add a recording of the speaker's voice (uploaded separately). no_recording: the user has none (then offer adding one later or a generated voice). generated: the user explicitly accepts a generated (robotic) voice instead of the speaker's. reuse: use an audio already sent in this conversation (voice_reuse choices) as this speaker's voice. decline_reuse: none of the offered audios contains this speaker's voice."New value: +"recording: the user has/will add a recording of the speaker's voice (uploaded separately). no_recording: the user has none (then offer a Kelpo catalogue actor, a character created with its voice, or a recording later). generated: refused — Kelpo never makes a talking video without a real voice. reuse: use an audio already sent in this conversation (voice_reuse choices) as this speaker's voice. decline_reuse: none of the offered audios contains this speaker's voice."
    • addedInput schema / $defs / EditClip / properties / pace
      Added value: +{
      +  "default": true,
      +  "description": "With a vertical frame pace above 1: false = this passage keeps x1 (other people visible: passers-by, a crowd, a street interview).",
      +  "title": "Pace",
      +  "type": "boolean"
      +}
    • addedInput schema / $defs / EditingLaterArgs / properties / frame
      Added value: +{
      +  "$ref": "#/$defs/Frame",
      +  "description": "Vertical frame for a 16:9 montage posted as a Short / Reel / TikTok (MONTAGE_FRAME): the picture at its native size in a 9:16 video on a plain background (color), a fixed title above it for the whole video (**accented words** in the accents colours), optional period look (grain, always subtle: light70/light80, film70, vhs80, bw60, broadcast90; title and subtitles stay sharp), speech pace 1-1.3 (1.1/1.2 livelier, same pitch; keep 1 when other people are visible: street, passers-by, crowd). Silences at the joins are cut, no frozen picture at the end. Pair it with subtitles=auto, subtitle_style=word (one word at a time). Write the title in the video's language. The scene must be 16:9 (aspect_ratio \"16:9\"), never 9:16. Refused as unavailable on the account: say so before promising it. Omitted = none."
      +}
    • addedInput schema / $defs / EditingLaterArgs / properties / subtitle_position
      Added value: +{
      +  "default": "keep",
      +  "description": "Where the subtitles sit on the picture: auto, bottom, center or top of the picture.",
      +  "enum": [
      +    "keep",
      +    "auto",
      +    "bottom",
      +    "center",
      +    "top"
      +  ],
      +  "title": "Subtitle Position",
      +  "type": "string"
      +}
    • addedInput schema / $defs / EditingLaterArgs / properties / subtitle_style
      Added value: +{
      +  "default": "shorts",
      +  "description": "shorts (default): bold capitals, a box follows the spoken word. word: one big lowercase word at a time, keywords in yellow or blue. bold: two or three big capital words at a time, keywords in yellow or blue, light shadow. classic: white outlined lines.",
      +  "enum": [
      +    "shorts",
      +    "classic",
      +    "word",
      +    "bold"
      +  ],
      +  "title": "Subtitle Style",
      +  "type": "string"
      +}
    • addedInput schema / $defs / EditingPlanArgs / properties / frame
      Added value: +{
      +  "$ref": "#/$defs/Frame",
      +  "description": "Vertical frame for a 16:9 montage posted as a Short / Reel / TikTok (MONTAGE_FRAME): the picture at its native size in a 9:16 video on a plain background (color), a fixed title above it for the whole video (**accented words** in the accents colours), optional period look (grain, always subtle: light70/light80, film70, vhs80, bw60, broadcast90; title and subtitles stay sharp), speech pace 1-1.3 (1.1/1.2 livelier, same pitch; keep 1 when other people are visible: street, passers-by, crowd). Silences at the joins are cut, no frozen picture at the end. Pair it with subtitles=auto, subtitle_style=word (one word at a time). Write the title in the video's language. 16:9 videos only. Refused as unavailable on the account: say so before promising it. Omitted = keep."
      +}
    • addedInput schema / $defs / EditingPlanArgs / properties / remove_frame
      Added value: +{
      +  "default": false,
      +  "description": "True = back to the video's own format (no vertical frame).",
      +  "title": "Remove Frame",
      +  "type": "boolean"
      +}
    • addedInput schema / $defs / EditingPlanArgs / properties / subtitle_position
      Added value: +{
      +  "default": "keep",
      +  "description": "Where the subtitles sit on the picture: auto (low on a vertical video, bottom of the picture in a vertical frame), bottom, center, top (center/top cover a face speaking to camera: only without a face there); keep = unchanged.",
      +  "enum": [
      +    "keep",
      +    "auto",
      +    "bottom",
      +    "center",
      +    "top"
      +  ],
      +  "title": "Subtitle Position",
      +  "type": "string"
      +}
    • changedInput schema / $defs / EditingPlanArgs / properties / subtitle_style / description
      Previous value: -"shorts (default): one text at a time, bold capitals, keywords coloured, a box follows the spoken word; hooks show alone before the speech, a label replaces the subtitles during its insert. classic: white outlined lines."New value: +"shorts (default): one text at a time, bold capitals, keywords coloured, a box follows the spoken word; hooks show alone before the speech, a label replaces the subtitles during its insert. classic: white outlined lines. word: one big lowercase word at a time on the voice, keywords in yellow or blue. bold: two or three big capital words at a time, keywords in yellow or blue, light shadow. With subtitles=keep, a style given here restyles the existing subtitles."
    • changedInput schema / $defs / EditingPlanArgs / properties / subtitle_style / enum
      Previous value: -[
      -  "shorts",
      -  "classic"
      -]New value: +[
      +  "shorts",
      +  "classic",
      +  "word",
      +  "bold"
      +]
    • addedInput schema / $defs / Frame
      Added value: +{
      +  "additionalProperties": false,
      +  "description": "Vertical frame (MONTAGE_FRAME): a 16:9 montage shown in a 9:16 video for TikTok / Reels / Shorts, at its native size\n(never reduced or enlarged), on a plain background, with a fixed title above it for the whole video.",
      +  "properties": {
      +    "accents": {
      +      "description": "Colours #rrggbb of the accented words, in turn (e.g. [\"#2ed6f2\"] cyan, [\"#ffe14d\",\"#ff5c8a\"]). Empty = cyan.",
      +      "items": {
      +        "type": "string"
      +      },
      +      "maxItems": 4,
      +      "title": "Accents",
      +      "type": "array"
      +    },
      +    "color": {
      +      "default": "#000000",
      +      "description": "Background of the frame #rrggbb (default black).",
      +      "pattern": "^#[0-9a-fA-F]{6}$",
      +      "title": "Color",
      +      "type": "string"
      +    },
      +    "grain": {
      +      "default": "none",
      +      "description": "Period look on the picture only (title and subtitles stay sharp), always subtle (no blur): light70 / light80 = a little grain, warm 70s or vivid 80s (recommended when a look is wanted); film70 = 70s film colours; vhs80 = 80s VHS colours and faint scan lines; bw60 = black and white 60s; broadcast90 = early 90s TV. none (default).",
      +      "enum": [
      +        "none",
      +        "light70",
      +        "light80",
      +        "film70",
      +        "vhs80",
      +        "bw60",
      +        "broadcast90"
      +      ],
      +      "title": "Grain",
      +      "type": "string"
      +    },
      +    "pace": {
      +      "default": 1,
      +      "description": "Speech pace of every passage at normal speed: 1 (default), 1.1 or 1.2 (a bit livelier, same voice pitch). The duration shrinks accordingly. Keep 1 for a shot where other people are visible (street, passers-by, crowd: sped up they look fake), or set that clip's pace false.",
      +      "maximum": 1.3,
      +      "minimum": 1,
      +      "title": "Pace",
      +      "type": "number"
      +    },
      +    "title": {
      +      "default": "",
      +      "description": "Fixed title above the picture for the whole video (max 120, in the video's language), e.g. \"Why I always give **three prices**\". **words** (or *words*) = accent colour; several accented groups take the accents colours in turn. Empty = no title.",
      +      "maxLength": 120,
      +      "title": "Title",
      +      "type": "string"
      +    },
      +    "title_case": {
      +      "default": "as_written",
      +      "enum": [
      +        "as_written",
      +        "upper"
      +      ],
      +      "title": "Title Case",
      +      "type": "string"
      +    },
      +    "title_color": {
      +      "default": "",
      +      "description": "Title colour #rrggbb. Empty = white (black on a light frame).",
      +      "pattern": "^(#[0-9a-fA-F]{6})?$",
      +      "title": "Title Color",
      +      "type": "string"
      +    },
      +    "title_font": {
      +      "default": "classic",
      +      "description": "classic (bold sans, the usual Shorts title), modern (rounded geometric), serif (elegant, old-money), typewriter (archive, report), condensed (tall news headline).",
      +      "enum": [
      +        "classic",
      +        "modern",
      +        "serif",
      +        "typewriter",
      +        "condensed"
      +      ],
      +      "title": "Title Font",
      +      "type": "string"
      +    },
      +    "title_size": {
      +      "default": "medium",
      +      "description": "medium (default); small for a long title, large for 3-6 words.",
      +      "enum": [
      +        "small",
      +        "medium",
      +        "large"
      +      ],
      +      "title": "Title Size",
      +      "type": "string"
      +    }
      +  },
      +  "title": "Frame",
      +  "type": "object"
      +}
    • changedInput schema / $defs / Op_plan_video / properties / op / description
      Previous value: -"API/MCP only. ONE quote for the whole video = every shot (1 to 6) + Kelpo's montage (see montage_credits in the receipt): for a video with several\nshots (the montage is always included, never ask whether they want it), or for a single shot only when the person asked\nfor a montage. Call it last, after the scene is prepared (finish_review first when needed). Nothing is generated or\ncharged here. The receipt data.shoot lists each shot (title, duration, credits), montage_credits and the total `credits`;\nits english_directions_needed lists, per shot, the texts to write SILENTLY in English for the accept call (never shown to\nthe person). Show every line and the total in the person's language, get their explicit yes, and call kelpo_accept_shoot\nONCE (max_credits = total, english per shot). The delivery is the final edited video (kelpo_get_shoot montage_job_id). A\nsingle shot without a requested montage never uses this: kelpo_quote_scene then kelpo_generate."New value: +"API/MCP only. ONE quote for the whole video = every shot (1 to 6) + Kelpo's montage (see montage_credits in the receipt): for a video with several\nshots (the montage is always included, never ask whether they want it), or for a single shot only when the person asked\nfor a montage. Call it last, after the scene is prepared (finish_review first when needed) and only once the person has\nseen the script and approved it in their own words (31/10: never with a « yes » you wrote for them). Nothing is generated or\ncharged here. The receipt data.shoot lists each shot (title, duration, credits), montage_credits and the total `credits`;\nits english_directions_needed lists, per shot, the texts to write SILENTLY in English for the accept call (never shown to\nthe person). Show every line and the total in the person's language, get their explicit yes, and call kelpo_accept_shoot\nONCE (max_credits = total, english per shot). The delivery is the final edited video (kelpo_get_shoot montage_job_id). A\nsingle shot without a requested montage never uses this: kelpo_quote_scene then kelpo_generate."
    • changedInput schema / $defs / Op_shoot_retake / properties / op / description
      Previous value: -"The user asks to redo one or several shots of the delivered multi-shot video (« refais le plan 2 », « refais les mains\nsans le tampon »). Prepares a fix offer limited to these shots, with its price in credits, shown with the two answers\n(do the fixes / it's fine). Nothing is generated or charged here. Quote the price in the reply and ask « on le fait ? »."New value: +"The user asks to redo one or several shots of the delivered multi-shot video (« refais le plan 2 », « refais les mains\nsans le tampon »). Prepares a fix offer limited to these shots, with its price in credits, shown with the two answers\n(do the fixes / it's fine). Nothing is generated or charged here. Quote the price in the reply and ask « on le fait ? ».\nEvery new take is paid, a faulty one too (defect true): Kelpo never redoes or offers a take; an AI take is not always\nright the first time and a shot can take a few tries."
    • changedInput schema / $defs / OperationsInput / properties / message / description
      Previous value: -"The user's actual message for this turn, verbatim."New value: +"The person's own words for this turn, verbatim. Never written by you: if they have not answered yet, ask them and wait."
    • addedInput schema / $defs / ShootRetakeArgs / properties / defect
      Added value: +{
      +  "default": false,
      +  "description": "True when the user complains about a problem in these takes (lips off the voice, wrong or missing words, deformed hands, distorted face, frozen picture, wrong product). The new take is paid like any other (an AI take is not always right the first time). False for a change the user wants.",
      +  "title": "Defect",
      +  "type": "boolean"
      +}
    • addedInput schema / $defs / Shot / properties / angle
      Added value: +{
      +  "default": 1,
      +  "description": "Fixed camera of this shot. Keep 1 for every shot (one locked camera: the whole video looks like the same take) unless the person explicitly asks for other camera angles or two/three fixed cameras for the montage: then number them; the same number = exactly the same camera, and `camera` describes it.",
      +  "maximum": 3,
      +  "minimum": 1,
      +  "title": "Angle",
      +  "type": "integer"
      +}
    • addedInput schema / $defs / backend__engine__ops_shoot___video_args_model___locals___Args / properties / reference_2k
      Added value: +{
      +  "default": false,
      +  "description": "True only when the person chose the 2K reference image (offered as reference_2k_option in the receipt, with its credits): the image every shot starts from (person, outfit, place, framing) is taken from a short silent 2K render, even for a Normal video; its credits are in this quote. Needs 2 shots or more showing the person.",
      +  "title": "Reference 2K",
      +  "type": "boolean"
      +}
    • addedInput schema / properties / body / type
      Added value: +"object"
  7. Changed6 schema fields changed
    • addedInput schema / $defs / EditPanel
      Added value: +{
      +  "additionalProperties": false,
      +  "properties": {
      +    "at_words": {
      +      "description": "Spoken words on which the panel opens (copied exactly, no figures).",
      +      "title": "At Words",
      +      "type": "string"
      +    },
      +    "lines": {
      +      "description": "1 to 6 short lines, in the order they are said.",
      +      "items": {
      +        "$ref": "#/$defs/EditPanelLine"
      +      },
      +      "title": "Lines",
      +      "type": "array"
      +    },
      +    "scale": {
      +      "default": 1,
      +      "description": "Zoom while the face slides aside, 1-1.3 (1 = no zoom, sharpest; default).",
      +      "title": "Scale",
      +      "type": "number"
      +    },
      +    "side": {
      +      "default": "auto",
      +      "description": "Side of the text; the face slides to the other side. auto (default) = close panels share ONE slide (the face moves once) and the side alternates from one slide to the next.",
      +      "enum": [
      +        "auto",
      +        "left",
      +        "right"
      +      ],
      +      "title": "Side",
      +      "type": "string"
      +    },
      +    "title": {
      +      "default": "",
      +      "description": "Yellow title of the panel (max 48), e.g. \"La niche réunionnaise\".",
      +      "title": "Title",
      +      "type": "string"
      +    },
      +    "until_words": {
      +      "default": "",
      +      "description": "Optional spoken words after which it closes. Empty = until the next panel, long enough to read.",
      +      "title": "Until Words",
      +      "type": "string"
      +    }
      +  },
      +  "required": [
      +    "at_words"
      +  ],
      +  "title": "EditPanel",
      +  "type": "object"
      +}
    • addedInput schema / $defs / EditPanelLine
      Added value: +{
      +  "additionalProperties": false,
      +  "properties": {
      +    "at_words": {
      +      "default": "",
      +      "description": "Spoken words on which this line appears, copied exactly as said; avoid words written in figures by the voice recognizer (numbers): take the words just before. Empty = spread over the panel.",
      +      "title": "At Words",
      +      "type": "string"
      +    },
      +    "text": {
      +      "description": "Short line written on screen (max 80), *word* = yellow accent, e.g. \"Abattement de *50 %* sur les bénéfices\".",
      +      "title": "Text",
      +      "type": "string"
      +    }
      +  },
      +  "required": [
      +    "text"
      +  ],
      +  "title": "EditPanelLine",
      +  "type": "object"
      +}
    • addedInput schema / $defs / EditingLaterArgs / properties / panels
      Added value: +{
      +  "description": "« Cours » panels for an explainer / YouTube lesson (face to one side, darkened side, yellow title, lines written on their words); max 8.",
      +  "items": {
      +    "$ref": "#/$defs/EditPanel"
      +  },
      +  "title": "Panels",
      +  "type": "array"
      +}
    • addedInput schema / $defs / EditingPlanArgs / properties / panels
      Added value: +{
      +  "description": "With set_panels: « cours » panels (max 8), for an explainer / YouTube lesson, ONLY on the key facts the viewer must remember (figures, rates, legal conditions, deadlines, eligibility, concrete steps), never on arguments, opinions or the call to action (face only): the face slides to one side, the other side darkens, a yellow title then short lines appear one by one on their spoken words. Subtitles, hooks and zooms are not shown during a panel. Leave face-only moments between panels.",
      +  "items": {
      +    "$ref": "#/$defs/EditPanel"
      +  },
      +  "title": "Panels",
      +  "type": "array"
      +}
    • addedInput schema / $defs / EditingPlanArgs / properties / set_panels
      Added value: +{
      +  "default": false,
      +  "description": "True = replace the « cours » panels by `panels` ([] removes them).",
      +  "title": "Set Panels",
      +  "type": "boolean"
      +}
    • changedInput schema / $defs / Op_plan_video / properties / op / description
      Previous value: -"API/MCP only. ONE quote for the whole video = every shot (1 to 6) + Kelpo's montage (flat fee): for a video with several\nshots (the montage is always included, never ask whether they want it), or for a single shot only when the person asked\nfor a montage. Call it last, after the scene is prepared (finish_review first when needed). Nothing is generated or\ncharged here. The receipt data.shoot lists each shot (title, duration, credits), montage_credits and the total `credits`;\nits english_directions_needed lists, per shot, the texts to write SILENTLY in English for the accept call (never shown to\nthe person). Show every line and the total in the person's language, get their explicit yes, and call kelpo_accept_shoot\nONCE (max_credits = total, english per shot). The delivery is the final edited video (kelpo_get_shoot montage_job_id). A\nsingle shot without a requested montage never uses this: kelpo_quote_scene then kelpo_generate."New value: +"API/MCP only. ONE quote for the whole video = every shot (1 to 6) + Kelpo's montage (see montage_credits in the receipt): for a video with several\nshots (the montage is always included, never ask whether they want it), or for a single shot only when the person asked\nfor a montage. Call it last, after the scene is prepared (finish_review first when needed). Nothing is generated or\ncharged here. The receipt data.shoot lists each shot (title, duration, credits), montage_credits and the total `credits`;\nits english_directions_needed lists, per shot, the texts to write SILENTLY in English for the accept call (never shown to\nthe person). Show every line and the total in the person's language, get their explicit yes, and call kelpo_accept_shoot\nONCE (max_credits = total, english per shot). The delivery is the final edited video (kelpo_get_shoot montage_job_id). A\nsingle shot without a requested montage never uses this: kelpo_quote_scene then kelpo_generate."
  8. First observed

TDQS

A3.7/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations give only readOnly/destructive/idempotent/openWorld flags; the description adds genuinely new behaviour: all-or-nothing atomicity ('All operations succeed together or none is saved'), the receipt contract ('status rejected + code/message explains why'), server-side recording of message/attachments/response, and the no-charge guarantee. The retry instruction ('resend the identical request_id/body after a timeout') adds recovery semantics the annotations cannot express, though it sits in mild tension with idempotentHint=false, which the schema resolves by noting request_id is an optional idempotency key.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loads the atomicity contract, then behaviour, then a compact JSON example that earns its space by showing the exact call shape. The mid-sentence operation enumeration is somewhat listy and ends in 'etc.', but no sentence is pure filler.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a tool whose schema embeds 39 discriminated operation types, the description covers the invocation contract well (atomicity, receipts, retry, no charge) but leaves the operation surface uncovered, delegating entirely to the $defs. No output schema exists, and the receipt shape (beyond 'status rejected + code/message') is only sketched, which leaves an agent guessing on the success path.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Top-level schema coverage is 0% (path.id and body carry no inline descriptions), so the description must compensate, and it does: it gives a concrete literal payload showing path.id, body.version, the verbatim message rule, and the operations array shape with an edit_scene example. It does not explain the `attachments`, `response`, or `request_id` fields, relying on the schema for those, which keeps it below 5.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a specific verb and scope: 'Apply one conversational turn atomically with the same operations and validations as the Kelpo studio', then enumerates operation families (classify uploads, edit the scene, prepare steps, choose actors/outfits/voice, ask_user). This is clearly the batch-write entry point for a conversation turn, distinguishable from the read/quote/generate siblings. It stops short of 5 because the operation list trails off with 'etc.' and the resource is abstract ('operations') rather than a concrete entity.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Usage is implied rather than stated: 'one conversational turn', 'the exchange is recorded by this call', and the negative boundary 'Never generates or charges: generation stays an explicit user approval (quote + generate)' which does route generation work away. There is no guidance on how to choose among the ~39 operations or when a turn should produce no operations, and no directory pointer to the operations catalog sibling.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources