Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault

No arguments

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

Clips Kitty turns long videos into short vertical clips, entirely on this computer. Queue work with queue_video or queue_local_file, follow it with job_status, then read the results with list_clips and export_clip. Processing a long stream takes tens of minutes, so never block on it: queue, then check back.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

CapabilityDetails
tools
{
  "listChanged": false
}

Tools

Functions exposed to the LLM to take actions

NameDescription
youtube_statusA

Whether Clips Kitty can publish to YouTube right now, which channel, and how many uploads are left today. Check this before planning a batch.

woopsocial_statusA

Whether Clips Kitty can post to social platforms through the person's WoopSocial account. Check this before offering to schedule anything.

schedule_clips_planA

Work out what posting a set of clips would do: which clips, which platforms, and the exact time each one would go out. Creates nothing and posts nothing. Give video_id to take every rendered clip from a video, or clip_ids for specific ones. Spacing: per_day for a daily budget ("5 a day" is per_day 5, the default when no spacing is given, because WoopSocial allows about 5 YouTube posts a day), or every_hours for a flat gap (24 is one a day). Always show the result and get a clear yes — the person confirms in the app.

schedule_clips_executeB

Actually schedule the posts worked out by schedule_clips_plan. This posts publicly and cannot be undone.

uploadpost_statusA

Whether Clips Kitty can publish to several platforms at once through the user's Upload-Post account, and which profile. Check this before offering to publish anywhere other than YouTube.

uploadpost_publishB

Send one clip to several platforms in a single upload: YouTube, TikTok, Instagram, Facebook, X, Threads, LinkedIn, Pinterest, Bluesky. The same title and description go everywhere unless the person asks otherwise. Uploads cannot be taken back, so get a clear yes first.

uploadpost_resultB

Per-platform state and links for a clip published through Upload-Post: which platforms went out, which failed and why.

publish_planA

Work out what publishing these clips would do: final titles, descriptions and publish times. Creates nothing. Give start_at (with a timezone offset) and every_hours to space them out, or neither to upload as soon as each is ready. ALWAYS show the plan and get a yes before executing it.

publish_plan_executeA

Upload the clips in a plan, one job each, so one failure does not stop the rest. Only call this after the person has agreed to the plan. Uploads cannot be undone and each one spends their daily quota.

publish_statusB

Where a clip's upload has got to, or the recent uploads when no clip is named.

queue_videoA

Hand Clips Kitty a YouTube, Twitch or Kick link and it finds the moments worth posting, crops them to vertical and captions them, all on this computer. Returns a job id; processing a long stream takes a while.

queue_local_fileB

Same pipeline for a video already on disk. Nothing is uploaded. Set channel when you know it: creator learning keys off it.

job_statusB

Where a job has got to: queued, running, done, failed or cancelled.

queue_statusA

What the queue is doing, how much room is left, and whether it is paused. Check this before reporting that nothing is happening.

list_videosB

Videos Clips Kitty has processed, newest first, with clip counts.

list_clipsB

The clips from one video: id, score, timestamps and title. Scores are the model's ranking, 0-100.

clip_captionsB

The transcript of one clip, as burned-in caption lines with times.

export_clipC

Copy a finished clip out to a folder, with its final filename.

engine_statusB

Whether the engine is up, and which version it is.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

TDQS

B3.1/5.0

Scored across 19 tools

Disambiguation2/5

There are two near-identical plan/execute pairs (schedule_clips_plan/execute and publish_plan/publish_plan_execute) plus uploadpost_publish, and the descriptions do not make it obvious which publishing path an agent should pick. The status family is also crowded: youtube_status, woopsocial_status, uploadpost_status, publish_status, uploadpost_result, job_status, queue_status and engine_status all answer 'what state is X in', and distinguishing publish_status from uploadpost_result or job_status from queue_status requires careful reading.

Naming Consistency3/5

There is a recognizable <noun>_status pattern for the seven state tools, but verbs are ordered inconsistently elsewhere: schedule_clips_plan and uploadpost_publish are verb-first, while publish_plan and clip_captions are noun-first. Multi-word compounds pile up (publish_plan_execute vs schedule_clips_execute) so the naming is readable but not predictable.

Tool Count3/5

19 tools for a pipeline-plus-publishing server is on the heavy side, and much of the weight comes from supporting three separate publishing backends (YouTube direct, WoopSocial, Upload-Post), each with its own status/plan/execute/result tools. It is defensible for the domain but feels borderline bloated.

Completeness3/5

The core lifecycle is covered: ingest (queue_video, queue_local_file), monitor (job_status, queue_status), inspect (list_videos, list_clips, clip_captions), export, and multi-platform publishing with plan/execute. However, there is no cancel or delete operation for jobs, scheduled posts, or clips, and no way to revise a plan once created, which are real dead ends for an agent managing a queue.

Maintenance

ActivityActive
ResponsivenessSyncing