agentleFS
VideoTranscriberAIVideoTranscriberAI/videotranscriber-skills/skills/setup-videotranscriber/SKILL.md

setup-videotranscriber

Skill14 starsChanged 4 months agoMITView on GitHub

Connect Video Transcriber AI MCP, check a connection, or recover from missing tools and API Key errors. Use for setup in local Codex, Claude Code, WorkBuddy, Cursor, Windsurf, Gemini CLI, Copilot, VS Code, Cline, or Roo Code. Does not require an existing MCP connection.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read only the matching section of client setup.

    line 13
  2. Distinctive:

    When a credential is missing, give one concrete next action: “请登录官网,在 API 控制台获取 Key,然后粘贴到我打开的本地配置窗口,点击连接。我会继续处理刚才的任务。” Link Video …

    line 14

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Connect Video Transcriber

Help the user reach a working connection with the fewest manual steps. Reply in their language, describe the current state simply, and give one actionable next step at a time.

Connect

  1. Identify the active client from the host context and discover any existing Video Transcriber connection. Reuse a healthy connection. Configure the active client only; other installed apps are not evidence of which client is running. Ask which client they use only when the host cannot be identified. Preserve the intended account and environment; clarify an ambiguous account before proceeding.
  2. Read only the matching section of client setup. The endpoint is https://videotranscriber.ai/mcp; authentication is a Bearer API Key. Installing a Skill alone does not supply the credential. This service has no OAuth login flow.
  3. When a credential is missing, give one concrete next action: “请登录官网,在 API 控制台获取 Key,然后粘贴到我打开的本地配置窗口,点击连接。我会继续处理刚才的任务。” Link Video Transcriber AI. Explain once that transcription uses independent API Quota; website plans do not cover it. Do not initiate a purchase.
  4. For a supported local client, follow local setup and launch the packaged helper with that client's --client value. Resolve known host network/user-config permission requirements before collecting a Key. It opens a hidden local input, validates the Key with bounded in-process retries, and writes that client's user-level connection. Prefer this one-paste path over instructions to set environment variables or edit files. A supported native credential dialog is also suitable. The Key stays out of chat, tool arguments, command history, and repository files. Do not print saved configuration; use the helper's safe status. For config_selection_required, locate the active user config through that client's MCP settings and rerun with --config; retain the user's task. Hosted clients need their own credential UI.
  5. After credential_verified, inspect fresh host tool discovery/status. Prefer native create_transcription, get_transcription, and get_transcription_result tools when available. If a real host reload API is available, it may load them; do not attempt to control Codex's own window through Computer Use. When native tools remain unavailable, use the packaged same-session MCP caller with the saved Key. It calls the same official service without restarting the client. Continue the user's original request through transcribe-media; do not stop at a restart instruction or write an ad hoc HTTP client. client_connection_verified: false concerns the native host only, not the authenticated service connection.
  6. Explain the actual state: authenticated service ready, and whether native tools or the packaged caller will be used. Invite a public URL only if the user has not already provided one. Setup alone never authorizes a paid submission. If neither native tools nor the packaged caller can run, report the specific blocker and one concrete remedy; manual refresh is a final fallback for that situation.

Recover

While retrying_verification or retrying_mcp_call is reported, wait for the same process to finish; keep the task and do not open a second Key window. It retries transient failures at most twice using the Key held in memory. The helper prefers installed curl and otherwise uses Python, with complete-response checks. For permission_required, resolve the specific execution/network/config permission before collecting another Key. On exhausted retries, explain the failed operation and safe error; investigate connectivity before restarting credential input. A failed network/response check does not establish that the Key is invalid. Reuse an already saved valid Key.

For invalid_key or key_revoked, use the same helper with --replace or the client's native credential dialog. For account_disabled, direct the user to their API account settings or support. For access_blocked, explain that the gateway/access policy refused the connection; preserve the configuration and do not ask for another Key as a remedy. HTTP 401/403 alone does not establish that a Key is invalid. For insufficient quota, explain the separate API balance and link the official site; do not change options, buy quota, or repeatedly submit. Cancellation ends setup without changes. If private input is unavailable, give the concrete terminal fallback from local setup rather than only a documentation link. Describe the observed user-facing state and one next action.

Analyze a user-supplied transcript or subtitle directly without API credentials. If the user asks for a demo and has supplied no transcript, use the text-only first-run demo.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.