Skip to main content
Agent skill

Plaud

Read Plaud recordings — list and search recordings, read transcripts, AI summaries and highlights, and get a temporary audio URL — through a Plaud account the user's team has connected to Nuphos via OAuth. Use whenever the user asks about a meeting or call they recorded with Plaud (Plaud Note, NotePin), wants a summary, action items, a follow-up draft or quotes from a recording, or asks what was said in a conversation. Nuphos backend hands the agent a short-lived Plaud access token; the agent calls the Plaud API directly with curl.

Get started

Get the complete skill folder, including its scripts and reference files.

When to use this skill

Use Plaud when an AI agent needs a reusable SKILL.md workflow for this job: Read Plaud recordings — list and search recordings, read transcripts, AI summaries and highlights, and get a temporary audio URL — through a Plaud account the user's team has connected to Nuphos via OAuth. Use whenever the user asks about a meeting or call they recorded with Plaud (Plaud Note, NotePin), wants a summary, action items, a follow-up draft or quotes from a recording, or asks what was said in a conversation. Nuphos backend hands the agent a short-lived Plaud access token; the agent calls the Plaud API directly with curl.

When not to use it

Skip Plaud when the task is outside the coding category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.

How to install

Where will you use it?

Install scope
Claude Code folder setup
mkdir -p ~/.claude/skills/plaud
# Save SKILL.md to ~/.claude/skills/plaud/SKILL.md
  1. Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/plaud/SKILL.md. For a shared project, commit the folder at .claude/skills/plaud/ instead.
  2. Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
  3. For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.

Full install guide for Claude, Cursor, and Codex →

Try a starter prompt

After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.

Starter prompt
Use the Plaud skill for this task: Use Plaud when this documented workflow matches the task. Confirm any required inputs or files before making changes.

What supports this recommendation?

70/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.

Documentation
77/100
Practical detail
68/100
Source evidence
63/100
Source signals
70/100
Model compatibility evidence
  • claudeuntested

    No model-specific signal or recorded compatibility test was found.

  • chatgptuntested

    No model-specific signal or recorded compatibility test was found.

  • geminiuntested

    No model-specific signal or recorded compatibility test was found.

  • copilotuntested

    No model-specific signal or recorded compatibility test was found.

  • llamauntested

    No model-specific signal or recorded compatibility test was found.

  • perplexityuntested

    No model-specific signal or recorded compatibility test was found.

  • mistraluntested

    No model-specific signal or recorded compatibility test was found.

  • grokuntested

    No model-specific signal or recorded compatibility test was found.

Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.

What this skill does

# plaud

## Session isolation

Keep CLI credentials and settings inside the current session's `HOME` (`NUPHOS_SESSION_HOME`) and respect the supplied CLI config environment variables. Do not copy another session's or the runtime owner's credentials. Without `NUPHOS_SESSION_HOME`, CLI defaults may use shared runtime configuration; be aware of the affected scope.

Changing runtime-global settings is possible, but strongly discouraged unless the user understands the impact on other sessions and explicitly requests it. Explain the shared scope first; do not unset session isolation variables, write to the runtime owner's home, use a shared OS credential store, or modify shared shell startup files as routine setup. This is configuration isolation, not an OS security boundary.

Plaud recordings are private conversations. Read only the recordings the request needs, quote sparingly, and never copy transcripts into places the user did not ask for. The access is read-only.

## Setup

The session's credential context lists each enabled Plaud account with an exact `setupCommand`. Run it once per conversation (and again on a 401):

```bash bash skills/plaud/scripts/setup-credentials.sh <teamId> <bindingId> && source ~/.plaud/nuphos.env ```

It writes `PLAUD_ACCESS_TOKEN`, `PLAUD_API_BASE_URL` and `PLAUD_MCP_URL` to a private env file. Never print the token. Every shell call is a fresh process, so source the env file in each call that uses it.

If no Plaud account is listed, ask the user to enable one for this session, or to have an administrator connect Plaud under Integrations.

A 403 from the setup script means the account is not selected for this session or the user is not on its access list; a 404 means it was disconnected.

## API

All calls are `GET` with `Authorization: Bearer $PLAUD_ACCESS_TOKEN` against `$PLAUD_API_BASE_URL` (`https://platform.plaud.ai/developer/api`).

```bash source ~/.plaud/nuphos.env plaud() { curl -sS --fail-with-body -H @<(printf 'Authorization: Bearer %s\n' "$PLAUD_ACCESS_TOKEN") -H 'Accept: application/json' "$PLAUD_API_BASE_URL$1"; }

plaud /open/third-party/users/current # the connected account plaud '/open/third-party/files/?page=1&page_size=20' # newest recordings first plaud "/open/third-party/files/<fileId>" # one recording, with content blocks ```

`files/` also takes `start_date` and `end_date` as epoch milliseconds (inclusive) to filter by recording time. Page with `page` until a page comes back short. To find a recording by name, list a date range and match `name` locally; there is no server-side text search.

A recording carries `id`, `name`, `created_at`, `start_at`, `duration` (seconds), `presigned_url` (a temporary audio URL, when available), `source_list` and `note_list`:

| Where | `data_type` | What it is | | ----------- | -------------------- | ---------------------------------------------- | | source_list | `transaction` | Transcript segments | | source_list | `transaction_polish` | Cleaned-up transcript | | source_list | `outline` | Topic outline with timestamps | | source_list | `mark_memo` | Moments flagged with the device's highlight key | | note_list | `auto_sum_note` | The AI summary (markdown) |

Each block holds its content inline in `data_content`, or behind a short-lived signed URL in `data_link` — fetch that URL with plain `curl` and **no** Authorization header. Transcript content is a JSON array of segments with `start_time` / `end_time` (milliseconds), `speaker` and `content`:

```bash source ~/.plaud/nuphos.env plaud "/open/third-party/files/$FILE_ID" | python3 -c ' import json, sys, urllib.request file = json.load(sys.stdin) file = file.get("data", file) block = next((b for b in file.get("source_list") or [] if b.get("data_type") == "transaction"), None) if not block: raise SystemExit("No transcript yet; available: %s" % [b.get("data_type") for b in file.get("source_list") or []]) raw = block.get("data_content") or urllib.request.urlopen(block["data_link"]).read().decode() for seg in json.loads(raw): ms = seg.get("start_time") or 0 print("[%02d:%02d] %s: %s" % (ms // 60000, ms // 1000 % 60, seg.get("speaker") or "?", seg.get("content", ""))) ' ```

Prefer the `auto_sum_note` summary when the user only needs the gist; read the transcript for quotes, action items or details the summary may drop. A new recording may not have a transcript or summary until Plaud finishes processing it.

## Errors

- **401**: the token expired; rerun the setup command. If it persists, the user revoked access in Plaud and an administrator must reconnect Plaud. - **Empty `source_list`**: the recording has not been transcribed yet.

Intended uses

  • Use Plaud when this documented workflow matches the task.

Implementation notes

  • Review the source instructions and adapt inputs before running the workflow.

Try it, then tell us how it went

A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.