Podcast Workflow

Full podcast post-production pipeline. Give it a transcript or audio URL and it produces show notes, chapter markers, social media clips, SEO title/descripti...

Security Scan

The skill's requirements and runtime instructions are coherent with a podcast post‑production pipeline and do not request unrelated credentials or risky installs.

Purpose & Capability

Name/description (podcast post‑production) match the runtime instructions: ingest transcript or URL, generate show notes, chapters, social posts, SEO metadata, and a newsletter blurb. The only external requirement is curl, which is reasonable for fetching a provided URL.

Instruction Scope

Instructions remain within the stated purpose. One operational gap: the skill tells the agent to 'use curl to fetch the page and extract any available transcript or description' but gives no concrete method for extracting transcripts from services like YouTube or Spotify (may require APIs or more advanced scraping). Also, fetching arbitrary user-provided URLs means the agent will make outbound HTTP requests (expected for this use case but worth noting for privacy/SSRF/logging concerns).

Install Mechanism

Instruction-only skill with no install spec and no code files — lowest-risk installation footprint. README mentions a 'clawhub install' command, but that is documentation-only; there is no automated install step in the package metadata.

Credentials

No environment variables, credentials, or config paths are requested. This is proportionate to the stated purpose (processing transcripts / fetching public URLs).

Persistence & Privilege

always is false and there are no special privileges requested. The skill can be invoked autonomously by the agent (platform default), which is expected behavior for a user-invocable workflow skill and is not combined with other red flags.

Assessment

This skill appears internally consistent and doesn't ask for secrets or installs. Before installing or using it, consider: 1) Do not paste sensitive or private transcript content you wouldn't want processed by the agent. 2) If you supply a URL, the agent will make outbound HTTP requests (those requests will be visible to the remote host in server logs and may be subject to network policies); avoid giving private/internal URLs to prevent SSRF/exposure. 3) Extraction from YouTube/Spotify may be unreliable using only curl — expect imperfect transcript retrieval for some URLs. 4) The README mentions a clawhub install command but the registry entry has no install spec; this is a documentation inconsistency rather than a security problem. If any of these concerns matter for your environment, test the skill with non-sensitive content first or request additional implementation details from the skill author.

When to use this skill

Use this skill when the user:

• Pastes a podcast transcript or provides an audio/video URL
• Asks for show notes, chapter markers, or timestamps
• Wants social media posts from a podcast episode
• Needs a newsletter summary of an episode
• Says anything like "process my podcast", "episode notes", "clip this"

Do NOT use this skill for general writing tasks unrelated to podcasts.

What you produce

Run all steps below in sequence. Deliver everything in a single structured response.

---

Step 1 — Ingest content

If the user pastes a transcript:

• Use the text directly. Note approximate word count.
• Estimate total episode duration: assume ~130 words per minute of spoken audio.

If the user provides a URL (YouTube, Spotify, RSS, etc.):

• Use curl to fetch the page and extract any available transcript or description.
• If no transcript is available, tell the user: "No transcript found at this URL. Please paste the transcript text directly."

If the user provides neither:

• Ask: "Please paste your episode transcript or share a URL to the episode."

---

Step 2 — Show notes

Write professional show notes in this exact format:

## [Episode Title — infer from content if not given]

[2–3 sentence hook that captures the core insight of the episode. Write for someone who hasn't listened yet.]

### What we cover
- [Key topic 1]
- [Key topic 2]
- [Key topic 3]
- [Key topic 4]
- [Key topic 5]

### Key takeaways
[3–5 bullet points with the most actionable or surprising insights from the episode]

### Guest / Host
[Name and one-line description if mentioned in transcript. Skip if solo episode with no guest info.]

### Resources mentioned
[List any books, tools, websites, or people mentioned in the transcript. Format: Name — brief description. Skip section if none found.]

Rules:

• Write in present tense ("In this episode, [Host] explains...")
• No filler phrases like "fascinating", "incredible", "game-changing"
• Keep the hook under 60 words
• Total show notes: 200–400 words

---

Step 3 — Chapter markers

Generate timestamp-based chapters suitable for YouTube, Spotify, and podcast players.

Format each line as:

00:00 Introduction
[MM:SS] [Chapter title — max 50 characters]

Rules:

• Estimate timestamps based on content position in transcript (use the ~130 wpm estimate)
• Minimum 4 chapters, maximum 12
• Chapter titles must describe the content, not just label it (e.g. "Why cold outreach fails" not "Topic 2")
• First chapter always starts at 00:00
• Last chapter should be a clear conclusion/outro

---

Step 4 — Social media clips

Write 3 social posts, each optimized for a different platform:

LinkedIn post:

• 150–250 words
• Professional tone, insight-first
• End with a reflective question to drive comments
• No hashtags in body — add 3 relevant ones at the end on their own line

X (Twitter) thread:

• 5–7 tweets
• Tweet 1: hook (under 280 chars, must stand alone)
• Tweets 2–6: one insight per tweet, concrete and specific
• Final tweet: call to action (link to episode or subscribe)
• Number each tweet: 1/ 2/ etc.

Instagram caption:

• 100–180 words
• Conversational, personal tone
• 5–8 hashtags at the end

---

Step 5 — SEO metadata

Provide:

Title (under 60 chars): [title]
Meta description (under 155 chars): [description]
Primary keyword: [keyword]
Secondary keywords: [keyword1], [keyword2], [keyword3]

Rules:

• Title must include the primary keyword naturally
• Meta description must include a benefit or outcome
• Keywords should reflect what someone would search to find this episode

---

Step 6 — Newsletter summary

Write a 100–150 word newsletter blurb suitable for pasting into a weekly digest:

**[Episode title]**

[2-sentence summary of what the episode covers]

[1 key insight or quote from the episode]

[1-sentence reason why subscribers should listen]

🎧 [Listen / Watch] → [leave placeholder: INSERT_LINK]

---

Output format

Deliver all 5 outputs in order, each under a clear H2 heading:

1. ## Show notes
2. ## Chapter markers
3. ## Social media
4. ## SEO metadata
5. ## Newsletter

Separate each section with a horizontal rule (---).

After all sections, add:

---
✅ Podcast workflow complete. All assets ready to publish.
Want me to adjust tone, length, or regenerate any section?

---

Error handling

• Transcript too short (under 500 words): Complete all steps but note: "This transcript is short — outputs may be less detailed than usual."
• Non-English transcript detected: Process in the source language, then add a note offering to translate all outputs to English.
• No clear topic or guest: Infer from context. Do not ask clarifying questions unless transcript is completely ambiguous.
• Explicit or adult content: Decline and explain this skill is for general-audience podcasts only.

---

Version history

• 1.0.0 — Initial release. Full pipeline: show notes, chapters, social (LinkedIn/X/Instagram), SEO metadata, newsletter.