conversation-to-post

Conversation To Post

Use this skill when a user asks to publish, summarize, or document an AI conversation as a dev.io post.

Inputs

• Conversation transcript or the current task result.
• Optional title, author, source, topic, and tags.

Procedure

1. Identify the useful result, decisions, implementation details, and verification evidence.
2. Remove secrets, tokens, credentials, personal data, and irrelevant transcript noise.
3. Write a specific title and a factual summary.
4. Put longer context in conversation; keep the summary readable without it.
5. Use publish_post with explicit title and summary.
6. Confirm the returned file path and read the generated file with read_post if validation is needed.

Post rules

• Use source: ai-conversation unless the user provides a more accurate source.
• Use a topic that describes the result, not the model brand.
• Prefer 2-5 lowercase, stable tags such as mcp, typescript, codex, or claude.
• Never include credentials or unreviewed private conversation content.
• Publishing writes to the Kubernetes-backed Markdown volume and optionally publishes to DEV.to when DEV_TO_PUBLISH=true is configured and the verified DEV.to/Forem adapter reports success.

Example tool intent

publish_post({
  title: "Add MCP publishing workflow",
  summary: "Implemented and verified a Kubernetes-hosted Streamable HTTP MCP server that writes AI conversation results as Markdown posts.",
  conversation: "Relevant decisions and test evidence...",
  source: "ai-conversation",
  topic: "mcp-publishing",
  tags: ["mcp", "typescript", "claude", "codex"]
})