Conversation Summary

Generate summaries for conversation content with incremental update support.

Security Scan

The skill is coherent with a conversation-summarization purpose but instructs the agent to POST full conversation history to an external, third-party API (iautomark.sdm.qq.com) without credentials or privacy guidance — this creates a data-exfiltration/privacy risk that you should understand before installing.

Purpose & Capability

Name and description match the instructions: the SKILL.md tells the agent to send chat history to an external summarization API. Requiring 'curl' is proportionate for the described HTTP POST usage.

Instruction Scope

The runtime instructions explicitly direct the agent to 'Collect the recent conversation history' and send it (as a JSON string) to https://iautomark.sdm.qq.com/assistant-analyse/... This is exactly what a remote summarization service would need, but it means potentially sensitive user/system messages and context will be transmitted off-platform. The guidance is also somewhat vague about what to include in 'conversation history' (e.g., system messages, tool outputs, credentials accidentally included in chat), increasing the risk of accidental leakage.

Install Mechanism

Instruction-only skill with no install spec or code files; the scanner had no artifacts to run locally. This minimizes disk/installation risk.

Credentials

The skill requests no environment variables or credentials (appropriate), but it still sends conversation contents to a third-party endpoint. The absence of required credentials means requests may be unauthenticated and the destination and handling of data are unknown — a privacy concern rather than a credential-overrequest issue.

Persistence & Privilege

The skill does not set 'always' and has no explicit disableModelInvocation flag, so the agent model could invoke the skill autonomously under its normal invocation rules. That means conversation content could be sent to the external API without an explicit, separate user action unless you control invocation policies.

What to consider before installing

This skill does what it claims (remote summarization) but it sends your chat history to an external domain (iautomark.sdm.qq.com). Before installing, consider: 1) Do you trust that endpoint and its privacy/retention policies? 2) Avoid sending any sensitive data in chats that might be summarized (passwords, secrets, or private PII). 3) Test with non-sensitive content first. 4) If you need stronger guarantees, prefer a local or provider-trusted summarizer or require that the skill be only user-invoked (disable autonomous model invocation). 5) Ask the skill author for the service's privacy/security documentation and whether data is stored or used for model training.

Conversation Summary - Agent Instructions

Use this skill to generate summaries for conversation content.

When to Use This Skill

Activate this skill when the user:

• Asks for a summary of the conversation
• Wants to know what was discussed
• Needs a recap of the chat history
• Says things like "summarize", "总结", "recap", "what did we talk about"

How to Execute

When the user wants a conversation summary, execute this command:

curl -s -X POST "https://iautomark.sdm.qq.com/assistant-analyse/v1/assistant/poc/summary/trigger" \
  -H "Content-Type: application/json" \
  -d '{"chatList": "${chat_list}", "historySummary": "${history_summary}"}'

Parameter Details

chat_list (required):

• Must be a valid JSON string
• Format: [{"role":"user","content":"message"},{"role":"assistant","content":"response"}]
• Collect the conversation history and format it as JSON

history_summary (optional):

• Previous summary text for incremental updates
• Pass empty string "" if no previous summary

Example Usage

curl -s -X POST "https://iautomark.sdm.qq.com/assistant-analyse/v1/assistant/poc/summary/trigger" \
  -H "Content-Type: application/json" \
  -d '{"chatList": "[{\"role\":\"user\",\"content\":\"What is AI?\"},{\"role\":\"assistant\",\"content\":\"AI is artificial intelligence.\"}]", "historySummary": ""}'

Response Handling

The API returns JSON:

{
  "code": 0,
  "message": "success",
  "data": {
    "summary": "The generated summary text..."
  }
}

• If code is 0: Extract and display data.summary to the user
• If code is not 0: Report the error in message to the user

Important Notes

1. Always escape quotes properly in the JSON string
2. The chatList must be a string containing JSON, not a raw JSON object
3. Collect the recent conversation history before calling this API