Conversation Summary

Generate summaries for conversation content with incremental update support.

Security Scan

The skill implements conversation summarization by sending conversation data to a hard-coded external API (iautomark.sdm.qq.com); that is coherent with a summary service but the hard-coded remote endpoint, absence of explicit consent/warning about data exfiltration, and packaging/metadata inconsistencies are concerning.

Purpose & Capability

The stated purpose (generate conversation summaries) matches the included Python script which POSTs chat content to a remote summary API. However there are inconsistencies: the SKILL.md declares runtime requirements (python3, pip and the Python 'requests' package) while the registry metadata lists no required binaries/env; package.json (an npm manifest) lists a 'requests' dependency (which is a Python package here) — this mismatch between packaging and runtime is incoherent and suggests sloppy or incorrect metadata.

Instruction Scope

The SKILL.md directs the agent to run the included script with the entire conversation (chat_list JSON). The script will transmit the full conversation content (and optional previous summary) to an external service without any mention in SKILL.md of privacy implications or where data is sent. The instructions do not read or transmit other local files or env vars, but they do enable exfiltration of conversation content to a third-party endpoint.

Install Mechanism

There is no install spec (instruction-only skill) which is low-risk. A package.json is present with a postinstall chmod command and an npm-style dependency on 'requests' — odd because the runtime is Python and the real dependency is the Python requests package. The presence of package.json is unexpected but the repo does not contain a remote download/extract step. This is likely a packaging oversight but is inconsistent.

Credentials

The skill requests no environment variables or credentials, which superficially seems minimal. However, it sends unencrypted conversation data to a hard-coded external API (https://iautomark.sdm.qq.com/assistant-analyse/v1/assistant/poc/summary/trigger) without authentication or an explicit consent mechanism — that may be disproportionate if users intend to summarize sensitive content. No justification or privacy notice is provided.

Persistence & Privilege

The skill does not request persistent presence (always: false), does not modify other skills or system config, and does not require elevated privileges. It is only executable by explicit invocation (user-invocable).

What to consider before installing

This skill will send the conversation text you pass to it to an external service at iautomark.sdm.qq.com to produce the summary. If you will be summarizing sensitive or private conversations, do not use this skill unless you trust that external endpoint and understand its data handling policies. Note the repository metadata is inconsistent (package.json lists an npm-style dependency 'requests' while the runtime is Python and expects the Python requests package) — this is likely a packaging mistake but indicates the author may not have polished the package. Before installing: (1) confirm you are comfortable with sending conversation content to the listed remote URL and verify the provider/policy, (2) avoid sending secrets or sensitive PII to the skill, (3) if you prefer local processing, request or implement a local summarizer instead. If you need greater assurance, ask the author for a privacy statement and a way to run the summarization locally or on a vetted endpoint.

Conversation Summary - Agent Instructions

Use this skill to generate summaries for conversation content.

Usage

When the user requests any of the following:

• "Summarize this conversation"
• "Generate a summary"
• "What did we talk about"

Use the summarize_conversation tool to call the summary API.

How to Call

python3 scripts/conversation_summary.py '<chat_list_json>' '<history_summary>'

Parameters

Parameter	Type	Required	Description
chat_list	string	Yes	JSON formatted conversation content
history_summary	string	No	Previous summary for incremental update

chat_list Format Example

[
  {"role": "user", "content": "How is the weather today?"},
  {"role": "assistant", "content": "It is sunny, 25 degrees."}
]

Response

The script returns JSON with:

• status: "completed" or "error"
• summary: Generated conversation summary
• error: Error message if failed

Error Handling

• If the API returns a non-zero code, report the error message to the user
• If the request fails, check network connectivity
• Ensure chat_list is valid JSON format before calling