search-agent-conversation

Search past agent conversations by running a Python script that queries dumped agent data, chats, and prompt history. It retrieves specific interactions using filters for time range, agent type, and session ID.

Yes, filtering agent chat logs by time range and specific session ID is supported. The search query can also target specific agent types to narrow down the retrieved conversation data.

Yes, you need the prerequisite skill 'dump-agent-data-for-rg' to prepare the detailed technical data. This initial data dump is required before the search script can retrieve and filter agent history.

Retrieve prompt-only results from agent history by applying the prompt-only search option. You can also choose to include reasoning data if needed, giving you control over the output of your search query.

Querying a codex agent's session is best done by filtering the agent data by specific session ID and agent type. This narrows the search to return summaries of actions and discussions from that exact session.

A key limitation is the dependency on the prerequisite skill to first dump the agent data. Without completing the data dump beforehand, the search script cannot access or filter the historical agent conversations.