Post

How do y'all avoid having seven+ terminal windows open with agent conversations that are probably over? but maybe you might want to pick back up a few days from now? seems a shame to throw away all that exploration & context, right?

Tilmon Edwards (Dirigo)

@tilmonedwards.com

11 Sept

My starting point for working with agents is that context windows are short-term working memory and disposable. A context window must be able to bootstrap within the first one or two messages in a blank context. Done using a combination of AGENTS md (esp. subdirectories) and repo skills.

Tilmon Edwards (Dirigo)

@tilmonedwards.com

The harnesses I use (polytoken at home and pi at work) allow me to resume a conversation if I really want that specific thread again, but I'd rather exercise the muscle of being able to start a fresh conversation any time I want and have it be productive.

11/9/26, 7:55 pm1 repost4 likes

joel

@jmoonz.bsky.social

12 Sept

I think this is the right approach. Context should be ephemeral. The stored history is documentation, git log, ticket summaries, and the code.