You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
PR #1 changed eds wf job logs <id> from an SSE stream to a paginated call
with new --limit/--before flags. The PR's follow-up commit removed the
now-false "(logs streams to stdout)" note from the command table in skill/SKILL.md, but didn't add anything describing the new flags or
paginated semantics.
Problem
skill/SKILL.md is the CLI's documented contract for AI agents driving it
(per CLAUDE.md: "agents rely on it verbatim"). An agent reading it currently
sees eds wf job logs <id> with no mention of --limit/--before or that
output is capped at --limit lines (default 10) per call — it has no way
to know it needs to paginate to see a full log.
Ask
Add --limit/--before to the eds wf job logs row (or a short note
below the table) describing how to page back through a job's log, once the
live-tail question (#2) is resolved — the doc fix should describe whatever
the final behavior ends up being, not just the current interim call shape.
Activity