What This Looks Like
The agent keeps revising, planning, checking, branching, or reconsidering instead of returning a final answer. It may repeatedly say more work is needed, open new paths, ask for another pass, or keep refining the same output without reaching a stable completion point.
Why It Matters
Agent work needs convergence. If the agent cannot settle, the user cannot know when the task is done or which output should be trusted. This can turn a bounded request into an open loop of partial work, repeated review, and unfinished decisions.
Structural Signal
The workflow lacks a stable finalization point. The issue is not that the agent needs one reasonable correction; it is that the agent’s process does not converge on a final answer under the declared task constraints.
Common Triggers
- The task has no declared done condition
- The agent is allowed to keep improving without a stopping rule
- Review or self-check loops create new work each pass
- The agent treats uncertainty as permission to continue indefinitely
- Tool results or critiques create new branches instead of closing the task
- The workflow does not distinguish final output from draft or analysis state
When to Use This Issue
Use this Issue when the agent’s main failure is inability to settle on a final answer, final artifact, or completed state.
When Not to Use This Issue
Do not use this Issue when the agent expands the task but still completes a final output. Do not use it for one bad answer or one ordinary revision cycle. This Issue applies when the agent cannot converge.