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
{{ message }}
Repository navigation
qwen3:8b says it clicked a button when it never called the click tool, or after the click was declined #402
Seen while testing #382 (browser sign-up labels) on qwen3:8b over WS:
On a page with a monthly subscription, qwen replied that the button "was clicked" without calling browser_click at all.
After the user denied a click, qwen replied it "was clicked, but the action was declined". The click never happened.
claude-code/sonnet reported both cases correctly.
#165 / #380 check the outcome of actions that ran. This is the other side: the reply claims an action that never ran, or one that was denied.
Fix idea: after a turn, compare the reply's claims about actions (clicked, sent, saved, booked, deleted) with the tool calls that actually succeeded in that turn. If a claimed action has no successful call:
correct the reply in plain words, for example "I didn't click it: you declined." or "I haven't clicked anything yet.";
or ask the model to rewrite it once.
A small-model-friendly prompt line that says declined or missing actions must be reported as not done may help too. Keep it cheap: no extra model call when the reply makes no action claims.
Done when: on qwen3:8b, a declined click and a never-made click are both reported as not done. Run the real test from #382's local pages over WS, and check sonnet still reads naturally.
I would like to take this on. I will focus on reconciling a turn's stated action outcome with its actual browser tool-call result, add focused offline regression coverage for declined and missing clicks, and keep the change small.
Activity
syedmustafan commentedon Oct 16, 2026
I would like to take this on. I will focus on reconciling a turn's stated action outcome with its actual browser tool-call result, add focused offline regression coverage for declined and missing clicks, and keep the change small.
chore: use changelog fragment for existence-master#402