Make prompt capture fast and resilient

Hop-State: A_06FN8D0CK06GY5XBDYZ632R
Hop-Proposal: R_06FN8CY2FJBV5QA0WFEDWHR
Hop-Task: T_06FN8C4PMFJRT8APV4G03Y8
Hop-Attempt: AT_06FN8C4PMFRJEE4B7KKQ740
This commit is contained in:
cyph3rasi
2026-07-11 19:42:05 -07:00
committed by Hop
parent ff62e89596
commit 4364e1d8a4
4 changed files with 54 additions and 15 deletions
+7 -2
View File
@@ -58,8 +58,13 @@ uses it as the default session and identifies the runtime as `codex` unless
headers, and credential-bearing connection strings before persistence.
Read the returned `HOP_STATE_ID`, `HOP_TASK_ID`, `HOP_ATTEMPT_ID`, and workspace.
If capture fails or `hop` is unavailable, stop without project effects and
report the error.
Allow at least 120 seconds for capture. If it times out or fails transiently,
retry once with the same message, agent, and session; Hop session binding makes
the retry idempotent for the task. If the retry fails, inspect Hop's error and
locks, run `hop doctor`, and repair a safe local operational problem when
possible. Stop without project effects only when Hop remains unavailable or
unsafe after recovery; do not abandon the user's request after the first
timeout.
If Hop reports redactions, never repeat the credential in output, summaries,
commands recorded as evidence, or proposal text. Refer to its environment