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
Copy file name to clipboardExpand all lines: MEMORY.md
+1Lines changed: 1 addition & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -10,6 +10,7 @@
10
10
- ty/ruff LSP servers may request `workspace/configuration`; respond with per-item empty config objects or diagnostic requests can hang.
11
11
- pi-statusline is display-only; avoid prompt interception or customization commands unless intentionally reintroduced.
12
12
- In Pi extensions, do not call action methods such as `getThinkingLevel()` during the factory load; defer them to `session_start` or later handlers.
13
+
- Symptom: extension accept/execute actions from `agent_end` may not trigger a new turn. Cause: `pi.sendMessage({ triggerTurn: true })` only triggers when idle. Fix: use `pi.sendUserMessage(..., { deliverAs: "followUp" })` when `ctx.isIdle()` is false.
13
14
- Extension statusline entries should be activity-based: only show an extension in status when it is actively running, retrying, or needs attention; avoid permanent “configured/ready/on” statuses.
14
15
- Codex usage can be queried without Codex CLI by sending Pi's `openai-codex` bearer token to `https://chatgpt.com/backend-api/wham/usage`; response uses Codex `RateLimitStatusPayload` snake_case fields.
15
16
-`pi-codex-usage` statusline must select a rate-limit bucket by current model id/name; `gpt-5.3-codex-spark` can use its own returned bucket instead of primary `codex`.
0 commit comments