Enhance context filtering and response handling with refactors - #3
Merged
Merged
Conversation
Previous hardcoded list missed <local-command-stdout> (seen in actual user sessions) and would miss any future tag Claude Code adds. Switch to a pattern match for (local-)command-* and system-* so the filter keeps working without manual additions. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Simpler than the previous pattern (command-* / system-* etc.) and more robust to future Claude Code tag additions. Inner text is kept so context previews still show what the user actually typed, minus the plumbing wrappers. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Closes a race: if the user responded locally (typed new prompt, ESC'd with transcript growth, etc.) between our last check_local_response and ch.poll() returning a callback, we'd previously return the channel decision and overwrite the local action. Now we re-check immediately after the poll returns an update. If local activity is detected, return "local" instead of the callback data — the message resolves as "🖥️ Handled locally" on TG. Applied to poll_callback, poll_question_answer, and the Stop hook. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- test: 2 new tests exercise the TOCTOU fix (210 total) — verify that poll_callback and poll_question_answer return "local" when transcript grows between the local-response check and ch.poll returning a callback - CLAUDE.md Known Issues: new entry #8 documenting that ESC in Claude Code sends SIGKILL to pending hooks, which we can't catch — so TG buttons linger until the next hook runs or the user resolves on TG Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Match the concision of the other entries in the list. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Multi-line user input used to wrap awkwardly in the channel context block. Collapse all whitespace (newlines, tabs, repeated spaces) to single spaces so each context turn renders as one clean line. Also applies the same collapse to assistant turns (previously we took the first line only — same net effect but simpler code path). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Previously: Continue clicked → clear buttons → transition to "Waiting for instruction..." → send_reply_prompt. If the prompt send failed (transient API error), original buttons were already gone and no prompt existed to reply to — user stuck until timeout. Now send_reply_prompt runs FIRST. On failure the original Continue / Dismiss buttons stay intact so the user can retry or dismiss. Only commit to the "Waiting" transition when the prompt actually lands. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Both notification.py and stop.py previously hardcoded max_turns=1 for idle context, ignoring the configured context_turns (default 3). Now they read from cfg — same knob affects every hook, which is what the documented contract already suggested. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.