Repository navigation
fix: answer a worker's errand when its classic.Stop is bypassed - #93
Merged
Merged
Conversation
On a machine with managed settings, Claude Code's builtin cc-plugin-sec-default sends classic.Stop on past user-installed plugins, so worker-answer never read a Stop and the worker never answered. A session that has not seen a Stop now reports from turn.complete, held back while $.agent.list() shows a subagent at work.
omartelo
added a commit
to lichdotdev/lich
that referenced
this pull request
Oct 21, 2026
2 tasks done
omartelo
added a commit
to lichdotdev/lich
that referenced
this pull request
Oct 21, 2026
) * fix(terminal): do not read the terminal's query replies as a draft The window's terminal answers a TUI's startup queries (OSC 11, XTVERSION, kitty graphics) through the same input path as the keyboard. escapeSpan read each control string as Alt+key followed by typed text, so a fresh Claude Code session looked drafted for up to a minute and a handoff (Resolve conflicts, Fix CI errors, issue worktree) waited for an Enter. * docs(ceilings): a managed plugin can hide the Stop a worker's answer waits on See lichdotdev/lich-plugin#93.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
On a machine with managed settings (Team and Enterprise organizations), Claude Code 2.1.306's builtin
cc-plugin-sec-defaultseats itself outermost and sendsclassic.Stopon past the plugins a person installs. The engine marks the mod's hookskipped("bypassed by") and queues a meta notice (lich: its classic.Stop hook was skipped. "cc-plugin-sec-default" runs first and sent the event on past it.), but no mod call says so.worker-answer.jsreports only from aturn.completethat follows aclassic.Stop, so a worker in such a session never answered its errand and the caller waited until the ticket expired.Change
classic.Stopreports from a main-loopturn.completewith reasonanswer: itsanswer, orunanswered: "blank"for a blank one. The bypass follows the settings the session started with, so once a Stop has been seen the Stop rule holds for the rest of the session.background_tasks. The report is held while$.agent.list()lists a subagentrunningorwaiting, or cannot be read.docs/mod-answer.mdand the module header: a background shell, monitor or workflow is not listed by$.agent.list(), so a turn that handed one off answers early and lich keeps that first answer.Test plan
node --test tests/*.test.mjs(459 pass), with cases for a Stop seen, a bypassed Stop with an answer and a blank one, a subagent at work holding the report, and a subagent's turn never reportingclaude plugin test .(33 pass), including a worker whoseclassic.Stopnever fires answering fromturn.complete