Repository navigation
boot-agents.sh reports "launched + armed" while the pane is waiting on a Claude Code prompt (#24) - #54
Closed
simon-inkie wants to merge 5 commits into
Closed
boot-agents.sh reports "launched + armed" while the pane is waiting on a Claude Code prompt (#24)#54simon-inkie wants to merge 5 commits into
simon-inkie wants to merge 5 commits into
Conversation
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Signed-off-by: Claude <noreply@anthropic.com>
…#24) boot-agents.sh now samples each claude agent's pane after launch and reports it ready, waiting on a Claude Code prompt, or not running. The folder trust screen is named with the choice to make. Any agent waiting on a prompt gives exit 3, and auto-arm skips panes that are not ready. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Signed-off-by: Claude <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Signed-off-by: Claude <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Signed-off-by: Claude <noreply@anthropic.com>
Addresses the review of #54: a launch refused with skipped:<reason> is reported as not launched instead of already up, readiness is polled in rounds so one stuck agent cannot starve the rest into a false exit 3, and INSTALL.md first boot is corrected and documents the timeout. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Signed-off-by: Claude <noreply@anthropic.com>
Owner
Author
|
Review leg: Aggy (AI review agent). Verdict: changes needed (superseded by earlier PR #46). |
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
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.
Requested by Simon · project thread
Closes #24
Before:
boot-agents.shprinted "N launched + armed" as soon as it had typed a launch command. A fresh agent stuck on Claude Code's folder trust screen counted as a success, and auto-arm could press Enter on "No, exit".After: the script checks each Claude agent's pane after launch and reports it as ready, waiting on a Claude Code prompt, or not running. For the trust screen it names the choice to make and warns about Enter. If any agent is waiting on a prompt, it exits 3. Auto-arm skips panes that are not ready. INSTALL.md tier 3 has a new "First boot" section, which also covers the permission mode.
How: there is a pure classifier,
boot_pane_state, over captured pane text. A pane that is not ready is one with no rule-anchored composer, so prompts are not enumerated; only the trust screen is named. A pane thatcheck_idle(frombin/pane-idle) reads as busy counts as ready. Fresh panes are polled against one shared deadline (BOOT_READY_TIMEOUT_SECS, default 30). Already-running Claude panes are sampled once.Known limits:
claudeagents are checked. Codex keeps its own launch verification.SPEC:
.specs/24-boot-agents-status-u181xn/SPEC.mdTest plan:
packages/base/test/boot-agents.test.sh. They cover trust-screen fixtures (unnumbered, numbered, older boxed), an unknown offer, a menu under a rule, ready, busy, and prose that quotes the label. The poller's early exits are covered too. Two end-to-end runs use a stub tmux: trust exits 3 and is not armed, ready exits 0 and is armed.pnpm testhas 4 failures that fail the same way onmainin the same container: create-agent "CANNOT VERIFY", and the inbox, inbox-push and pane-idle shell suites. That container hasclaudeon PATH and no UTF-8 locale.🤖 Generated with Claude Code
https://claude.ai/code/session_018wj1S6xHUPXDVTCw8vQxHx
Generated by Claude Code