v0.9.3

Overview

This is a maintenance release with three fixes, all found in real feature runs. Planning now goes straight on to development, the review loop now reports the effort level of a Copilot review, and review-loop agents no longer leave a background wait running for hours. Upgrade if you plan features with the plugin or run Copilot review loops, mainly with subagents. No change breaks an existing setup.

Fixes

• Review-loop subagents no longer leave a wait running for hours. Before, a subagent that was not sure the harness would wake it made up its own wait, often a background until ! pgrep -f "await-copilot-review.sh" loop. That loop matches its own command line, so it never ends, and finished subagents still showed as running in the task view. Step 3 of copilot-review-loop now says how any agent waits: the one wait script is the whole wait, and the exit code of the script decides the next step. If the agent cannot count on a wake, it runs the script in the foreground with a bounded timeout. A second watcher is forbidden, and no background shell may still run when the agent reports back. fanning-out-with-worktrees no longer lets a dispatch prompt add wait instructions of its own. (#98)
• The review loop now reports the effort level of a review. Before, the skill said that a Copilot review does not record its effort level, so an agent never reported it. A review in the current format states its level in a **Review effort:** line of its body. The agent now reports the level from that line and names the line as its source. If the line is absent, the agent reports no level, and it does not guess lite. (#96)
• Planning no longer stops to ask how to execute the plan. Before, after planning-a-feature saved the plan, an agent could ask the user to choose "subagent-driven" or "native" execution, because that question comes from superpowers:writing-plans. The agent then did not go on to developing-a-feature. Step 6 now ends the sub-skill at its self-review, and step 9 replaces its handoff. (#94)

Changelog

• fix(skills/planning-a-feature): stop the writing-plans handoff question by @sourcehawk in #94
• fix(skills/copilot-review-loop): read the effort level from the review body by @sourcehawk in #96
• fix(skills/copilot-review-loop): stop review waits that never exit by @sourcehawk in #98
• chore(.claude-plugin): bump version to 0.9.3 by @sourcehawk in #99

Full Changelog: v0.9.2...v0.9.3