Plugin marketplace
Reflection/judge layer for OpenCode & Claude Code — catches premature stops, enforces workflow gates (tests, CI, PRs). 78% of agent stops are premature.
Add this marketplace
Then install any plugin below by name — each page has its exact command.
End-of-session retrospective: finds scripts, CLI tools, fixes and corrections from the session and proposes the right home for each (new or edited skill, AGENTS.md/CLAUDE.md, memory, hook, command, subagent). Nothing is written without approval. Also ships an opencode plugin.
1 command · hooks
When the agent stops, a cheap model audits the whole conversation against ~/.agents/AGENTS.md. Rule violations are fed back to the agent on your next message.
hooks
Warns or blocks a new exported declaration when an indexed repo already provides it. The index is built from a repo list and includes real call sites.
hooks
Blocks the end of a turn while GitHub CI for a commit this session pushed is red or still running, and reviews git and GitHub writes against plain-English rules.
hooks
Injects a short-reply, plain-English instruction into every prompt so replies don't overwhelm. Also ships an opencode plugin.
hooks
After 30 minutes of work in a session, inject a reflection prompt (goal, acceptance criteria, what is wrong, is the plan still right, can you unblock yourself) and tell the agent to act on it. Idle sessions are never touched.
hooks
Re-prompts Claude Code when it stops prematurely — catches PERMISSION-SEEKING, STOPPED-WITH-TODOS, and FALSE-COMPLETE failure modes (78% of real agent stops are premature), and injects targeted recovery instructions via the Stop hook.
Shows how many plugins this repo ships, and links here.