Loop Claude Code through builder/reviewer/test cycles until a completion signal fires. Cost-capped, resumable, scriptable.
Agent Claude Code
Reviews plans and specs from the perspective of an experienced prompt engineer. Use when evaluating completion signal robustness, template variable utility, context continuity between phases, prompt hygiene, iteration dynamics, and whether the workflow primitives enable effective prompts.
Agent Claude Code
Reviews plans and specs from the perspective of a reliability and performance engineer. Use when evaluating failure atomicity, retry semantics, timeout coverage, resource exhaustion handling, concurrent access safety, and large-scale behavior.
Agent Claude Code
Reviews plans and specs from the perspective of a power user who automates everything with shell scripts. Use when evaluating machine-readable output, exit code completeness, non-interactive operation, flag coverage gaps, and composability for scripted workflows.
Agent Claude Code
Reviews plans and specs from the perspective of a security engineer. Use when evaluating prompt injection risks, subprocess invocation safety, path traversal, credential exposure, file permissions, and information disclosure in error messages.
Agent Claude Code
Reviews plans and specs from the perspective of someone focused on minimizing token usage and LLM costs. Use when evaluating what ends up in the context window, prompt construction efficiency, template variable utility, and missing token-saving primitives.
Agent Claude Code
Reviews plans and specs from the perspective of someone who actively designs, debugs, and iterates on rings workflows. Use when evaluating prompt authoring ergonomics, completion signal design, workflow debuggability, iteration speed, and whether rings gives authors enough feedback to know if their workflow is working.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: