ponybunny

Use to reconstruct PonyBunny failure paths from logs, traces, state, and runtime evidence in order to identify the first bad transition.

Use when writing ADRs, migration notes, harness principles, verification records, or other technical documentation for PonyBunny.

Use to generate high-signal evaluation cases for PonyBunny changes, including success paths, failure modes, edge cases, and regression-sensitive behaviours.

Use when defining or reviewing runtime events, lifecycle payloads, and event-order expectations for PonyBunny.

Use when designing or reviewing PonyBunny as a harness-first system, especially around boundaries, evaluation flow, observability, and migration structure.

Use when mapping PonyBunny's current state against harness-first best-practice gaps and turning those gaps into actionable engineering inputs.

Use when turning repeated failures, weak evals, weak handoffs, or trace blind spots into concrete improvements to the development harness.

Use when designing or reviewing MCP servers and MCP-exposed capabilities used by PonyBunny.

Use when structuring work so that planning, candidate generation, and verification remain separate and explicit.

Use to identify likely regression areas after a PonyBunny change, especially around contracts, transitions, traces, retries, and observability.

Use to prevent premature completion claims and require evidence-backed status reporting for PonyBunny work.

Use when concluding or pausing work so the next Claude Code session can continue reliably without hidden context.

Use when designing or reviewing run, task, or step states and allowed transitions in PonyBunny.

Use when breaking PonyBunny work into small, ordered, verifiable tasks with explicit dependencies and acceptance criteria.

Use when defining or reviewing tool interfaces, skill payloads, executor-facing contracts, and structured request-response shapes in PonyBunny.

Use to review whether traces, logs, and runtime evidence are sufficient to explain behaviour and support debugging or verification.

Market listener analysis.guess_author_role weak role guesser

Market listener control.tick orchestrator

Market listener data.store_record persistence + dedupe

Market listener source.read_stream for multi-source ingestion

Market listener text.detect_problem_signal classifier

Market listener text.extract_problem_block verbatim extractor

Discover and install skills from skills.sh marketplace

Example skill demonstrating the skill system

Query current weather information for any location

Get current weather using wttr.in (no API key needed)