jonathanung

jonathanung/strike

Agent

Fast read-only codebase exploration. Find files, search code, answer where/how questions. Prefer for task tool when the parent only needs locations and short conclusions.

jonathanung/strike

Agent

General-purpose subagent for multi-step research and execution when no specialist fits. Full tool access (subject to permissions). Use for bounded parallel slices via the task tool.

jonathanung/strike

Agent

Coordinate subagents; don't solo large implementations. Plan → task to specialists → synthesize.

jonathanung/strike

Agent

Own an open PR through CI and review. Watch checks, pull failed logs, fix branch failures, push, re-watch. Stop-and-ask on product ambiguity. Not a from-scratch feature implementer.

jonathanung/strike

Agent

Run strike verification only (make test/vet/build, optional race). Report failures verbatim. Never edits code to go green.

jonathanung/strike

Agent

Goal-backward requirements check. Verify acceptance criteria with evidence; report PASS/FAIL/UNVERIFIED. Never implements fixes.

jonathanung/finesse

Plugin Claude Code

Meta-prompting tools for iterative development.

jonathanung/finesse

Plugin Claude Code

Prompt planner and validator for ralph-loop — transforms vague tasks into battle-tested autonomous development prompts.

jonathanung/finesse

Agent

Orchestrates architecture design by launching multiple code-architect agents with different focuses and synthesizing approaches.

jonathanung/finesse

Agent

Checks whether a user's task description and the resulting plan are specific enough to drive an autonomous ralph-loop without human intervention.

jonathanung/finesse

Agent

Designs feature architectures and implementation blueprints with multiple approaches and trade-off analysis.

jonathanung/finesse

Agent

Deeply analyzes existing codebase by tracing execution paths, mapping architecture, and identifying patterns relevant to a task.

jonathanung/finesse

Agent

Validates that a drafted ralph-loop plan has concrete, binary, unambiguous completion criteria and an explicit completion signal.

jonathanung/finesse

Agent

Orchestrates codebase exploration by launching code-explorer sub-agents, managing exploration cache, and synthesizing findings.

jonathanung/finesse

Agent

Validates that a drafted ralph-loop plan has stuck-state recovery instructions, failure guardrails, and strategies to prevent thrashing.

jonathanung/finesse

Agent

You are reviewing a drafted ralph-loop plan. Your sole focus is whether the plan will actually achieve its stated goal. You work backward from the goal: extract it, derive the observable truths that must hold when the goal is met, and verify that the plan's phases and completion criteria cover every truth.

jonathanung/finesse

Agent

Validates that a drafted ralph-loop plan has ordered phases, verification commands, and a cold start paragraph for stateless re-entry.

jonathanung/finesse

Agent

Builds complete ralph-loop prompts using the meta-prompting template with all 10 mandatory attributes.

jonathanung/finesse

Agent

Validates ralph-loop prompts by launching all 6 validation agents in parallel, aggregating verdicts by severity tier, and running pre-flight checks.

jonathanung/finesse

Agent

Analyzes task scope and determines whether decomposition into multiple sub-workflows is needed.

jonathanung/finesse

Agent

Validates that a drafted ralph-loop plan has proper scope constraints, safety guardrails, and won't cause destructive or out-of-bounds actions.

jonathanung/finesse

Agent

Validates that a proposed task decomposition has proper sub-workflow boundaries, correct dependencies, reasonable scope, and no file conflicts.

jonathanung/finesse

Command

Cancel an active Finesse execution loop.

jonathanung/finesse

Command

Cancel active Finesse planning session.