Write, debug, and optimize custom GLSL shaders for the Ghostty terminal. Covers Ghostty's Shadertoy-compatible GLSL ES 3.0 dialect, its terminal-specific uniforms (cursor position and color, focus state, event timestamps), the unsupported Shadertoy uniforms, focus-gated performance patterns, and adapting shaders from…
Reference for deeper Jujutsu (jj) topics beyond the everyday describe/advance workflow. Load when working with conflicts, the operation log, bookmarks, revsets, fork/PR workflows, colocated Git repos, multi-remote setups, divergent changes, or jj config. Also the place to look up fileset glob syntax, non-interactive…
Describe and finalize work with jj-vcs — drafting a description for the current change, splitting mixed work into focused commits, and advancing to a clean working copy. Use when asked to commit, when finishing a unit of work, or when starting new work on top of undescribed changes.
Build comprehensive understanding of a problem by gathering context from GitHub issues, codebase exploration, git history, and linked references. Use when starting work on an issue or investigating a problem.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
Design an implementation as a task graph with explicit dependencies, parallel phases, and agent team topology. Use after gathering context and agreeing on an approach, when the work is large enough that sequencing it badly costs real time.
Create a draft or ready pull request with gh for a jj revision — pushes the bookmark, reviews the full changeset, and drafts a title and body explaining why the change was made. Use when asked to open a PR.
Use when creating, updating, or triaging GitHub issues. Writes issues from a product perspective — user stories, acceptance criteria, context, and dependencies — without prescribing implementation.
Draft, review, or improve a prompt for an LLM or agent against a research-backed rubric — explicit instructions, instruction/data separation, output contracts, reasoning scaffolds, grounding, verification, and evals. Use when writing a prompt, auditing one that misbehaves, or explaining a prompting principle.
Validate an implementation plan by checking for gaps, ambiguity, incorrect assumptions, and missing dependencies. Use after creating a plan and before execution.
Pressure-test a decision — surface constraints, expose trade-offs, and recommend a path. Use proactively when the user explores ideas, directions, or priorities without a clear problem or urgency, and for trade-off analysis. Checks proposals against docs/vision.md and docs/core-principles.md when the project has them.