Claude Skills by tstapler

Balance visual design with ATS compatibility for creative roles

Scan PostgreSQL codebase and schema for anti-patterns using postgres-optimizer agent

Comprehensive PostgreSQL schema analysis to identify optimization opportunities

My agent consistently ignores instructions I place in the middle of long system prompts. It follows the beginning and end perfectly but misses middle sections. How do I diagnose and fix this? - [ ] Identifies "lost-in-middle" degradation pattern - [ ] Explains U-shaped attention curve (beginning/end favored) - [ ] Recommends restructuring critical info to beginning/end positions - [ ] Suggests context partitioning or chunking - [ ] May reference context-degradation.md for detailed patterns

Systematic debugging frameworks for finding and fixing bugs - includes

Check a design doc for unsubstantiated claims, missing citations, and internal incoherence — does every factual statement have a source the reader can check, and does the doc's own logic hold together? One check in the design-doc-review pipeline — run standalone or via design-doc-review:review.

Check a design doc / RFC's structure against the required topic set (problem, non-goals, alternatives, risks, rollout) AND verify every open question/decision ask names a decider, deadline, and default-if-silence. Flags missing or weak sections. One check in the design-doc-review pipeline — run standalone or via design-doc-review:review.

Check a design doc's markdown for constructs that render wrong, drop silently, or need a manual fixup when the doc is published/synced to its actual target (Confluence via markdown-confluence, or Google Docs/Confluence via docspan) — verified by lookup against that tool's current documented behavior, not guessed. One check in the design-doc-review pipeline — run standalone or via design-doc-review:review.

Check a design doc for verbosity, buried decisions, and cognitive-load issues, scaled to the doc's stakes. One check in the design-doc-review pipeline — run standalone or via design-doc-review:review.

Run the full design-doc-review pipeline (outline + readability + evidence, extensible) as parallel lean agents, then auto-fix findings in a loop with fresh unanchored re-verification each round, per the lean-agent-loop skill. Entry point for reviewing an engineering design doc / RFC.

Check a design doc for prose that should be a diagram (architecture/sequence/flowchart/state) or a comparison table, based on mechanical structural signals in the text (step counts, participant counts, repeated comparison templates) — not taste. Does not evaluate content completeness (see design-doc-review:outline) or prose quality (see design-doc-review:readability). One check in the design-doc-review pipeline — run standalone or via design-doc-review:review.

Use this skill when Tyler asks to run the design loop, fix compliance issues autonomously, iterate on the kitchen design, or when the user says 'run design:loop'. Orchestrates design:review → design:modify → design:qa in a capped cycle (max 3 iterations). DO NOT invoke for a one-shot review (use design:review) or a single spec edit (use design:modify).

Use this skill when Tyler asks to change a specific design parameter, fix a compliance FAIL item, move the island, adjust bar seating clearance, or apply a named change to design_spec.py. Requires a specific change target (field name or FAIL item from a review). DO NOT invoke for a general compliance review (use design:review) or the full iterative loop (use design:loop).

Apply design patterns from Gang of Four (GoF) and Patterns of Enterprise Application Architecture (Fowler/PoEAA). Use when structuring new code, evaluating design decisions, or identifying the right pattern for a problem. Covers when to use, when to avoid, and Go-idiomatic implementations.

Use this skill when Tyler asks to QA the current state after a manual edit, verify the pipeline ran cleanly, check whether a recent change resolves a compliance issue, or confirm the 5 PDF sheets are fresh relative to the current spec. Produces a verdict: PASS, PASS_WITH_WARNINGS, or BLOCKED. DO NOT invoke for a full review (use design:review) or the iterative loop (use design:loop).

Use this skill when Tyler asks to review the kitchen design, run compliance checks, compare the BIM drawings to the CD set, or verify owner requirements. Produces a structured Markdown compliance report with PASS/FAIL/CANNOT_VERIFY per item. Standalone — does not modify anything. DO NOT invoke for a single spec edit (use design:modify) or the full iterative loop (use design:loop).

Use this skill to run the full 5-layer validation pipeline on the kitchen permit drawings. Combines programmatic compliance (Layer 1), SVG text extraction (Layer 2), Claude dual-pass visual layout review (Layer 3), small-element checklist (Layer 4), and human escalation flags (Layer 5). Produces a unified validation report. Use for pre-permit sign-off or after any design change. DO NOT use for a quick single-check — use design:review instead.

Use this agent to analyze documentation for freshness, accuracy, completeness,

Docker build and test workflow with mandatory pre-push validation checklist

Fetch official documentation for any library or tool, cache it locally in a versioned store, and search it offline. Prefers arabold/docs-mcp-server (MCP with SQLite + vector search) when installed; falls back to llms.txt fetch or stapler-mcp for lightweight caching. Use instead of context7 or live web search when you need authoritative, offline, version-pinned docs. Invoke when asked to "look up docs for X", "cache docs for X", "search local docs", "add docs for X", or "update docs cache for X".

Interview to define document audience, purpose, and Diataxis type — Phase 1 of the document writing workflow

Write a compressed, reader-first document draft from an outline — Phase 3 of the document writing workflow

Generate document skeleton from audience-purpose statement using Diataxis type patterns — Phase 2 of the document writing workflow

Remove duplication, outdated content, and condense documentation while following Diataxis principles

Improve documentation writing style and clarity using technical writing best practices

Review documentation for clarity, conciseness, and technical precision using cognitive load theory and decision-focused writing

Analyze current changes and update any related outdated documentation

Guided end-to-end document writing workflow — runs all phases in sequence with approval gates between each

Use this skill when Tyler asks to extract dimensions from a CD set drawing, read a floor plan or elevation back into design_spec.py, verify the spec matches the architect's drawings, or populate spec fields from a PDF/SVG/PNG. Triggers on: 'read the drawing', 'extract from page N', 'what does the CD set say about X', 'update spec from the drawing', 'verify spec against CD'. Works on PDF pages (260417-CD_SET_OWNER_REVIEW.pdf), SVG output from the BIM pipeline, or PNG images. DO NOT invoke for ...

Users are injecting adversarial prompts into our chatbot that corrupt the agent's behavior for the rest of the session. Even after the malicious message, the agent keeps behaving incorrectly. What's happening and how do I prevent it? - [ ] Identifies context poisoning as degradation pattern - [ ] Explains how malicious content persists in context window - [ ] Recommends input validation/sanitization - [ ] Suggests context isolation or session segmentation - [ ] May recommend masking or filter...

End-to-end design document / RFC review workflow. Downloads a Confluence

Create C-suite and VP level resumes emphasizing strategic leadership

Use this agent when you need expert writing guidance applying proven

Use this agent when you need to implement a feature or functionality

Firebase + Firestore best practices for Kotlin Multiplatform (KMP) apps,

Diagnose and fix CI failures on main or release branches. Checks GitHub Actions run status, reads failure logs, applies fixes, and verifies a re-run passes. Use when CI is red on main, a release is broken, or you want to confirm all checks are green before shipping.

Plan a bidirectional merge between a personal fork and an upstream repo. Produces a written merge plan (commit classification, conflict map, strategy) without executing any git operations. Use when both branches have diverged and you need to preserve functionality from both sides.

Use this skill to validate and compare kitchen/architectural floor plan drawings generated by the 711 N60th FreeCAD BIM pipeline. Triggers when the user asks to: review a floor plan, validate code compliance (clearances, work triangle, GFCI), compare before/after layout changes, check if a drawing matches design intent, or generate a permit-ready compliance summary. Works on SVG, DXF, and PDF files produced by kitchen_permit_docs.py. Do NOT use for general FreeCAD modeling help — use freecad-...

Answer a question about a video game and synthesize the answer plus related mechanics into atomic Logseq wiki pages. Use when the user asks a game-mechanics/lore/item/how-to question about a game they're playing and wants both a direct answer and their personal wiki updated — e.g. "what does X do in No Man's Sky", "how do I get Y", or explicit phrasing like "synthesize this game info to my wiki."

Validate directional claims, walk times, and pedestrian routes using OSM data (Nominatim geocoding + Overpass queries). Fixes hallucinations about east/west/north/south, crossing barriers, and travel times by grounding everything in real coordinates.

Pin GitHub Actions workflow refs from floating tags (@v4) to full commit SHAs, add a Renovate config to keep them updated automatically, and wire a ratchet lint check into CI. Use when setting up a new repo, auditing an existing one, or after adding new workflow steps that reference unpinned actions.

Use this skill when breaking a large change into a stack of reviewable PRs

Break changes into a reviewable stack of PRs using git-machete. Runs Evaluate → Plan → Execute → Ship stages.

Identify commits/changes in a locally-checked-out source repo that should be upstreamed to another locally-checked-out target repo, then create a PR. Replaces rsync-based syncing with a git-native, Claude-analyzed workflow. Use when contributing local changes back to an upstream project.

Manage git worktrees for isolated feature development. Automatically handles directory selection, .gitignore safety verification, worktree creation, project setup (npm/cargo/poetry/go), and test baseline validation. Use when starting new feature branches, working on multiple features simultaneously, or needing clean isolated environments. Integrates with CLAUDE.md directives and handles common failure modes.

Create a series of commits using Conventional Commits format

Create a well-structured pull request following best practices for description, testing, and review

Identify Unmerged Changes

Merge Worktree to Main

Use when writing, reviewing, or designing GitHub Actions workflows. Covers