Shared Context Desktop

The context editor

The desktop app for agent skills, agents, and markdown files.
Local. Simple. Free

Download for Mac

Shared Context

Claude Code

113Filter⌘F

A/B Test SetupWhen the user wants to plan, design, or implement an A/B test or experiment.

Ad CreativeGenerate, iterate, and scale ad creative — headlines, hooks, scripts.

Adaptive ResizeAdapt designs to work across screen sizes, devices, and a11y requirements.

Linear MCPRead and write Linear issues, projects, and cycles from any agent.

AI SEOOptimize content for AI-driven search engines and answer surfaces.

Standup RecapSummarize a week of standups into a one-pager. Posts to #eng-updates.

Analytics TrackingSet up, improve, or audit analytics tracking across the product.

AnimateReview a feature and enhance it with purposeful motion and micro-interactions.

Github MCPPRs, issues, reviews, and CI status surfaced inline in chat.

AutoplanAuto-review pipeline — reads CEO/design/eng reviews and synthesizes a plan.

Behavioral ProductApply behavioral economics — friction, defaults, salience, social proof.

BenchmarkPerformance regression detection. Establishes per-route baselines.

Library

Drag a skill from Claude to Codex.

Manage a growing library of skills, agents, MCPs and plugins — across every agent. No cp, no version drift, no half-stale copies in three folders.

Shared Context

Claude Code

113

Filter⌘F

Ad Creative3d

Adapt1w

AI SEO2w

Analytics Tracking3w

Animate1mo

Audit1mo

Autoplan2mo

Auditcopy

Same file. Edit once, both agents pick it up.

Claude Code › Skills › A/B Test Setup

Visual Source

---

name: ab-test-setup

description: Plan, design, or implement an A/B test or experiment.

allowed-tools: [Read, Edit, Bash(git:*)]

---

# A/B test setup

Use this skill when the user wants to plan, design, or implement an experiment — or build a growth experimentation program.

## When to invoke

- User mentions "test", "experiment", "treatment vs. control", or names a metric they want to move.

- A new feature is being scoped and the lift is uncertain.

- Anyone is about to ship a change that touches conversion, retention, or monetization.

## Process

Lead with the **hypothesis**, not the variant. The format is always:

`If <change>, then <metric> will <direction> because <mechanism>`

A/B Test Setup

Plan, design, or implement an A/B test or experiment — or build out a growth-experimentation program from scratch.

User Invocable: Yes
Version: 1.0.3
Allowed Tools: [Read, Edit, Bash(git:*)]

When To Invoke

Test or experiment — user mentions “test”, “experiment”, or “treatment vs. control”.
Metric is named — conversion, retention, activation, or revenue is on the line.
Lift is uncertain — a new feature is being scoped and impact is not obvious

Process

Lead with the hypothesis, not the variant. The format is always: If <change>, then <metric> will <direction> because <mechanism>.

Saved 2s ago·14 versionsMD · 1.2 KB · UTF-8·Ln 12 · Col 84

Files

Markdown that humans can read.

Read any agentic file just like a doc. Skills, plans, memories and more. Agentic code in the back, human in the front.

Projects

Every project gets its own context.

Point Shared Context at a folder and it picks up the CLAUDE.md, the AGENTS.md, and any per-project skills inside .claude/.

curve-shift — Shared Context

curve-shift

~/code/curve-shift· main·12 resources

ResourcesInstructions 2HistorySharing

CLAUDE.md~/code/curve-shift/CLAUDE.md

# Curve Shift
Internal A/B testing platform. Next.js + Postgres.
## Conventions
Prefer server actions. Migrations live in db/migrations/.
All PRs reference an experiment ID

AGENTS.md~/code/curve-shift/AGENTS.md

# Agent guidance
Codex / OpenAI Code agents read this on every run.
Never modify db/migrations/ without an experiment ID.
Run pnpm test:integration before any deploy script.

Skills 4.claude/skills/

Migration ReviewReviews schema or framework migrations before they ship.4d

PRD TemplateDrops the team's PRD scaffold into the current file.2w

RFC FormatReformats an RFC into the in-repo template with the right headers.3w

Sprint NotesGenerates the sprint summary from PR history and design reviews.1mo

Agents 2.claude/agents/

Test RunnerRuns the test matrix, summarizes failures, suggests fixes.—

Deploy CheckerPre-deploy smoke check across staging and prod parity.—

Get it

Make your AI context visible.

Free. Local. No account required.

Download for Mac View the changelog