Dalt's Bazar

tekierz's Claude Code plugin & skill marketplace — a home for plugins and skills worth sharing: CLI delegation, design languages, and whatever comes next.

Add the marketplace once and every plugin here is an install away:

/plugin marketplace add tekierz/dalts-bazar

Plugins

Plugin	What it does	Install
cursor	Use the Cursor CLI (cursor-agent) from Claude Code for reviews and task delegation (docs below)	/plugin install cursor@dalts-bazar
codex	Use Codex from Claude Code for reviews and task delegation (docs below)	/plugin install codex@dalts-bazar
cyber-deck-ui	Cyber-deck design language skill — cyberpunk / retro-future / CRT / EVA instrument-panel UI (repo)	/plugin install cyber-deck-ui@dalts-bazar

After installing any plugin, reload:

/reload-plugins

---

The Cursor Plugin

Use the Cursor CLI (cursor-agent) from inside Claude Code for code reviews or to delegate tasks to the Cursor agent.

This plugin is for Claude Code users who want an easy way to hand work to Cursor from the workflow they already have. It is a port of OpenAI's Codex plugin adapted to the Cursor CLI; this marketplace ships both plugins, so /codex:* stays available here too (see The Codex Plugin).

What You Get

• /cursor:review for a normal read-only Cursor review
• /cursor:adversarial-review for a steerable challenge review
• /cursor:rescue, /cursor:transfer, /cursor:status, /cursor:result, and /cursor:cancel to delegate work, hand off sessions, and manage background jobs

Requirements

• Cursor subscription or Cursor API key.

  • Usage will contribute to your Cursor usage limits.

• Node.js 18.18 or later

• The Cursor CLI (cursor-agent) installed on your machine:

  curl https://cursor.com/install -fsS | bash

Install

Add the marketplace in Claude Code (if you haven't already):

/plugin marketplace add tekierz/dalts-bazar

Install the plugin:

/plugin install cursor@dalts-bazar

Reload plugins:

/reload-plugins

Then run:

/cursor:setup

/cursor:setup will tell you whether the Cursor CLI is ready. It does not install anything for you; if cursor-agent is missing, it prints the curl one-liner above so you can run it yourself.

If the Cursor CLI is installed but not logged in yet, run:

!cursor-agent login

or set CURSOR_API_KEY in your environment.

After install, you should see:

• the slash commands listed below
• the cursor:cursor-rescue subagent in /agents

One simple first run is:

/cursor:review --background
/cursor:status
/cursor:result

Usage

/cursor:review

Runs a normal Cursor review on your current work.

Note

Code review especially for multi-file changes might take a while. It's generally recommended to run it in the background.

Use it when you want:

• a review of your current uncommitted changes
• a review of your branch compared to a base branch like main

Use --base <ref> for branch review. It also supports --wait and --background. It is not steerable and does not take custom focus text. Use /cursor:adversarial-review when you want to challenge a specific decision or risk area.

Examples:

/cursor:review
/cursor:review --base main
/cursor:review --background

This command is read-only and will not perform any changes: it runs in Cursor's plan mode, which cannot execute shell commands, so the plugin collects the diff itself and the reviewer works from that plus file reads. When run in the background you can use /cursor:status to check on the progress and /cursor:cancel to cancel the ongoing task.

/cursor:adversarial-review

Runs a steerable review that questions the chosen implementation and design.

It can be used to pressure-test assumptions, tradeoffs, failure modes, and whether a different approach would have been safer or simpler.

It uses the same review target selection as /cursor:review, including --base <ref> for branch review. It also supports --wait and --background. Unlike /cursor:review, it can take extra focus text after the flags.

Use it when you want:

• a review before shipping that challenges the direction, not just the code details
• review focused on design choices, tradeoffs, hidden assumptions, and alternative approaches
• pressure-testing around specific risk areas like auth, data loss, rollback, race conditions, or reliability

Examples:

/cursor:adversarial-review
/cursor:adversarial-review --base main challenge whether this was the right caching and retry design
/cursor:adversarial-review --background look for race conditions and question the chosen approach

This command is read-only. It does not fix code.

/cursor:rescue

Hands a task to the Cursor agent through the cursor:cursor-rescue subagent.

Use it when you want Cursor to:

• investigate a bug
• try a fix
• continue a previous Cursor task
• take a faster or cheaper pass with a smaller model

Note

Depending on the task and the model you choose these tasks might take a long time and it's generally recommended to force the task to be in the background or move the agent to the background.

It supports --background, --wait, --resume, and --fresh. If you omit --resume and --fresh, the plugin can offer to continue the latest rescue chat for this repo.

Examples:

/cursor:rescue investigate why the tests started failing
/cursor:rescue fix the failing test with the smallest safe patch
/cursor:rescue --resume apply the top fix from the last run
/cursor:rescue --model gpt-5.6-sol-high investigate the flaky integration test
/cursor:rescue --model 'claude-opus-4-8[effort=high]' redesign the retry logic
/cursor:rescue --background investigate the regression

You can also just ask for a task to be delegated to Cursor:

Ask Cursor to redesign the database connection to be more resilient.

Notes:

• if you do not pass --model, your Cursor account's default model is used.
• there is no --effort flag. Effort is encoded in the model id (e.g. gpt-5.6-sol-high) or in bracket parameters (e.g. --model 'claude-opus-4-8[effort=high]'). Run cursor-agent models to see what's available.
• the gpt-5.6 series comes in three flavors — gpt-5.6-sol-*, gpt-5.6-terra-*, gpt-5.6-luna-* — each with effort tiers none/low/medium/high/xhigh/max and -fast variants (e.g. gpt-5.6-luna-max, gpt-5.6-terra-xhigh-fast).
• some models are gated behind Cursor's Max Mode (currently the gpt-5.6/gpt-5.5/gpt-5.4 families, grok-4.5, and the claude-* tiers). Without Max Mode those runs fail at turn start with Max Mode Required — the plugin appends a hint when this happens. cursor-agent models lists gated models without marking them; models that run everywhere include gpt-5.3-codex-*, gpt-5.2-codex-*, gpt-5.1-codex-max-*, and composer-2.5*.
• follow-up rescue requests can continue the latest Cursor chat in the repo

/cursor:transfer

Creates a new Cursor chat primed with the current Claude Code session and prints a cursor-agent --resume <chat-id> command.

Use it when you started a debugging or implementation conversation in Claude Code and want to continue that same context directly in the Cursor CLI.

Examples:

/cursor:transfer
/cursor:transfer --source ~/.claude/projects/-Users-me-repo/<session-id>.jsonl

The plugin's existing SessionStart hook supplies the current transcript path automatically; --source is available as a manual override, and the source must be under ~/.claude/projects. Cursor has no session importer, so the plugin distills your Claude transcript into a compact handoff digest, creates a fresh chat, and primes it with that digest in one read-only turn. The Cursor agent acknowledges the current state and open threads, then waits for you in cursor-agent --resume <chat-id>.

/cursor:status

Shows running and recent Cursor jobs for the current repository.

Examples:

/cursor:status
/cursor:status task-abc123

Use it to:

• check progress on background work
• see the latest completed job
• confirm whether a task is still running

/cursor:result

Shows the final stored Cursor output for a finished job. When available, it also includes the Cursor chat ID so you can reopen that run directly in the Cursor CLI with cursor-agent --resume <chat-id>.

Examples:

/cursor:result
/cursor:result task-abc123

/cursor:cancel

Cancels an active background Cursor job by terminating its cursor-agent process.

Examples:

/cursor:cancel
/cursor:cancel task-abc123

/cursor:setup

Checks whether the Cursor CLI is installed and authenticated. It does not auto-install; if cursor-agent is missing it prints the install one-liner, and if you are logged out it suggests !cursor-agent login (or CURSOR_API_KEY).

You can also use /cursor:setup to manage the optional review gate.

Enabling review gate

/cursor:setup --enable-review-gate
/cursor:setup --disable-review-gate

When the review gate is enabled, the plugin uses a Stop hook to run a targeted Cursor review based on Claude's response. If that review finds issues, the stop is blocked so Claude can address them first.

Warning

The review gate can create a long-running Claude/Cursor loop and may drain usage limits quickly. Only enable it when you plan to actively monitor the session.

Typical Flows

Review Before Shipping

/cursor:review

Hand A Problem To Cursor

/cursor:rescue investigate why the build is failing in CI

Start Something Long-Running

/cursor:adversarial-review --background
/cursor:rescue --background investigate the flaky test

Then check in with:

/cursor:status
/cursor:result

Differences From The Codex Plugin

If you are coming from /codex:*, the surface is the same but the runtime is simpler — one-shot headless cursor-agent processes instead of a persistent app server. The differences that matter:

• Read-only means plan mode. Reviews and other read-only runs use Cursor's plan mode, which rejects all shell commands. The plugin gathers the git diff itself and hands it to the reviewer, which can still read files in your working tree.
• No --effort flag. The Cursor CLI encodes effort in model ids (gpt-5.6-sol-high) or bracket parameters (--model 'claude-opus-4-8[effort=high]'). Passing --effort to /cursor:rescue is an error that tells you exactly this. There is also no spark model alias.
• Transfer is create + prime. Cursor has no external-session importer, so /cursor:transfer creates a new chat and primes it with a distilled transcript digest rather than importing turns natively.
• Cancel kills the process. There is no turn-interrupt API; /cursor:cancel terminates the job's cursor-agent process tree.

Cursor Integration

The plugin delegates through the local cursor-agent binary — the same install, authentication state, repository checkout, and configuration you would use running the Cursor CLI directly.

Common Configurations

• User-level CLI config lives in ~/.cursor/cli-config.json (default model and other CLI settings).
• Project permissions live in .cursor/cli.json, where you can allow or deny specific tools and commands for agent runs in that project.
• Rules and context: cursor-agent reads .cursor/rules, AGENTS.md, and CLAUDE.md from your project, so the instructions you already maintain apply to delegated runs too.

Check out the Cursor CLI docs for more configuration options.

Moving The Work Over To Cursor

Delegated tasks and any stop gate run can be resumed inside the Cursor CLI by running cursor-agent --resume <chat-id> with the chat ID you received from /cursor:result or /cursor:status.

This way you can review the Cursor work or continue the work there.

The Codex Plugin

The original Codex plugin remains available, unchanged, in two places:

• from this marketplace: /plugin install codex@dalts-bazar
• from upstream: /plugin marketplace add openai/codex-plugin-cc

Its source lives in plugins/codex and is fully documented in the upstream README. The short version:

/codex:review

Runs a normal Codex review on your current work through Codex's built-in reviewer. Use --base <ref> for branch review; it also supports --wait and --background. It is not steerable and does not take custom focus text.

/codex:adversarial-review

Runs a steerable review that questions the chosen implementation and design. It uses the same review target selection as /codex:review, including --base <ref>, and can take extra focus text after the flags:

/codex:adversarial-review --base main challenge whether this was the right caching and retry design

/codex:rescue

Hands a task to Codex through the codex:codex-rescue subagent — investigate a bug, try a fix, or continue a previous Codex task. It supports --background, --wait, --resume, and --fresh:

/codex:rescue --model gpt-5.4-mini --effort medium investigate the flaky integration test

Notes:

• if you do not pass --model or --effort, Codex chooses its own defaults.
• if you say spark, the plugin maps that to gpt-5.3-codex-spark

/codex:transfer

Creates a persistent Codex thread from the current Claude Code session and prints a codex resume <session-id> command.

/codex:status

Shows running and recent Codex jobs for the current repository.

/codex:result

Shows the final stored Codex output for a finished job, including the Codex session ID.

/codex:cancel

Cancels an active background Codex job.

/codex:setup

Checks whether Codex is installed and authenticated. If Codex is missing and npm is available, it can offer to install Codex for you; if Codex is installed but not logged in, run !codex login. It also manages Codex's optional stop-time review gate:

/codex:setup --enable-review-gate
/codex:setup --disable-review-gate

FAQ

Do I need a separate Cursor account for this plugin?

If you are already signed into the Cursor CLI on this machine, that account works immediately here too. This plugin uses your local cursor-agent authentication — sign in with cursor-agent login, or set CURSOR_API_KEY for API-key auth. Run /cursor:setup to check whether Cursor is ready.

Does the plugin use a separate Cursor runtime?

No. This plugin runs your local cursor-agent binary headlessly on the same machine.

That means:

• it uses the same Cursor CLI install you would use directly
• it uses the same local authentication state
• it uses the same repository checkout and machine-local environment

Will it use the same Cursor config I already have?

Yes. The plugin picks up your ~/.cursor/cli-config.json, your project's .cursor/cli.json permissions, and the .cursor/rules, AGENTS.md, and CLAUDE.md files cursor-agent already reads.

Can I keep using both plugins side by side?

Yes. The commands live in separate namespaces (/cursor:* and /codex:*) and track their jobs separately, so installing both from this marketplace is fine.