Cursor IDE Configuration & Context Archive

Complete export of Cursor IDE configuration, skills, plans, MCP integrations, agent transcripts, and Codex skills from a macOS (darwin arm64) setup. Captured on March 27, 2026.

Repository Structure

.
├── cursor/                        # ~/.cursor/ configuration
│   ├── config/                    # Core IDE config files
│   ├── skills-cursor/             # Built-in Cursor agent skills (6 skills)
│   ├── plans/                     # AI-generated project plans (5 plans)
│   ├── mcp/                       # MCP (Model Context Protocol) server config
│   │   └── cursor-ide-browser/    # Browser automation MCP (33 tools)
│   ├── extensions/                # Installed extension manifest
│   ├── agent-transcripts/         # Agent conversation logs (JSONL)
│   ├── agent-tools/               # Captured agent tool output
│   └── terminals/                 # Terminal session snapshots
├── codex/                         # ~/.codex/ OpenAI Codex skills
│   └── skills/.system/            # System-level Codex skills
│       ├── skill-creator/         # Skill authoring framework
│       └── skill-installer/       # Skill installation from GitHub
└── README.md

---

Cursor Configuration (cursor/config/)

File	Description
cursor-gitignore	The .gitignore from ~/.cursor/ defining which internal files are tracked
argv.json	Permanent CLI arguments for Cursor (crash reporter config)
blocklist	Empty blocklist ([])
ide_state.json	Recently viewed files state (references Jobs, Brand Tracking, and property management projects)
unified_repo_list.json	Empty repo list ([])

---

Cursor Built-in Skills (cursor/skills-cursor/)

Six built-in skills managed by Cursor, defined in .cursor-managed-skills-manifest.json:

Skill	Description	Auto-invoked?
create-rule	Create .mdc rule files in .cursor/rules/ for persistent AI guidance	Yes
create-skill	Guide for authoring Agent Skills (SKILL.md format)	Yes
create-subagent	Create custom subagents (specialized AI assistants with custom system prompts)	No
migrate-to-skills	Convert "Applied intelligently" rules and slash commands to Agent Skills format	No
shell	Execute literal shell commands via /shell invocation	No
update-cursor-settings	Modify Cursor/VSCode settings.json user preferences	Yes

Key Concepts from Skills

• Rules (.cursor/rules/*.mdc): YAML frontmatter + markdown, can be alwaysApply, glob-scoped, or "applied intelligently"
• Skills (.cursor/skills/*/SKILL.md): Progressive-disclosure modular guides with name and description frontmatter
• Subagents (.cursor/agents/*.md): Custom AI assistants with focused system prompts, project or user level
• Settings path (macOS): ~/Library/Application Support/Cursor/User/settings.json

---

AI Plans (cursor/plans/)

Five completed project plans with YAML frontmatter tracking todos:

Plan	Project Context
Brand Tracking Excel Charts	Python script (generate_charts.py) generating Excel workbooks with embedded charts for 7 markets (US, UK, DE, FR, PL, BR, TR), 2024-2025 brand survey data
Fix Missing Trust Values	Bug fix: missing localized trust-positive answer translations for FR 2024 ("I have confidence") and BR 2024 ("I trust him")
Audit Fix All Data	Systematic audit of 14 CSVs finding 3 remaining bugs in BR 2024 (Likert mapping, recency mapping)
Training Q&A Dataset	Extract Q&A pairs from 6 property .docx files across 3 phases (Finn Bot, Tenant FAQ, Maintenance) into 18 training markdown files
Master HTML Narrative Report	10-chapter lifecycle narrative report for Opera Mobile users with embedded Matplotlib charts

---

MCP Browser Automation (cursor/mcp/cursor-ide-browser/)

The cursor-ide-browser MCP server provides 33 browser automation tools for frontend testing and web interaction:

Tool Categories

Category	Tools
Navigation	navigate, navigate_back, navigate_forward, reload, tabs
Interaction	click, fill, fill_form, type, press_key, hover, drag, select_option, scroll
State	snapshot, take_screenshot, is_visible, is_enabled, is_checked, get_attribute, get_bounding_box, get_input_value
Session	lock, unlock, resize, wait_for, handle_dialog, highlight
Debugging	console_messages, network_requests, search
Performance	profile_start, profile_stop

Key Workflow

1. browser_navigate to URL
2. browser_lock the tab
3. browser_snapshot to get accessibility tree with element refs
4. Interact using refs (click, fill, type, etc.)
5. browser_unlock when done

---

Installed Extensions (cursor/extensions/)

15 extensions installed (summary in installed-extensions.json):

Extension	Purpose
ms-python.python	Python language support
ms-python.debugpy	Python debugging
anysphere.cursorpyright	Cursor's Pyright fork
ms-toolsai.jupyter + renderers + keymap + slideshow + cell-tags	Full Jupyter stack
mtxr.sqltools	SQL client
cweijan.vscode-mysql-client2 + cweijan.dbclient-jdbc	MySQL/JDBC database tools
qwtel.sqlite-viewer	SQLite viewer
mechatroner.rainbow-csv	CSV highlighting
tomoki1207.pdf	PDF viewer
openai.openai-chatgpt-adhoc	OpenAI ChatGPT (VSIX, pinned)

---

Agent Transcripts (cursor/agent-transcripts/)

JSONL conversation logs from 5 Cursor workspaces:

Workspace	Transcript Size	Context
All-Brand-Tracking-2021-2025	210 KB	Brand survey analysis across 7 countries
Jobs	157 KB	Job tracking with Finn.no scraping
Mobile-Lifecycle	327 KB	Opera Mobile user lifecycle survey analysis
drive-download-*	230 KB	Property management AI training data extraction
Github	<1 KB	This workspace (just initialized)

---

Agent Tool Outputs (cursor/agent-tools/)

Captured stdout from agent tool executions:

File	Size	Content
All-Brand-Tracking-*	305 KB	Matplotlib font warnings from build_html_report.py
drive-download-*	43 KB	Topic-mismatch audit for Finn bot markdown

---

Terminal Sessions (cursor/terminals/)

Saved terminal session snapshots with metadata (pid, cwd, exit codes):

Terminal	Workspace	Content
601082.txt	Brand Tracking	Python http.server on port 8767
69698.txt	Brand Tracking	Build/run output
834013.txt	Brand Tracking	Additional session
781221.txt	Jobs	Python Finn.no scraping session

---

Codex Skills (codex/skills/.system/)

Two system-level OpenAI Codex skills:

skill-creator

Comprehensive guide for creating AI agent skills with:

• Progressive disclosure design (metadata → SKILL.md body → bundled resources)
• Three resource types: scripts/, references/, assets/
• Validation via quick_validate.py
• Initialization via init_skill.py
• agents/openai.yaml generation via generate_openai_yaml.py

skill-installer

Install skills from GitHub repositories:

• Curated list from github.com/openai/skills
• Supports private repos via GITHUB_TOKEN/GH_TOKEN
• Direct download with git sparse checkout fallback
• Scripts: list-skills.py, install-skill-from-github.py

---

Workspaces Referenced

The IDE state and transcripts reference these active workspaces:

Workspace Path	Description
~/Desktop/Cursor/Github	This repository
~/Desktop/Cursor/Jobs	Job tracking & Finn.no automation
~/Desktop/Cursor/All Brand Tracking 2021-2025	Multi-country browser brand survey analysis
~/Desktop/Cursor/Mobile Lifecycle	Opera Mobile user survey lifecycle analysis
~/Desktop/Cursor/drive-download-*	Property management AI (Norwegian rental properties)

---

Environment

• OS: macOS (darwin 24.6.0, arm64)
• Cursor: Latest as of March 2026
• Python: Available via system
• Git: 2.39.5 (Apple Git-154)
• Shell: zsh