SkillsAdd

ferueda/agent-skills · Archived

cursor-cli

>- Run Cursor Agent headlessly and delegate work to another Cursor agent over the CLI. Use when the user asks to call, ask, invoke, run, or delegate to Cursor (e.g. "call cursor", "ask cursor", "invoke cursor agent"); when automating Cursor from scripts or agent-to-agent flows; or when the user mentions `agent -p`, headless Cursor CLI, or `cursor-agent.mjs`.

First seen Jun 21, 2026

Installation

$ npx skills add ferueda/agent-skills --skill cursor-cli
View on skills.sh ↗ GitHub repository ↗

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

review-spec Same source · more installs
ferueda/agent-skills

Review a spec document against codebase reality, identifying gaps and ensuring sound, robust im…

86 installs
research-codebase Same source · more installs
ferueda/agent-skills

Research codebase comprehensively using parallel sub-agents. Use when the user asks to "researc…

57 installs
ask-questions Same source · more installs
ferueda/agent-skills

Clarify requirements before implementing. Do not use automatically, only when invoked explicitl…

43 installs
summarize-work Same source · more installs
ferueda/agent-skills

Summarize work done in a spec/plan document.

30 installs

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

find-skills Same topic
vercel-labs/skills

Helps users discover and install agent skills when they ask questions like "how do I do X", "fi…

3.3M installs
agent-browser Same topic
vercel-labs/agent-browser

Browser automation CLI for AI agents. Use when the user needs to interact with websites, includ…

810.4K installs
vercel-react-best-practices Same topic
vercel-labs/agent-skills 697.8K installs
web-design-guidelines Same topic
vercel-labs/agent-skills

Review UI code for Web Interface Guidelines compliance. Use when asked to "review my UI", "chec…

617.3K installs
microsoft-foundry Same topic
microsoft/azure-skills

Build, deploy, evaluate, optimize, fine-tune, and manage Microsoft Foundry agents, models, and …

576.5K installs
azure-prepare Same topic
microsoft/azure-skills

Prepare azd-based Azure projects for deployment: generates azure.yaml, infrastructure (Bicep/Te…

568.3K installs

Also in this package

Other skills from ferueda/agent-skills · top by installs.

npx skills add ferueda/agent-skills
review-spec 86 research-codebase 57 ask-questions 43 summarize-work 30 create-plan 26 implement-plan 24 react-to-review 11 review-implementation 10 interrogator 7 theory-building-review 5

Browse all from ferueda/agent-skills

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Stars 2
License LICENSE
Default branch main
Open issues 0
Status Archived

Skill metadata

Parsed from SKILL.md frontmatter.

Declared agents cursor

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 4,957 B
  • docs SUMMARY.md 375 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 4 installs

SKILL.md

Cursor CLI (Headless)

Delegate work to a Cursor agent from the shell without the IDE chat. Use when the user wants another agent to call, ask, invoke, or run Cursor on a task.

Global script: ~/.agents/skills/cursor-cli/scripts/cursor-agent.mjs Optional launcher: cursor-agent

Requires agent on PATH. Local dev: agent login once. CI/servers: CURSORAPIKEY.

When to use

Situation Use this
User says "call cursor", "ask cursor", "invoke cursor", etc. cursor-agent "…"
Already inside Cursor IDE chat Task tool — not CLI
Long-lived typed service @cursor/sdk

Quick start

# Check auth + CLI version
cursor-agent

# Invoke (read-only by default — no file edits)
cursor-agent "explain the backend routing"

# Allow edits
cursor-agent --force "add a regression test for auth"

# Typed JSON response
cursor-agent \
  --schema-json '{"type":"object","required":["verdict"],"properties":{"verdict":{"type":"string"}}}' \
  "Review auth.ts"

# Continue a session
cursor-agent --resume <session-id> "follow up"

Caller flow: parse stdout → check status → read structuredOutput or result → save sessionId for --resume.

If the launcher is not installed, call the script directly:

node ~/.agents/skills/cursor-cli/scripts/cursor-agent.mjs "your task"

Output

Default format is TOON on stdout (token-efficient). Use --format json when debugging.

Success example:

status: completed
sessionId: uuid
durationMs: 8783
usageSummary: 45271 in, 76 out
result: truncated answer preview…
help[1]: Run `cursor-agent --resume <id> "follow up"` to continue

With --schema / --schema-json:

status: completed
sessionId: uuid
structuredOutput:
  verdict: pass
  summary: …

Errors also go to stdout (not stderr):

status: failed
error: prompt is required
help[1]: Run `cursor-agent "your task"` or --prompt-file / --stdin

Exit codes: 0 success, 1 error, 2 usage.

Envelope fields

Field When
status completed, failed, timed_out
sessionId Resume with --resume
result Answer text (~800 char preview; use --full for all)
structuredOutput With --schema — validated JSON
error Failure message
help Suggested next commands
usageSummary Token counts when available

Flags

Flag Purpose
--force Allow file/shell changes (off by default)
--model <id> e.g. composer-2.5
`--mode plan\ ask` Read-only modes
--workspace <path> Repo root (default: cwd)
--resume <id> Continue a session
--schema <path> JSON Schema file
--schema-json '<json>' Inline schema (mutually exclusive with --schema)
`--format toon\ json` Output format (default: toon)
--full Untruncated result
--verbose Include raw usage object
--max-runtime-ms <n> Overall wall-clock limit (default: 30 minutes)
--idle-timeout-ms <n> Optional no-output timeout; disabled by default because Cursor may be silent while tools run
--quiet Suppress stdout
--prompt-file, --stdin Prompt input
`--sandbox enabled\ disabled` Sandbox override

The wrapper always passes --trust --approve-mcps to Cursor CLI. Use --force only when the delegated agent should edit files or run mutating commands.

Structured output

Cursor CLI has no native JSON-schema mode. The wrapper injects schema instructions into the prompt, parses JSON from the answer, and validates a basic subset (type, required, properties, items, enum).

  • --schema — reusable schema file in the repo
  • --schema-json — one-off inline schema

On success with a schema, prefer structuredOutput over result.

Auth

Context What you need
Your machine (after agent login) Nothing extra
CI / headless server CURSORAPIKEY

Verify: agent status

Install globally

Install the skill folder at ~/.agents/skills/cursor-cli. Node 18+, no npm dependencies.

Optional launcher:

mkdir -p ~/.local/bin
ln -sf ~/.agents/skills/cursor-cli/scripts/cursor-agent.mjs ~/.local/bin/cursor-agent
chmod +x ~/.agents/skills/cursor-cli/scripts/cursor-agent.mjs

Ensure ~/.local/bin is on PATH.

Raw CLI escape hatch

agent -p --trust --approve-mcps --workspace . "prompt"
agent -p --force "apply changes"

Docs: https://cursor.com/docs/cli