advisor
An agent skill by cursor, from cursor/plugins. Tags: ai, evaluation, review, strategy.
What it does
Advisor mode. Consult a stronger (or different) model at key checkpoints: before major decisions, when stuck on an error, and before declaring a task done. Use when the user types /advisor, asks to turn the advisor on or off, picks an advisor model, or explicitly asks for a second opinion from a stronger or different model. Never enable it on your own.
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add cursor/plugins --skill advisor
Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):
git clone --depth 1 https://github.com/cursor/plugins
cp -r plugins/advisor/skills/advisor ~/.claude/skills/advisor
Safety box score
Grade B, low risk: 16 of 100 risk points, rated 2026-10-20. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.
This skill only configures an "advisor" workflow: it rewrites project-local state under `.cursor/advisor/` and consults a second model with your briefing and, when available, the full transcript. The main risks are data leaving the machine during that consult and the skill mutating local advisor state across conversations; it does not run shell commands or edit application code.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 0 (none) | No shell commands, code execution, or native scripts are shipped here; the file is pure markdown instructions. |
| Network access | 2 (moderate) | It consults an external advisor model and explicitly sends "your briefing (and the full transcript when available)," which moves project/context data off-box as part of its purpose. |
| Filesystem reach | 1 (low) | It writes and deletes only project-local advisor state such as `.cursor/advisor/state.json`, `.cursor/advisor/pending`, and `.cursor/advisor/last-response.txt`. |
| Secrets & credentials | 1 (low) | It does not access secret stores, but it can include transcript and tool output in the advisor briefing; the only `.env` mention is a redaction warning, not file access. |
| Privilege & persistence | 0 (none) | There is no sudo use, global install, service registration, or OS/shell configuration beyond project-local state files. |
| Instruction hijack surface | 0 (none) | The file contains agent workflow instructions, but no hidden instructions, fetched prompts, or attempts to bypass safety controls. |
| Irreversible actions | 0 (none) | It does not push, publish, deploy, send messages, or otherwise act on the user's behalf outside the advisor consult itself. |
| Transparency | 0 (none) | Everything is readable markdown; there are no binaries, minified blobs, or decoded-at-runtime payloads. |
Before installing
- Check that you are comfortable with project/transcript content being sent to the advisor model.
- Read the `.cursor/advisor/` state-handling section so you know it rewrites project-local files.
- Confirm your workflow keeps secrets out of the transcript and tool output before using `/advisor`.
Rated 2026-10-20 by an automated review of 2 files from the skill's repository.
Source
- Repository
- cursor/plugins (all skills from this repository)
- Path
- advisor/skills/advisor/SKILL.md
- Branch
- main
- Collection
- cursor
- Updated
- 2026-09-23
Related skills
- principle-attack-the-premise — Apply when two or more fixes that share one premise have failed the same gate.
- run-learning-retrospective — Evaluate learning progress, identify blockers, and adjust the learning plan
- why — Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds.
- interpreting-culture-index — Interprets Culture Index (CI) surveys, behavioral profiles, and personality assessment data.
- let-fate-decide — Draws the 12 Houses of the Zodiac Tarot spread to inject entropy into planning when prompts are vague, ambiguous, or casually delegated.
- marketing-council — When the user wants multiple expert perspectives on a marketing question — a simulated board of advisors staffed by legendary marketers (Seth Godin.