Skip to content
Back to skills

Mode

ASecurity

Show or switch the session mode (chat, deep, delegate, or off). Modes are enforced by hooks, not just by instructions. Use for "/shipright:mode", "what mode am I in", or "switch mode".

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 7, 2026
developmentpythonbash

Works with

  • mcp

Security analysis

A100/100

Scanned November 7, 2026

npx -y skills add agoopintsai/shipright --skill mode --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Mode?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Mode
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/agoopintsai-mode/badge)](https://www.skillsdirectory.com/skills/agoopintsai-mode)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: mode
description: Show or switch the session mode (chat, deep, delegate, or off). Modes are enforced by hooks, not just by instructions. Use for "/shipright:mode", "what mode am I in", or "switch mode".
---

# /shipright:mode

Three mutually exclusive modes. The current mode lives in a per-session flag file; hooks inject a banner each turn and check every tool call. It persists until you switch it or the session ends. No mode set means no quotas.

| Mode | Lookups per turn | What is blocked |
|---|---|---|
| chat | 3 (Read, Grep, Glob, Bash, WebFetch, WebSearch) | Replies of more than 300 words are sent back for a rewrite (`chat 200` changes the limit). Agent, Workflow, edits and artifact or document tools are denied. |
| deep | 10 (the same tools plus Agent) | Only read-only subagents (`shipright:scout`, `scout`, `Explore`, `Plan`, `claude-code-guide`). Workflow and edits are denied. |
| delegate | 5 personal lookups | Edits ask first. Agent is not limited but is monitored: more than 4 at once, more than 8 per turn, or one running over 10 minutes raises an alert and a "Subagent status:" line in the reply. |

Skill, AskUserQuestion, ToolSearch, `mcp__ccd_*`, SendMessage, ListAgents and ScheduleWakeup are never counted. Tool calls made inside subagents are not counted either, but edits stay denied in chat and deep.

The red-zone gate (see `/shipright:grill`) is independent of modes and always applies.

## How

- `/shipright:mode <chat|deep|delegate>`: run `python3 "${CLAUDE_PLUGIN_ROOT}/scripts/shipright_hooks.py" set <mode>` (for chat, an optional word limit may follow: `set chat 200`).
- `/shipright:mode off`: run `python3 "${CLAUDE_PLUGIN_ROOT}/scripts/shipright_hooks.py" clear`.
- Status: run `python3 "${CLAUDE_PLUGIN_ROOT}/scripts/shipright_hooks.py" status`.
- Relay the one-line output to the user. Claude cannot change the model itself: remind the user to switch the model and effort in the model menu to match the mode.
- Modes only work in a project with a valid `shipright.yaml`; otherwise the hooks stay inert.

Arguments: `$ARGUMENTS`

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.