Skip to content

sotayamashita/bridge-mode

v0.0.2MIT

Outcome-first responses with only the detail the task needs.

Agent skills

bridge-mode 🖖

Complete your request with only the explanation and input it needs.

Motivation

Coding agents often narrate routine work, repeat information, and add unsolicited advice. Bridge Mode takes its cue from the Star Trek bridge computer: answer the request, report what matters, and stop.

Principle

Keep the information you need to understand, act, verify, or decide.

Lead with the answer or result. Follow with the necessary evidence, conditions, and uncertainty. Detailed explanations belong when the task requires them; brevity follows from removing what does not help.

Install

Codex

Run in your terminal:

codex plugin marketplace add sotayamashita/skills
codex plugin add bridge-mode@sotayamashita-skills

Claude Code

Run in Claude Code:

/plugin marketplace add sotayamashita/skills
/plugin install bridge-mode@sotayamashita-skills

Skill-only fallback

For clients that support Agent Skills but not plugins:

npx skills add sotayamashita/skills --skill bridge-mode

FAQ

How does it differ from Caveman?

Both remove filler. Bridge Mode focuses on what to include and what comes first. Caveman focuses on shorter wording, using short sentences and fragments when the meaning stays clear.

You can use them together: Bridge Mode selects and orders the information; Caveman compresses the wording while preserving necessary detail.

Further reading

DocumentPurpose
DesignPrinciples, influences, and design decisions
ContributingHow to propose and validate changes
Evaluation casesExpected behavior and regression checks
ChangelogChanges to the skill's behavior