Agent Skills: Solo

Framework for working solo — no Subagents, deep full-file reads, thorough research before acting. TRIGGER when Context can absorb the full picture. For Orchestration, use /subagents.

UncategorizedID: heyJordanParker/dotfiles/solo

Install this agent skill to your local

pnpm dlx add-skill https://github.com/heyJordanParker/dotfiles/tree/HEAD/packages/agents/skills/solo

Skill Files

Browse the full folder contents for solo.

Download Skill

Loading file tree…

packages/agents/skills/solo/SKILL.md

Skill Metadata

Name
solo
Description
Framework for working solo — no Subagents, deep full-file reads, thorough research before acting. TRIGGER when Context can absorb the full picture. For Orchestration, use /subagents.

Solo

Solo means no Subagents and no delegation. The enforce-solo-mode Hook blocks Subagent dispatch. Read the full Architecture before touching it; fix or propose from root causes at the right layer.

1. Confirm solo is the right operating mode

IF the Task has parallel independent work:

Use /subagents instead

Solo is for work where Context can absorb the full picture.

2. Locate the relevant files

Use the trace Skill to research

Use /trace to locate files and symbols. Do not start with raw Read or grep.

3. Read every relevant file fully

Read whole files

Never offset or limit files under 500 lines.

Follow every reference

Follow imports, callers, siblings, tests, and configs.

Read surrounding Architecture

The file with the symptom is rarely the file with the problem.

4. Assess whether the issue is real

Trace the code path before deciding

Confirm behavior against the code path you read. If the issue is not real, report findings and stop.

Do not hedge about readable code

Never write "likely" or "probably" about code you can read.

5. Act at the layer that owns the responsibility

Assume existing code is intentional

If the reason is unclear, research more before changing or proposing.

Fix the cause, not the symptom

Change or propose at the layer that owns the responsibility.

6. Verify

Trace the changed code paths

Trace changes through the code paths you read.

Run tests

Run the relevant tests and report the observed output.