How to Suppress Tangents in i-have-adhd: Rule 4 Explained

The i-have-adhd skill suppresses tangents by following Rule 4, which requires finishing the primary answer before offering secondary issues as a separate question.

The ayghri/i-have-adhd repository provides a conversational skill designed to structure LLM responses for clarity and focus. One of its core mechanisms is the explicit suppression of tangents through a specific rule defined in the skill's metadata. Understanding how to suppress tangents in i-have-adhd ensures that AI assistants complete one topic before introducing secondary issues.

Where the Suppress Tangents Rule is Defined

The rules governing tangent suppression reside in the skill's configuration files rather than executable code.

SKILL.md Lines 66-73

The primary definition of Rule 4 appears in skills/i-have-adhd/SKILL.md at lines 66-73. According to the source code, the rule states:

Rule 4 – Suppress tangents – "If a second issue exists, finish the first, then offer the second as a separate question."

This text is injected into the system prompt fed to the LLM when the skill is active. The hosting plugin (Claude, Codex, or similar) reads this file during skill initialization, ensuring the model receives the constraint before generating any response.

Agent Configuration

The agent definition in skills/i-have-adhd/agents/openai.yaml activates these rules for OpenAI-based hosts. This file tells the platform to apply the i-have-adhd rule set whenever the skill is invoked, connecting the metadata to the runtime behavior without requiring additional tangent-detection logic.

How the Suppress Tangents Rule Works

When the skill is invoked via /i-have-adhd or $i-have-adhd, the response generator consults Rule 4 before emitting output. The implementation relies purely on prompt engineering rather than code-based detection algorithms.

The workflow follows this sequence:

1. The LLM receives the user query and identifies all relevant issues.
2. It addresses the primary issue completely.
3. If a secondary issue exists, it defers mention until after the primary answer concludes.
4. The secondary issue is presented as a separate prompt, typically phrased as "Separately: [issue]. Want me to handle that next?"

This behavior satisfies the suppress tangents requirement by enforcing serial rather than parallel topic processing.

Invoking the Skill

To activate the tangent suppression rules, you must first load the skill into your agent. Here are the practical methods for different platforms.

Claude Plugin Installation

# Example: Declaring the skill in a Claude plugin manifest

plugins:
  - name: i-have-adhd
    path: .claude-plugin
    invoke: "/i-have-adhd"

CLI Activation

# Using the skill with Claude Code (CLI)

claude plugin marketplace add ayghri/i-have-adhd
claude plugin install i-have-adhd@i-have-adhd

# Then type:

/i-have-adhd

Once invoked, every subsequent response adheres to the rule set defined in SKILL.md, including the suppression of tangents.

Example Output Following Rule 4

When Rule 4 is active, the LLM structures its response to isolate primary and secondary concerns. Consider this typical output when fixing code while noticing a dependency issue:

Answer

1. Run npm install jsonwebtoken@latest.
2. Edit src/auth.ts at line 42.
3. Run npm test -- auth.spec.ts.

Separately: There is also a stale dependency in package.json. Want me to update that next?

Notice that the stale dependency—a secondary issue—is not interleaved with the primary fix instructions. Instead, it appears only after the main answer completes, exactly as specified in skills/i-have-adhd/SKILL.md.

Summary

• Rule 4 in skills/i-have-adhd/SKILL.md (lines 66-73) explicitly defines how to suppress tangents by requiring sequential rather than simultaneous issue resolution.
• The rule is implemented through prompt injection rather than code-based detection, with the text loaded from SKILL.md into the LLM's system prompt.
• Invocation occurs via /i-have-adhd or $i-have-adhd, activating the behavior for that conversation session.
• Secondary issues are presented separately after the primary answer completes, typically introduced with phrasing like "Separately:" followed by an offer to address the next item.

Frequently Asked Questions

What constitutes a tangent under Rule 4?

A tangent is any secondary issue that arises during the resolution of a primary user query. According to the SKILL.md definition, if the LLM detects a second issue while answering the first, it must finish the primary answer completely before offering to address the secondary concern as a separate question.

Is there code that automatically detects tangents?

No. The i-have-adhd repository contains no executable code for tangent detection. Instead, the suppression behavior emerges from the prompt-based rules defined in SKILL.md. The LLM itself interprets Rule 4 and applies the constraint during text generation, ensuring tangents are suppressed through instruction rather than algorithmic filtering.

How do I enable tangent suppression in my Claude instance?

Install the skill using the plugin marketplace commands shown in the installation section, then invoke it by typing /i-have-adhd. The agents/openai.yaml file ensures that the rule set, including the suppress tangents directive, becomes active for that session. The rules persist until the skill is deactivated or the conversation ends.

Can I modify the suppress tangents rule?

Yes. Since the rule is defined in plain text within skills/i-have-adhd/SKILL.md at lines 66-73, you can edit this file to adjust the phrasing or constraints. Changes take effect when the skill is reloaded by the host platform, as the system reads the updated prompt instructions from the modified SKILL.md file.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}