How to Adjust Rule 9 in i-have-adhd to Allow Lists Longer Than 5 Items

Modify SKILL.md to increase or remove the 5-item cap on lists when workflows genuinely require more steps.

The i-have-adhd repository provides a portable skill that reshapes coding assistant responses for ADHD-friendly formatting. By default, Rule 6 caps all lists at five items to prevent overwhelm, but the skill is designed to be forked and customized. This guide shows exactly how to adjust that limit in the source.

Where Rule 9 (formerly Rule 6) Lives

The 5-item cap is defined in [SKILL.md], the core skill definition file. According to the source, the rule states:

6. Cap lists at five items

The repository's architecture separates the style guide from agent adapters, so changing this behavior requires editing only one file.

Step-by-Step: Increasing the List Limit

1. Locate and Open SKILL.md

git clone https://github.com/ayghri/i-have-adhd.git
cd i-have-adhd/skills/i-have-adhd
nano SKILL.md

2. Find Rule 6 in the Document

Search for the numbered rule about list caps. In SKILL.md, you'll find it approximately midway through the 10-rule style guide:

6. **Cap lists at five items** – Break longer procedures into sub-lists
   with clear headings, or ask the user which 5 items to tackle first.

3. Modify the Rule to Allow Longer Lists

Replace with your adjusted version. Options include:

Option A: Raise to 10 items

6. **Cap lists at ten items** – For complex migrations or refactors, up to ten
   bounded steps are acceptable. Beyond ten, split into titled phases.

Option B: Conditional cap based on context

6. **Cap lists at five items by default** – Allow 6–10 items when:
   - The user explicitly requests "all steps"
   - Steps are identical boilerplate (e.g., "run this on 8 files")
   - Use sub-lists or pagination for 11+ items.

Option C: Remove the cap entirely

6. **Keep lists scannable** – Prioritize 5 items, but match the true
   complexity of the task. Never artificially split coherent 7-step procedures.

4. Update the Pre-Send Checklist

SKILL.md includes a pre-send validation checklist. Locate it and adjust the list-capping item:

Before:

- [ ] No list exceeds 5 items

After (for 10-item variant):

- [ ] No list exceeds 10 items unless user requested "show all"

5. Commit and Deploy Your Fork

git add SKILL.md
git commit -m "Adjust rule 6: increase list cap to 10 for complex workflows"
git push origin main

Installing Your Modified Skill

Claude Code

claude plugin add /path/to/your/fork/skills/i-have-adhd

Or use the always-on hook with your custom version:

mkdir -p ~/.claude
ln -s /path/to/your/fork/skills/i-have-adhd ~/.claude/i-have-adhd-custom

Gemini CLI

cp /path/to/your/fork/skills/i-have-adhd/agents/gemini.toml ~/.gemini/commands/i-have-adhd-custom.toml

Edit the toml file to point skill_file to your modified SKILL.md.

Verification: Testing Longer Lists

Start a session and verify the new behavior:

User: List all 7 steps to migrate from Express 4 to 5

Expected output with adjusted rule:

1. Run `npm install express@5`.
2. Update error handling: replace `next(err)` patterns in `src/middleware/error.js`.
3. Migrate `body-parser` imports to built-in `express.json()`.
4. Update route string patterns in `src/routes/api.js`.
5. Replace deprecated `res.send(status)` with `res.sendStatus`.
6. Run `npm test` to catch signature changes.
7. Next: paste any `TypeError` from step 6.

If the assistant still truncates to 5 items, verify your skill file is loaded and no other rule (like "suppress tangents") is triggering premature cutoff.

Summary

• Primary change: Edit SKILL.md Rule 6 to adjust or remove the 5-item limit.
• Secondary change: Update the pre-send checklist to match your new threshold.
• Deploy: Point agent adapters to your forked skill file.
• Validate: Test with a 6+ step request to confirm the adjustment works.

The i-have-adhd skill is intentionally minimal and fork-friendly—all behavior changes flow from SKILL.md.

Frequently Asked Questions

How do I know if my change broke the skill?

The repository includes scripts/run_evals.py for validation. Run python3 scripts/run_evals.py validate to check your modified SKILL.md is well-formed, then python3 scripts/run_evals.py run --trials 3 with a small budget to verify output quality.

Can I have different caps for different agents?

Yes. Create agent-specific skill files: SKILL-gemini.md with stricter limits, SKILL-claude.md with relaxed ones. Update each agent's adapter (e.g., agents/gemini.toml) to point to its corresponding file.

Will removing the cap hurt the ADHD-friendly goal?

According to the source design, the cap prevents cognitive overload. If you remove it entirely, compensate with stronger sub-list grouping or progress indicators. The skill's effectiveness is measured by the evaluation rubric in scripts/run_evals.py—run evaluations before and after your change to quantify impact.

Does the rule number matter? The analysis mentions Rule 9.

The repository uses sequential numbering (1–10). Earlier descriptions referenced Rule 6 for list capping. Verify the current numbering in your SKILL.md version—the principle (cap lists at N items) remains the same regardless of its position in the sequence.

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