Skip to content
Back to skills

Response Format Rules

ASecurity

Sub-skill of core-context-management: Response Format Rules (+3).

  • 17 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 9, 2026
documentationgo

Security analysis

A100/100

Scanned September 24, 2026

npx -y skills add vamseeachanta/workspace-hub --skill response-format-rules --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Response Format Rules?

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

Security grade badge for Response Format Rules
[![Security: A β€” Skills Directory](https://www.skillsdirectory.com/api/skills/vamseeachanta-response-format-rules/badge)](https://www.skillsdirectory.com/skills/vamseeachanta-response-format-rules)

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: core-context-management-response-format-rules
description: 'Sub-skill of core-context-management: Response Format Rules (+3).'
version: 1.0.0
category: _core
type: reference
scripts_exempt: true
---

# Response Format Rules (+3)

## Response Format Rules


#### Output Constraints
- **Tables**: Maximum 10 rows, summarize remainder
- **Code blocks**: Maximum 50 lines, split larger into files
- **Lists**: Maximum 15 items, aggregate beyond
- **Large outputs**: Write to `.claude/outputs/`, return path only

#### Mandatory Response Ending
```
STATUS: [complete|in_progress|blocked] | NEXT: [action] | KEY: [metrics]
```

## Prohibited Actions


1. **No Echo**: Never repeat input data back
2. **No Redundancy**: Don't repeat previous findings
3. **No Over-Explanation**: Keep explanations ≀3 sentences
4. **No Raw Content**: Use file paths instead of pasting
5. **No Unbounded Lists**: Always cap with "and N more..."

## Context Health Indicators


| %ctx | Status | Action |
|------|--------|--------|
| 0-40% | 🟒 Healthy | Normal operation |
| 40-60% | 🟑 Elevated | Consider summarizing |
| 60-80% | 🟠 High | Archive older exchanges |
| 80-100% | πŸ”΄ Critical | Trim to essentials |

## Recovery Patterns


When context exceeds threshold:
1. Create checkpoint with current state
2. Write to `.claude/outputs/session-state.json`
3. Clear verbose history
4. Continue with checkpoint reference only

---

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.