hatch3r-context-health

Assess conversation context health and apply corrective actions during long coding sessions.

26|4|Updated Feb 27, 2026
One-click install
npx skills add https://github.com/hatch3r/hatch3r --skill hatch3r-context-health
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hatch3r-context-health
Source: https://github.com/hatch3r/hatch3r/tree/main/skills/hatch3r-context-health
Command: npx skills add https://github.com/hatch3r/hatch3r --skill hatch3r-context-health

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prevents silent drift and wrong decisions during long, multi-turn coding sessions by continuously assessing whether the model still has healthy, current context.

Core Features & Use Cases

  • Context health assessment: Evaluates task recall, progress clarity, error patterns, file currency, and scope alignment to determine Green/Yellow/Orange/Red degradation.
  • Corrective action routing: Refreshes context (re-read and re-plan), delegates to a sub-agent with a handoff, or checkpoints and stops with a progress postback to the tracked issue.
  • Context poisoning detection: Flags outdated file content, stale assumptions, contradictory context, and accumulated errors to avoid confident-but-wrong outputs.

Quick Start

Use hatch3r-context-health when you notice repeated errors or drifting behavior in a long coding session, and ask it to assess context health and apply the appropriate corrective action.

Frequently Asked Questions about hatch3r-context-health

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent task drift during long agentic coding sessions?

To prevent task drift during long agentic coding sessions, assess context health continuously using a five-check degradation model that monitors task recall, progress clarity, and error patterns to enforce corrective actions at Green/Yellow/Orange/Red thresholds.

What is context poisoning in AI agent orchestration and how is it detected?

Context poisoning in agent orchestration refers to outdated file content, stale assumptions, contradictory context, and accumulated errors causing confident-but-wrong outputs. It is detected by flagging these specific degradation signals during long multi-turn development sessions.

How do I handle repeated tool failures and stale file knowledge in a coding agent?

Handle repeated tool failures and stale file knowledge in a coding agent by applying corrective action routing, which refreshes context through re-reading and re-planning, delegates work to a sub-agent with a handoff, or checkpoints and stops with a progress postback.

When should I use a context health assessment for ambiguous or irreversible coding tasks?

Use a context health assessment for ambiguous or irreversible coding tasks when you experience many conversation turns, repeated tool failures, or stale file knowledge where delegation to a sub-agent or a complete session restart is needed to prevent wrong decisions.

Can I use model-aware profiles to configure corrective actions for different context degradation thresholds?

Yes, you can use model-aware profiles from configuration to enforce corrective actions based on Green, Yellow, Orange, and Red context degradation thresholds, ensuring long development sessions remain accurate and on track.

What is the best way to recover from silent drift in multi-turn codebase interactions?

The best way to recover from silent drift in multi-turn codebase interactions is to evaluate scope alignment and file currency, then apply corrective actions like context refreshing or agent handoff to prevent incorrect decisions and maintain codebase awareness.