foxctl-context-curator

Identify stale memory records and generate prioritized maintenance proposals.

3|Updated Apr 14, 2026
One-click install
npx skills add https://github.com/joshka0/foxctl --skill foxctl-context-curator
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: foxctl-context-curator
Source: https://github.com/joshka0/foxctl/tree/main/configs/skills-pack/foxctl-context-curator
Command: npx skills add https://github.com/joshka0/foxctl --skill foxctl-context-curator

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The foxctl-context-curator skill helps you prevent degraded agent performance caused by stale or low-utility memory records, outdated observations, unresolved tensions, and abandoned vault drafts.

Core Features & Use Cases

  • Deterministic context maintenance: Produces a unified maintenance report using age, confidence, count, and utility heuristics with no LLM calls.
  • Actionable proposals across stores: Flags items in Memory, Observations, Tensions, Handoffs, and Vault drafts with suggested actions like review, archive, address_or_close, promote_or_archive, and consolidate.
  • Dry-run first governance: Starts with dry-run reporting so you can review proposals before applying changes, then verifies after remediation.

Quick Start

Run foxctl_context_curator(mode="dry_run", stale_after_days=30) to generate a proposals report for stale context across the workspace.

Frequently Asked Questions about foxctl-context-curator

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

FAQPage Schema
How do I clean up stale memory records and outdated observations in agent workflows?

Context maintenance identifies stale memory records, low-confidence observations, lingering tensions, outdated handoffs, and neglected vault drafts using deterministic heuristics like age, confidence, count, and utility, then outputs prioritized proposals without LLM calls.

How do I prepare agent context for a milestone or handoff without losing important memory?

Pre-milestone cleanup and handoff preparation scan the context plane to flag outdated handoffs and neglected vault drafts, generating a dry-run report with store-scoped summary counts and per-item action proposals so you can review before applying changes.

Can I review context maintenance proposals before they are applied to the memory stores?

Yes, dry-run first governance starts with reporting so you can review proposals like archive, consolidate, or address_or_close across Memory, Observations, Tensions, Handoffs, and Vault drafts before applying any changes, then verifies after remediation.

What is the best way to resolve lingering tensions and triage low-confidence observations deterministically?

Deterministic heuristics using age, confidence, count, and utility metrics triage low-confidence observations and flag lingering tensions for resolution, producing a unified maintenance report with suggested actions like address_or_close and promote_or_archive without requiring LLM calls.

Does context maintenance work without LLM calls for periodic memory curation?

Yes, periodic maintenance runs deterministically without LLM calls, using age, confidence, count, and utility heuristics to identify stale context across the workspace and output a unified report containing store-scoped summary counts and per-item action proposals with record identifiers.