skills.rest
cacheforge-ai
cacheforge-aiOFFICIAL
Explore all skills by this author →
WebsiteGitHub

continuity-kernel

Inject bounded continuity data into llm_input with deterministic runtime proofs.

10|2|Updated Feb 14, 2026
Search Tags:#audit#runtime#continuity#deterministic#inject#llm_input#shadow-eval
One-click install
npx skills add https://github.com/cacheforge-ai/cacheforge-skills --skill continuity-kernel
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: continuity-kernel
Source: https://github.com/cacheforge-ai/cacheforge-skills/tree/main/skills/continuity-kernel
Command: npx skills add https://github.com/cacheforge-ai/cacheforge-skills --skill continuity-kernel

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

OpenClaw continuity kernel injects bounded llm_input context, preserving deterministic runtime contracts and recording shadow-mode receipts to enable fail-open reliability and auditability.

Core Features & Use Cases

  • Bounded continuity injection: injects a compact, restart-invariant set of fields into llm_input.
  • Deterministic proofs & receipts: generates runtime_contract_proofs and shadow-eval artifacts for audits and reproducibility.
  • Fail-open safety: preserves baseline payload and continues operation when components fail.
  • Shadow-mode evaluation: runs trace-backed scenarios and outputs repeatable quality snapshots for governance and QA.

Quick Start

Generate a deterministic runtime contract proof artifact and observe shadow receipts with the provided scripts.

Frequently Asked Questions about continuity-kernel

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

FAQPage Schema
How do I maintain deterministic state for AI agents across runtime restarts?▼

You maintain deterministic state for AI agents by injecting bounded continuity data into llm_input prompts. This approach persists Soul Card and Mission Ticket state across lifecycles, ensuring restart-invariant context and repeatable runtime behavior.

Can I generate audit receipts for shadow-mode evaluations of AI agents?▼

Yes, you can generate audit receipts by running trace-backed shadow-mode evaluations. The process outputs deterministic runtime contract proofs and repeatable quality snapshots, providing governance and QA artifacts for reproducibility.

What happens to agent continuity when a runtime component fails?▼

When a runtime component fails, fail-open safety mechanisms preserve the baseline payload and allow the agent to continue operating. A dual-route selector and attention-preserving options ensure robust fallback paths without interrupting the lifecycle.

How do I audit AI agent runtime state for compliance and reproducibility?▼

You audit AI agent runtime state by generating deterministic runtime contract proofs and shadow-eval artifacts. These receipts record bounded llm_input continuity data, providing durable storage and robust fallback paths for compliance verification.

Do I need external dependencies to implement deterministic runtime proofs?▼

No external dependencies are required to implement deterministic runtime proofs. The kernel operates independently to inject bounded continuity fields and generate shadow-mode receipts, satisfying requirements for deterministic fingerprints and durable storage natively.

Related Skills

Explore More Software Engineering Skills→
filipexyzfilipexyz
6

projects

Manage project-level alignment and context for AI agent workflows in the Ravi runtime.

Community
Advanced
Fandry96Fandry96
27

protect-mcp-governance

Evaluate Cedar policies for MCP tool calls and generate Ed25519 audit receipts.

Community
Advanced
almazomalmazom

codex-review

Wrap the local Codex review tool with glm_wp fallback and per-run artifacts.

Community
Advanced
jacob-balslevjacob-balslev

skill-evolution

Automate corpus maintenance for Skill Graph through prioritized triage and bounded improvement cycles.

Community
Advanced