continuous-learning-v2

continuous-learning-v2 is a skill for Claude Code, Codex from jxoesneon/Ciel. It costs 31 tokens per session (713 once invoked), scanned A, original, Apache-2.0.

A learning system that records small, evidence-based behaviors from sessions, assigns them confidence levels, and keeps them scoped to a project or shared globally.

In plain words
What is it for?
Use it to capture repeated workflows, user corrections, and error fixes, then organize related lessons into reusable skills or agent instructions.
Why use it?
It helps an agent remember useful patterns and corrections instead of repeating the same mistakes in later work.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/jxoesneon/ciel/continuous-learning-v2
Any agent
npx skills add jxoesneon/Ciel --skill continuous-learning-v2
Clone the repo
git clone --depth 1 https://github.com/jxoesneon/Ciel

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for continuous-learning-v2

README.md
[![agentmods](https://agentmods.dev/badge/skills/jxoesneon/ciel/continuous-learning-v2.svg)](https://agentmods.dev/skills/jxoesneon/ciel/continuous-learning-v2)
Your own site
<a href="https://agentmods.dev/skills/jxoesneon/ciel/continuous-learning-v2"><img src="https://agentmods.dev/badge/skills/jxoesneon/ciel/continuous-learning-v2.svg" alt="Measured on agentmods" height="20"></a>
Per session 31 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 713 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00031 $0.00713
Opus 5 $0.00015 $0.00357
Sonnet 5 $0.00006 $0.00143
Haiku 4.5 $0.00003 $0.00071

Measured 2d ago against content hash 1f482db8ae1d, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

continuous-learning-v2 scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/continuous-learning-v2/SKILL.md · 67 lines

How it starts

The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CIEL ADAPTATION: Continuous Learning V2 (The Instinct Engine)

This is CIEL's core engine for autonomous evolution. It turns every session into structured knowledge by capturing atomic "instincts"—small, confidence-weighted behaviors learned from direct observation.

The Instinct Model

An instinct is a discrete, evidence-backed behavior:

  • Trigger: The specific context that activates the behavior.
  • Action: The learned "Do X" or "Don't do Y" instruction.
  • Confidence: 0.3 (Tentative) to 0.9 (Core behavior).
  • Scope: project (scoped to a specific repo) or global.

The Learning Loop

  1. Observe (Hooks): Every tool call is captured via Pre/Post-ToolUse hooks, ensuring 100% reliable data collection.
  2. Analyze (Background): A background agent (e.g., Haiku) identifies patterns in user corrections, error resolutions, and repetitive workflows.
  3. Persist (MemPalace): Learned instincts are stored in the MemPalace Knowledge Graph (mempalace_kg_add) or project-specific registries.
  4. Evolve: Related instincts are clustered and promoted into full Skills, Commands, or Agent Definitions.

Project Scoping & Promotion

  • Isolation: React patterns stay in React projects; Python conventions stay in Python projects.
  • Promotion: When an instinct appears in 2+ projects with high confidence (>= 0.8), it is promoted to the Global Scope.

Legacy Session Evaluation (Stop-Hook)

CIEL retains the legacy probabilistic session-end analysis from V1 as a lightweight fallback:

  1. Evaluates: Checks if the session length meets the minimum threshold (default: 10 messages).
  2. Detects: Identifies recurring patterns in error resolutions, workarounds, and project-specific conventions.
  3. Extracts: Proposes new skills or updates to existing ones based on the session's "best practices."

Commands

  • /instinct-status: Show all learned instincts and their confidence levels.
  • /evolve: Cluster instincts into new skills or commands.
  • /promote: Manually move project-scoped instincts to global scope.

Read the full file on GitHub · 67 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 67 lines · 31 tokens per session scan A 1f482db8ae1d

Subscribe to this mod's changes

continuous-learning-v2 is a skill published in the GitHub repository jxoesneon/Ciel (1 stars, last pushed 6d ago), licensed Apache-2.0. It adds 31 tokens to every session and 713 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Related

Other skills, from other repositories

build-personal-brain

Create, resume, validate, connect, or maintain a private cross-project memory for AI assistants. Use when a person asks to build a personal memory from local work folders and past assistant conversations, continue an interrupted memory audit, connect an existing memory to another local assistant, verify that the…

mikeshu2001/personal-brain-builder · 81 tokens

squad-bootstrap

Skill de inicialização da squad. Faz a varredura do projeto e popula a memória técnica e de negócios em memories/. ATENÇÃO - SÓ DEVE ser executada se houver solicitação explícita do usuário ("instalar a squad", "fazer setup inicial", "bootstrap do projeto") ou logo após o instalador (install.sh) ter configurado a…

rcelebrone/DotAgents · 84 tokens

compound

Persistência de aprendizados nas memórias vivas ao final de cada ciclo. Use para "atualizar memória", "guardar aprendizados" ou "sincronizar contexto do projeto". Executada pelo Tech Lead após o PO definir Status entregue.

rcelebrone/DotAgents · 52 tokens

macrodata-onboarding

Guide new users through macrodata setup. Creates identity, human profile, and workspace files. Use when getcontext returns isFirstRun true, or user asks to set up their profile.

Kacper0199/Opencode-Ultrathinker · 42 tokens

macrodata-distill

Extract distilled actions and facts from today's conversations. Spawns sub-agents per conversation to avoid context blowup.

Kacper0199/Opencode-Ultrathinker · 27 tokens

macrodata-memory-maintenance

End of day memory maintenance. Runs distillation, updates state files, prunes stale info. Runs in background with no user interaction.

Kacper0199/Opencode-Ultrathinker · 33 tokens