Tired of Your AI Making the Same Mistakes?

AI that remembers what went wrong and learns not to repeat it. 7 skills for failure-anchored memory. For developers building autonomous agents with Claude Code, Cursor, or any LLM.

From Failure to Learning

Most AI forgets its failures. Same mistakes, session after session. Agentic skills fix that. When something goes wrong, it gets recorded. When the same mistake happens again, it becomes a pattern. When the pattern is confirmed by multiple sources, it becomes a rule.

AI learns from consequences, not instructions. Failures become constraints.

How It Works

Failure detected
Pattern recognized
Constraint generated
Rule enforced

Eligibility Criteria

3

Recurrence

Seen at least 3 times

2

Confirmations

Verified by multiple sources

< 20%

Disconfirmation rate

Not frequently disputed

The 7 Skills

failure-memory/fmStart Here

Records failures with R/C/D counters, detects recurring patterns.

constraint-engine/ceWhat Happens Next

Generates constraints from recurring failures, enforces them at runtime.

context-verifier/cvSupporting

Verifies file integrity via SHA-256 hashes, detects unauthorized changes.

review-orchestrator/roSupporting

Coordinates multi-perspective reviews (technical, creative, external).

governance/govLifecycle

Manages constraint lifecycle, triggers 90-day reviews.

safety-checks/scAdvanced

Validates model configs, enforces pinning, provides fallbacks.

workflow-tools/wtAdvanced

Detects infinite loops, evaluates parallel vs serial decisions.

Three-Layer Architecture

Layer 1: SKILL.md

Portable instructions. Works with any LLM agent.

Layer 2: Workspace

Shared files for cross-session state and memory.

Layer 3: Automation

Claude Code hooks for runtime enforcement.

Installation

GitHub (Recommended)

git clone https://github.com/live-neon/skills.git
cp skills/agentic/*/SKILL.md ~/.claude/skills/

Any LLM Agent

Copy the SKILL.md contents into your agent's system prompt. Compatible with Claude, GPT, Gemini, and any instruction-following model.