Cm
Install the CASS Memory System (cm) CLI skill to give your coding agent procedural memory with decay, anti-patterns, and safe cross-agent knowledge transfer.
Overview
cm (CASS Memory System) is an agent skill most often used in Build (also Ship, Operate) that gives AI coding agents procedural memory via a Bun/TypeScript CLI with decay, anti-patterns, and a trauma guard safety layer.
Install
npx skills add https://github.com/Dicklesworthstone/cass_memory_system --skill SKILL.mdWhat is this skill?
- Three-layer cognitive architecture for procedural memory tailored to AI coding agents
- Confidence decay so stale habits surface less often unless reinforced by outcomes
- Anti-pattern learning to block repeats of known bad fixes or risky commands
- Cross-agent knowledge transfer so lessons from one session or agent profile help another
- Trauma guard safety system to gate hazardous procedural memories before they execute
- Trauma guard safety system for hazardous procedural recalls
- Bun/TypeScript CLI packaging
Adoption & trust: 376 GitHub stars.
What problem does it solve?
Your coding agent forgets what worked last sprint and keeps suggesting fixes you already proved were wrong, because nothing persists as graded procedural memory.
Who is it for?
Solo builders running Claude Code, Cursor, or Codex daily who want local procedural memory with explicit safety framing around hazardous habits.
Skip if: Builders who refuse any local CLI touching agent memory stores, or teams that need a managed cloud memory product with no self-hosted Bun runtime.
When should I use this skill?
When you want persistent procedural memory, anti-pattern learning, and cross-agent knowledge transfer for AI coding agents via the CASS Memory System CLI.
What do I get? / Deliverables
You run the cm CLI so agents retain layered procedures, down-rank stale confidence, block anti-patterns, and share vetted knowledge across agents with safety gates before risky recalls.
- Persistent procedural memory records with confidence decay
- Anti-pattern entries and cross-agent transferable knowledge
- Safety-gated recall behavior via trauma guard rules
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Procedural memory is wired during Build as agent infrastructure, then pays off whenever you debug, review, or iterate because learned habits persist outside a single chat. Agent-tooling is the right canonical shelf for a Bun/TypeScript CLI that extends how agents remember and avoid repeated mistakes.
Where it fits
You install the cm CLI so your agent records successful refactor steps as layered procedural memory for this repo.
After review catches a bad migration pattern, you store it as an anti-pattern so the agent stops proposing the same shortcut.
A production rollback teaches a new safe deploy procedure and trauma guard blocks recall of the previous risky command sequence.
How it compares
A procedural-memory CLI skill package—not a generic notes MCP server and not a one-shot brainstorming workflow.
Common Questions / FAQ
Who is cm (CASS Memory System) for?
Indie and solo developers who treat AI agents as long-lived pair programmers and need durable, safety-aware procedural memory beyond chat transcripts.
When should I use cm?
During Build while wiring agent-tooling and memory stores; during Ship when code review should feed anti-patterns back into memory; during Operate when incidents should update procedures and trauma guard should block unsafe recalls.
Is cm safe to install?
Trauma guard is designed to reduce hazardous replays, but you should still review the Security Audits panel on this Prism page and audit what filesystem and shell permissions you grant the Bun CLI.
SKILL.md
READMESKILL.md - Cm
CASS Memory System - procedural memory for AI coding agents. Three-layer cognitive architecture with confidence decay, anti-pattern learning, cross-agent knowledge transfer, trauma guard safety system. Bun/TypeScript CLI. # cm { "name": "cm", "description": "CASS Memory System - procedural memory for AI coding agents. Three-layer cognitive architecture with confidence decay, anti-pattern learning, cross-agent knowledge transfer, trauma guard safety system. Bun/TypeScript CLI." }