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.
npx agentmods add instructions/codingagentsystem/cas/claude-mdgit clone --depth 1 https://github.com/codingagentsystem/casWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.02157 | $0.02157 |
| Opus 5 | $0.01078 | $0.01078 |
| Sonnet 5 | $0.00431 | $0.00431 |
| Haiku 4.5 | $0.00216 | $0.00216 |
Grade A, and why
cas CLAUDE.md 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 271 lines — stays where its author put it; the contents beside it link to each section on GitHub.
IMPORTANT: USE CAS FOR TASK AND MEMORY MANAGEMENT
DO NOT USE BUILT-IN TOOLS (TodoWrite, EnterPlanMode) FOR TASK TRACKING.
Use CAS MCP tools instead:
mcp__cas__taskwith action: create - Create tasks (NOT TodoWrite)mcp__cas__taskwith action: start/close - Manage task statusmcp__cas__taskwith action: ready - See ready tasksmcp__cas__memorywith action: remember - Store memories and learningsmcp__cas__searchwith action: search - Search all context
CAS provides persistent context across sessions. Built-in tools are ephemeral.
CAS - Coding Agent System
Unified context system for AI agents: persistent memory, tasks, rules, and skills across sessions.
Design Philosophy: CAS is built for AI agents as the primary users. Humans are observers who review agent activity, provide feedback, and guide direction - but the tools and workflows are optimized for agent consumption and production.
Use CAS for Task & Memory Management
Agents use CAS MCP tools instead of built-in TodoWrite:
mcp__cas__task action=create - Create/track tasks
mcp__cas__task action=start - Start working on a task (sets status to in_progress)
mcp__cas__task action=notes - Add progress notes (progress, blocker, decision, discovery)
mcp__cas__task action=close - Complete tasks
mcp__cas__memory action=remember - Store learnings and context
mcp__cas__search action=search - Find relevant context (filter with doc_type: entry/task/rule/skill)
Human CLI:
cas init # Initialize CAS in a project
cas serve # Start MCP server
cas config list # View configuration
cas doctor # Run diagnostics
cas update # Self-update
Project Structure
cas-cli/ # Rust CLI & MCP server (primary)
crates/ # Workspace crates (cas-core, cas-store, cas-search, etc.)
cas-cli Architecture
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.
- yesterday First seen · 271 lines · 2,157 tokens per session scan A e73b9d5e8319
cas CLAUDE.md is an instructions file published in the GitHub repository codingagentsystem/cas (151 stars, last pushed 5mo ago), licensed MIT. It adds 2,157 tokens to every session, about $0.0108 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 2026-08-30.
Other instructions, from other repositories
guardian-agent AGENTS.md
Instructions for Threat-Vector-Security/guardian-agent, covering repository guidelines, branching (critical), project structure & module organization, build, test, and development commands and runtime restart handoff (critical).
guardian-agent CLAUDE.md
Instructions for Threat-Vector-Security/guardian-agent, covering repository guidelines, branching (critical), project structure & module organization, build, test, and development commands and intent gateway (critical).
guardian-agent GEMINI.md
Instructions for Threat-Vector-Security/guardian-agent, covering repository guidelines, branching (critical), project structure & module organization, build, test, and development commands and intent gateway (critical).
hatch3r CLAUDE.md
Instructions for hatch3r/hatch3r, covering hatch3r — development instructions, architecture, development commands, two-axis pillar framework (2.0.0) and orchestrator self-discipline (bypass protection).
gemini-kit GEMINI.md
Instructions for nth5693/gemini-kit, covering gemini-kit: super engineer team, role & responsibilities, workflows, team members and workflow.
ai-assistant-instructions CLAUDE.md
Instructions for dryvist/ai-assistant-instructions: Canonical agent instructions live in AGENTS.md and agentsmd/rules/. Nix-managed machines auto-load both globally via nix-ai, so this file deliberately does not re-import them — doing so would load every byte twice in each session. Non-Nix contributors: read AGENTS.md…