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 agents/prefecthq/marvin/claudegit clone --depth 1 https://github.com/PrefectHQ/marvinWhat 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 | $0.00000 | $0.00200 |
| Opus 5 | $0.00000 | $0.00100 |
| Sonnet 5 | $0.00000 | $0.00040 |
| Haiku 4.5 | $0.00000 | $0.00020 |
Grade A, and why
CLAUDE 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.
What it actually says
Marvin Agents
AI agents with tool access, memory, and autonomous decision-making capabilities.
Core Concepts
- Agent: Autonomous AI entity that can use tools and maintain memory
- Tool Integration: Seamless function calling with type safety
- Memory Persistence: Conversation history and context retention
- Decision Making: Goal-oriented task execution
Key Components
- Agent class with tool orchestration
- Memory management for context persistence
- Tool registration and execution framework
- Event handling for agent actions
Usage Patterns
from marvin import Agent
# Create agent with tools
agent = Agent(tools=[my_tool_fn])
# Run autonomous task
result = agent.run("Analyze the data and create a report")
Design Notes
- Agents are stateful and maintain conversation context
- Tools are regular Python functions with type hints
- Memory is automatically managed and persisted
- Agents can chain tool calls to accomplish complex tasks
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 · 36 lines · 0 tokens per session scan A 65ebf7b80b06
CLAUDE is an agent published in the GitHub repository PrefectHQ/marvin (6,192 stars, last pushed 11d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 200 tokens. 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 agents, from other repositories
dynamic-agents
Dynamic agents use functions instead of static values for instructions, model, and tools. These functions receive runtime context and return the appropriate configuration for each operation.
strategic-advisor
Activated for negotiation prep, deal analysis, interpersonal strategy, and high-stakes decision-making. Combines game theory with psychological awareness.
cleanup
You are an account cleanup agent for X/Twitter using the XActions MCP server.
fsl-vacuity-reviewer
Use PROACTIVELY after adding or changing a .fsl spec under specs/ or examples/. Uses the working-tree native Rust CLI to detect hollowing, weak mutation kill-rate, vacuous properties, and weakened invariants. Read-only on specs; may run verifier commands.
agent_class
The Agent class is the core orchestrator in AgentForge. It loads configuration, renders prompts, invokes the LLM, and produces final outputs. Agents can be subclassed for custom logic.
Plan
Research and outline multi-step plans for zen analysis improvements.