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/wiseaidotdev/autogpt/generic-gptgit clone --depth 1 https://github.com/wiseaidotdev/autogptWhat 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 | $0.01411 |
| Opus 5 | $0.00000 | $0.00705 |
| Sonnet 5 | $0.00000 | $0.00282 |
| Haiku 4.5 | $0.00000 | $0.00141 |
Grade A, and why
generic-gpt 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.
How it starts
The opening of the file, as written. The whole thing — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GenericGPT
GenericGPT is the autonomous, self-improving software engineering agent that powers AutoGPT's default interactive shell. It is the primary agent a developer interacts with when running autogpt with no subcommands.
Unlike the specialised agents (BackendGPT, FrontendGPT, etc.) that each own a single discipline, GenericGPT is a generalist: it reasons about its task, decomposes it into steps, writes or edits files surgically, runs the project's build system, and learns from the outcome, all autonomously.
Capabilities
| Capability | Description |
|---|---|
| Reasoning pre-step | Plans before acting; emits a structured internal monologue before every task |
| File tooling | CreateFile, WriteFile, PatchFile, ReadFile, AppendFile, ListDir, FindInFile |
| Shell execution | RunCommand runs arbitrary shell commands inside the workspace |
| Build-and-verify | Detects the build system (cargo, npm, python, ...) and iterates on errors |
| Git commits | GitCommit action commits changes with a generated message |
| Persistent skills | Learns lessons across sessions and injects them into future prompts |
| Session continuity | Full conversation history is restored when you resume a session |
| Multi-provider | Switches between Gemini, OpenAI, Anthropic, xAI, Cohere, and HuggingFace at runtime |
The .autogpt Directory
GenericGPT maintains a hidden directory at its state root (typically ~/.autogpt/, but controlled by the AUTOGPT_WORKSPACE environment variable) that stores all persistent state:
~/.autogpt/
├── sessions/ # Markdown conversation snapshots, auto-saved after every response
│ ├── <uuid>.md
│ └── ...
└── skills/ # Learned lessons (TOML)
├── rust.toml
├── web.toml
├── python.toml
└── ...
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.
- 2d ago First seen · 131 lines · 0 tokens per session scan A debfe6a245de
generic-gpt is an agent published in the GitHub repository wiseaidotdev/autogpt (115 stars, last pushed 2mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,411 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
medical-learning-safety-reviewer
审校医学学习草稿的来源、教育边界、患者隐私和输出风险；只返回审校意见，不处理外部动作或状态写入。.
user-guide-expert
Use this agent when you need to create, update, or improve user-facing documentation in the docs/userguide directory. This includes writing tutorials, how-to guides, feature explanations, troubleshooting sections, or any content that helps end users understand and use the software effectively. Examples: Context: User…
lessons-learned
Hard-won knowledge from building this codebase. When you make a mistake or discover a non-obvious behavior, add it here.
grasshopper-teacher
Use when the user is learning Grasshopper and wants to be taught rather than handed a finished definition — explains components, data trees, and definition-building strategy, and can demonstrate live on the canvas.
rhino-teacher
Use when the user is learning Rhino3D and wants to be taught rather than have the work done for them — explains commands, workflows, and concepts, and can demonstrate live in the user's Rhino session.
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.