Skip to main content

🧪 Classified Agent — Synthesis Hackathon CLI

PyPI version Python 3.10+ License: MIT

Build + submit AI agents for Synthesis Hackathon ($75K prizes) in 60 seconds.

100% offline core. PyVax-powered Web3 scaffolding. Resilient state machine with automatic resume.


⚡ Quickstart (60 seconds to submission)

pip install classified-agent

classified-agent init
# → Creates classified.toml, agent.yaml, SKILL.md, workspace/, logs/, examples/

# Edit classified.toml (set API keys)
export ANTHROPIC_API_KEY='sk-ant-...'

classified-agent doctor
# → Verifies environment readiness (10 checks)

classified-agent run
# 🚀 Agent running!

🏗️ Architecture

classified_agent/
├── cli/             # Typer CLI: init, doctor, run, join-synthesis
├── config/          # Pydantic v2 models + TOML loader
├── core/            # Agent runtime (LLM, loop, context, memory)
├── tools/           # 14 built-in tools (fs, http, git, web3)
├── wallet/          # On-chain backends (PyVax local, mock, managed)
├── adapters/        # Synthesis.md hackathon integration
├── logging/         # Rich console + JSONL structured logs
├── templates/       # Scaffold assets for init (skills, examples, prompts)
├── examples/        # Example classified.toml configurations
└── tests/           # pytest test suite

🎯 CLI Commands

Command Description
classified-agent init Scaffold config + workspace + skills + examples
classified-agent doctor Verify environment readiness (10 checks)
classified-agent run Start the agent loop
classified-agent run --dry-run Simulate (no on-chain txs)
classified-agent run --verbose Debug-level logging
classified-agent join-synthesis --enable Join Synthesis hackathon mode

🔧 Configuration (classified.toml)

[agent]
name = "my-agent"
max_steps = 50
workspace_dir = "./workspace"

[llm]
provider = "anthropic"        # "anthropic" | "openai"
model = "claude-sonnet-4-20250514"
api_key_env = "ANTHROPIC_API_KEY"

[wallet]
backend = "pyvax_local"       # "pyvax_local" | "mock"
default_chain = "avalanche_fuji"

[wallet.policy]
max_native_per_tx = "0.1"     # AVAX per transaction
max_native_per_day = "1.0"    # daily spend cap

[synthesis]
enabled = false
track = "open"                # open | uniswap | base | lido

🛠️ Built-in Tools (14 + 3 Synthesis)

Category Tools
Filesystem fs_read, fs_write, fs_list
HTTP http_get (domain allowlist)
Git git_init, git_status, git_commit
PyVax pyvax_compile, pyvax_deploy, pyvax_call
Wallet wallet_get_balance, wallet_send_native, wallet_erc20_transfer, wallet_erc20_approve
Synthesis synthesis_load_skill, synthesis_register, synthesis_report_status

🔒 Safety & Resilience

  • Wallet Policy: Per-tx and daily spend caps, contract/method allowlists
  • Sandboxed FS: All file ops confined to workspace directory
  • HTTP Allowlist: Only whitelisted domains (synthesis.md, pyvax.xyz, GitHub, Avalanche)
  • Dry-Run Mode: --dry-run flag simulates all state-changing operations
  • API Resilience: Exponential backoff + retry for rate limits
  • State Checkpoints: Resume from exact failure point

🏆 Synthesis Hackathon Workflow

1. pip install classified-agent==1.2.1
2. classified-agent init
3. Set API keys in .env or export them
4. classified-agent doctor
5. classified-agent run  (or  join-synthesis --enable)
6. Agent reads SKILL.md → plans → acts → observes → done

📖 Skill Files

Project Classified bundles a SKILL.md that teaches agents how to operate:

  • Agent identity and capabilities
  • How to read classified.toml
  • How to use the wallet safely
  • Available tools and usage rules
  • How to log actions
  • How to execute missions

Copied to your project root during init. Edit to customise agent behaviour.

📦 Development

git clone https://github.com/ShahiTechnovation/classified-agent
cd classified-agent
pip install -e ".[dev]"
pytest classified_agent/tests/ -v

License

MIT — see LICENSE

Metadata

Release files for classified-agent 1.3.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for classified-agent 1.3.0
File Size Uploaded Provenance Actions
classified_agent-1.3.0.tar.gz 76.0 kB Without provenance Details for classified_agent-1.3.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for classified-agent 1.3.0
File Interpreter ABI Platform Actions
classified_agent-1.3.0-py3-none-any.whl Python 3 none any Details for classified_agent-1.3.0-py3-none-any.whl

Total release size: 170.4 kB

Release files / classified_agent-1.3.0.tar.gz

Download URL classified_agent-1.3.0.tar.gz
Size 76.0 kB
Tags Source
SHA-256 checksum
How to use checksums
3826c94e336d1cedb25e964c32f252d26eec4ef02cf042b213cf19c2ccb96bd1
BLAKE2b-256 checksum
How to use checksums
2da7222f4e9e6ab8ca76de180c2f7dd29c53dffa73662662a3467fffc8a3affe
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.12

Release files / classified_agent-1.3.0-py3-none-any.whl

Download URL classified_agent-1.3.0-py3-none-any.whl
Size 94.4 kB
Tags Python 3
SHA-256 checksum
How to use checksums
be2a7cce51bde343355111985cba4cbfe1e7e27a5c61d81f655e435d7db638fa
BLAKE2b-256 checksum
How to use checksums
73828be4f130c2e3ae87335c345e4627fef6d24947b1637aa23108a95ea0ca3c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.12

Security for classified-agent 1.3.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page