Solana autonomous trading & 24/7 mode: expected setup vs pip 0.1.6 — clarification needed

[sergeykgoncharov-crypto]

Summary

We're evaluating AutoHedge for fully autonomous Solana trading as described in the README:

"Current support: Full autonomous trading on Solana"
"AutoHedge is an enterprise-grade autonomous agent hedge fund that trades on your behalf..."

After installing autohedge==0.1.6 and testing on Windows (Python 3.11, venv), we can run the interactive REPL and get useful LLM-based analysis (Director → handoffs → sub-agents). However, we could not find a documented path to autonomous, continuous Solana execution as advertised.

We'd appreciate official guidance: is Solana auto-trading production-ready, roadmap, or requires custom integration?

---

Our environment

• OS: Windows 10/11, VPS-capable
• Package: pip install -U autohedge → 0.1.6
• Entry point: autohedge → interactive REPL (manual prompt per run)
• .env configured: OPENAI_API_KEY, EXA_API_KEY
• Not yet configured: JUPITER_API_KEY, SOLANA_PRIVATE_KEY

---

What works (confirmed)

• CLI REPL starts and accepts task prompts
• Multi-agent pipeline (Trading-Director → Quant / Sentiment / Risk / Execution via handoffs)
• EXA_API_KEY → Sentiment-Agent exa_search executes successfully
• Jupiter-related code exists in autohedge/tools/ (jupiter_search, jupiter_price, ultra_tools: get_order, execute_trade, get_holdings)

---

What we could not find / reproduce

1. No 24/7 or autonomous loop in the shipped package

• cli.py is a manual REPL (while True: input()), not a trading daemon
• No scheduler, cron example, or documented systemd/Windows service for continuous operation
• example.py only runs a single AutoHedge().run(task=...) call

Question 1: Is there an official way to run AutoHedge 24/7 without human prompts? If yes, which command, script, or branch should we use?

---

2. Jupiter / Solana execution tools are not wired to agents

• tools_registry.get_tools() returns Jupiter tools (search_tokens, get_token_price, execute_trade, get_holdings, get_order)
• workers.py does not attach these tools to any agent
• Only exa_search is attached to Sentiment-Agent
• Execution-Agent has no tools=[...] — it only generates text orders (stock-style prompts: symbol, limit, stop-loss)

We never observed get_order or execute_trade being called from the default autohedge CLI flow.

Question 2: What is the intended integration path for Jupiter Ultra (get_order → execute_trade)? Should Execution-Agent receive get_tools() from tools_registry.py? Is there a Solana-specific worker or prompt we missed?

---

3. Environment variable mismatch

• .env.example lists WALLET_PRIVATE_KEY
• ultra_tools.py reads SOLANA_PRIVATE_KEY (base58)

Question 3: Which variable is correct for signing swaps? Will .env.example be updated?

---

4. README vs implementation gap

README states "Full autonomous trading on Solana". Our testing shows:

Claim	Our observation (0.1.6)
Autonomous trading	Manual REPL only
Solana execution	Tools exist, not connected to agents
Risk-first execution	Risk/Execution output is LLM text, no on-chain tx

Question 4: Should README wording be "Solana execution tools available (beta)" until agent wiring ships? Or is a newer release/branch expected soon?

---

5. experimental/ folder

GitHub repo contains experimental/market_making.py, btc_agent.py, etc. These are not in the PyPI package and market_making.py appears to use simulated orders and CEX price APIs (Coinbase/Binance), not Jupiter/Solana.

Question 5: Is experimental/ the intended path for autonomous trading today? Should users clone the repo instead of pip install autohedge for Solana?

---

6. Recommended Solana onboarding

Question 6: Could you publish a minimal official Solana quickstart, e.g.:

1. Required .env keys
2. Devnet vs mainnet recommendation
3. Exact prompt or script to trigger a small test swap
4. How to run unattended (if supported)
5. Risk limits / kill-switch recommendations

---

7. Roadmap

Question 7: Is Coinbase or true 24/7 Solana auto-trading planned for a specific release (version / ETA)? Should we wait for an upcoming release rather than patching workers.py locally?

---

Additional context (optional for maintainers)

We needed small patches locally (Swarms 13 tools_list_dictionary shared default, litellm_wrapper fallback when tool_calls is None) to get the stock-analysis REPL working reliably. Happy to contribute PRs if useful — but first we'd like to understand the intended Solana architecture from the maintainers.

---

What would help us most

• Official Solana setup doc or example script that ends with an on-chain transaction
• Confirmation whether 0.1.6 is analysis-only or execution-ready
• Pointer to the correct branch/release if Solana auto-trading is already implemented elsewhere

Thank you for the project and for clarifying the Solana execution path.

Environment: autohedge==0.1.6, Python 3.11, Windows, OpenAI + Exa configured.

---

Community support

We're seriously evaluating AutoHedge for production use and are willing to support the project financially (GitHub Sponsors / donation / paid tier — whichever you recommend) once we can verify that Solana autonomous execution works as documented, or once we have a clear official roadmap and timeline from the maintainers.

We're not asking for preferential treatment — just honest guidance on whether v0.1.6 matches the README claims today.