A community-run Flowchain AI Agent that can act on-chain.
This 2026 product vision turns FLC into AI Agent gas: users fund tasks with FLC, the agent executes under strict guardrails, and activity is metered, auditable, and stoppable. The agent can perform approved on-chain actions on Base—like swapping FLC on Uniswap, adding liquidity, and other community-governed workflows.
中文摘要：2026 產品主軸是「社群營運的 Flowchain AI Agent」。使用者用 FLC 付費（gas / usage budget）啟動任務， Agent 依照政策與限額執行（可 swap / add liquidity 等），並留下可稽核的紀錄與可中止機制。
Complete product plan: Flowchain Community AI Agent
Think of this as an “AI operations layer” for the community: a shared agent that can run tasks for users, while the community defines rules, limits, and allowed actions. The key design is budgeted autonomy: the agent can act, but only within explicit constraints.
What users get
- Submit a task (“swap X”, “add liquidity”, “rebalance”, “monitor”, “report”).
- Pay task gas in FLC (budget + max fee + time limit).
- Get a verifiable receipt: actions taken, transaction hashes, and itemized gas usage.
What the agent is
- A goal-driven runner: plan → tool calls → execution → verification.
- Policy-driven: allowlists, caps, rate limits, and approval gates.
- Observable: logs, traces, and post-action validation.
What the community controls
- Which actions are enabled (e.g., swaps, LP add/remove, routing).
- Safety parameters (max slippage, max value, cooldowns).
- Upgrades and incident response processes.
FLC Gas model (how users pay)
Users fund tasks with FLC. The system meters agent work (AI calls + tool calls + on-chain actions), enforces caps, then returns a receipt. This makes cost predictable and prevents runaway autonomy.
Task Budget
Each task includes a user-defined budget: max FLC spend, max duration, and max on-chain value. If any cap is reached, the task stops safely.
Receipts & Refunds
Unspent FLC from a task budget can be returned according to community policy. Every task emits a receipt with itemized costs and transaction hashes.
How it works (end-to-end)
A production-style pipeline designed for agent safety: verify → simulate → execute → validate → publish receipt.
1Intent & quote
Agent parses the user goal, proposes a plan, and returns a quote: estimated FLC gas and required approvals.
2Policy check
Enforce allowlists, caps (slippage/value/tx count), and risk rules. Reject unsafe tasks early.
3Simulation
Simulate the trade/LP action before sending: expected output, worst-case under slippage cap.
4Execution & validation
Send transactions, confirm outcomes, and validate results (amounts, pool position, receipts).
Agent capabilities (community-governed)
The agent’s actions are modular. The community can enable/disable modules and set constraints. Below are examples of what can be supported on Base.
Swap FLC on Uniswap
Execute swaps via approved routers with strict parameters: token allowlist, max slippage, max size, and a simulation-first rule.
- Allowed routes only (e.g., FLC ↔ WETH / USDC if enabled).
- Hard cap on slippage and per-tx value.
- Post-trade validation + receipt.
Add liquidity on Uniswap
Add liquidity only under explicit constraints: approved pools, min/max ranges (if applicable), and value caps.
- Pool allowlist + min liquidity thresholds.
- Position limits: per-task and per-user.
- Risk disclosure + confirmations for advanced modes.
“etc.” modules
Additional modules can be added if the community approves and safety criteria are met.
- Monitoring / alerts (prices, pool health, exposures).
- Batch operations (scheduled, capped, auditable).
- Reporting (receipts, summaries, transparency pages).
Safety guardrails (non-negotiable)
A community agent must be safer than a typical trading bot. Guardrails protect users, the community, and the system itself.
Permissioning
- Token allowlist + contract allowlist (routers, pools).
- Least-privilege: approve only what is needed per task.
- Role separation: policy admin ≠ execution keys.
Limits & circuit breakers
- Max slippage, max value, max tx count, max duration.
- Per-user caps + cooldowns (anti-spam).
- Emergency stop button (pause all execution).
Transparency
- Receipts with tx hashes and itemized gas usage.
- Deterministic policy versioning (policy hash).
- Public status page + incident reports (when needed).
Human-in-the-loop (recommended)
For higher-risk actions (large swaps, LP changes, new pools), require confirmation. “Auto-mode” is only for low-risk, well-defined tasks.
MEV / price movement awareness
The agent should prefer protected routing where available, enforce tight slippage, and abort when quotes change beyond policy thresholds.
Community governance model
This agent is “community run” in a practical sense: governance defines policy, upgrades, and incidents. Execution is rule-bound, audited, and reversible where possible.
Policy proposals
Community votes/approves changes to:
- Allowed actions & modules
- Allowlisted contracts / pools
- Default caps (slippage/value/time)
Operations council (optional)
A small, rotating group can manage incident response under strict transparency rules (e.g., emergency pause + public postmortem).
Audits & upgrades
Upgrades should be staged: testnet → limited rollout → full rollout. Publish policy hashes and version history.
Verify official information
Use Base contract verification as your anchor. Only trust information published on this domain and the official Base token contract page.
Base token contract
basescan.org/token/0x320f7aaeefec65f29b7e4509739ff9c62ded9803
Beware spoofed domains and look-alike addresses.
Contact
Business inquiries: hello@flowchaincoin.org
We will never request private keys, seed phrases, or remote “screen share” access.
FAQ
Is this a trading bot?
It’s broader: a community-run agent that can perform defined tasks (including on-chain actions) under policy, limits, and receipts. No guaranteed outcomes.
Who holds the funds?
Product design should minimize custody risk: budgeted deposits, limited approvals, and strict caps. Avoid “one hot wallet with everything”.
What does “etc.” include?
Only what the community approves and what can be made safe: monitoring, reporting, automation, and other tool-driven workflows—always with guardrails.
中文：這是一個「可控、可稽核、可中止」的社群 Agent 產品藍圖。 任何 swap / LP 行為都有風險；我們用預算、限額、白名單與人為確認機制把風險關進護欄。