Liquidity Order Flow Analyst (4.2)
Assesses slippage risk and order-book imbalance from depth data, giving crypto trading decisions liquidity-side evidence.
Read-only analytical agent (slippage/order-book imbalance/POC) with no fund movement or external side effects; repo-level SECURITY.md and AGPL-3.0 license exist, risks visible. Deducted: SKILL.md does not disclose NexusDataClient data flow, key handling or least-privilege details; persona admits falling back to 'inferred scores' when depth data is unavailable, potentially emitting unreliable estimates without explicit flagging; confirmation and rollback mechanisms undefined.
SKILL.md and persona.md are internally consistent (Tier-0 contract, JSON schema, threshold rules, few-shot examples), and README/CI/pytest indicate test infrastructure. Deducted: no dedicated tests found for 4.2's key paths; stub/inferred-score fallback on missing depth data weakens determinism and failure feedback; error-message quality on abnormal input is unevidenced.
Scenario is clear (crypto liquidity/slippage risk), query interface and trigger examples explicit, and the desk is consistently positioned in the README Tier-0 list. Deducted: no declared capability boundaries or non-fit ranges; no Chinese support; core dependency on Nexus depth data (overseas service, mainland-China reachability unverified), degrading to inference when unavailable.
Docs are basically layered and usable (SKILL.md + persona.md, output contract with schema_version tier0/v1); repo has LICENSE, CI and maintenance signals. Deducted: no per-skill version history/changelog/known-limitation disclosure; parameter naming stability and FAQ absent; publisher unverified, maintenance ownership unclear.
Output contract is concrete (slippage score, imbalance, POC price as JSON), offering some marginal value over manual order-book inspection. Deducted: static review cannot verify representative outputs are directly usable; only two illustrative few-shot examples; accuracy on real depth data and cost/benefit proportionality unevidenced.
Repository provides CI workflows, test suites and deterministic Nexus fixtures — auditable primary material. Deducted: fixtures only contain a quant_summary shape (order_imbalance/slippage/poc); no test cases targeting the 4.2 agent itself and no third-party reproduction evidence; facts vs. inference not separated in skill docs.
- When Nexus depth data is unavailable, agent 4.2 falls back to inferred/stub scores; reliability drops significantly and the degradation may not be clearly flagged — verify data provenance before live decisions.
- The skill depends on the overseas Nexus service; mainland-China reachability is unverified, and there is no Chinese documentation or output support.
- Static review found no dedicated tests for this skill; key-path reproduction is left to the user.
- The slippage threshold (≥80 → bear vote) is a hardcoded heuristic with limited adaptability and must not be treated as trading advice.
What does this skill do, and when should you use it?
This skill is one of 14 bundled in the olaxbt/ai-market-maker repository, located at src/agents/operator/4.2_liquidity_order_flow/. It reads order book snapshots from the Nexus depth endpoints (plus exchange L2 data when available) and, for a given ticker, returns slippage risk, order imbalance, and the POC price. When depth data is unavailable it falls back to stub estimators, so outputs may be rough estimates rather than real depth calculations. It follows a Tier-0 output contract, making it embeddable in the AIMM multi-agent trading system or callable directly via its query interface.
Calls NexusDataClient depth endpoints to fetch order book snapshots, falling back to stub estimators when no depth data exists; answers direct natural-language queries such as "What's the slippage risk on BTC?" and "Is the order book balanced?"; serves a query interface of the form /liquidity_order_flow?ticker=BTC/USDT returning slippage risk, order imbalance, and POC price; exposes an analyze(ticker, market_data, nexus_context) function under the Tier-0 contract.
- A crypto trader wants a quick read on whether slippage risk on a pair like BTC/USDT is acceptable before placing an order.
- A quant desk checks order book bid/ask imbalance before executing a large trade.
- An upstream agent in the AIMM multi-agent system invokes this operator skill as a pre-execution liquidity risk check.
- A researcher extracts POC price from order book structure as a market-microstructure signal.
What are this skill's strengths and limitations?
- Clean, minimal contract — one query interface and one analyze function.
- Defined fallback behavior when depth data is missing, so it doesn't hard-fail.
- Part of an AGPL-3.0 open-source repo, so logic is auditable.
- SKILL.md shows only the interface, no implementation or algorithm details.
- Depends on Nexus Skills API access; without Nexus you get stub estimates of uncertain reliability.
- No independent test evidence or accuracy data — the repo's example backtest figures belong to the whole system, not this skill.
- Examples are crypto-only (BTC/USDT); support for equities is not documented in the SKILL.md.
How do you install this skill?
The skill ships inside the ai-market-maker repository. Clone the collection: git clone https://github.com/olaxbt/ai-market-maker.git; the skill file lives at src/agents/operator/4.2_liquidity_order_flow/SKILL.md. OpenClaw installation is also supported for the collection: claw install https://github.com/olaxbt/ai-market-maker (the README documents no separate standalone install for this sub-skill).
How do you use this skill?
Two routes: 1) ask directly, e.g. "What's the slippage risk on BTC?" or "Is the order book balanced?"; 2) call the query interface /liquidity_order_flow?ticker=BTC/USDT, which returns slippage risk, order imbalance, and POC price. Prerequisite: NexusDataClient depth endpoints must be reachable (Nexus Skills API access is listed as optional); when depth data is unavailable, results come from stub estimators and should be treated cautiously.