Ifrs9 Loss Engine

计算IFRS 9预期信用损失（ECL），支持Vasicek单因子前瞻性调整、Kaplan-Meier生存分析计算PD及贷款摊销计划生成，满足Basel III减值合规要求。

Security Scan

The skill claims to be an IFRS 9 ECL engine but its instructions mix in trading/backtest pipelines and host-level preconditions without declaring required credentials or an install; these mismatches warrant caution before installing or running it.

Purpose & Capability

The name/description state IFRS 9 ECL/regtech compliance, but the SKILL.md and human_summary repeatedly reference quant trading, ZVT backtests, data recorders, and a pipeline that includes 'trading_execution' and MACD-based use cases. That combination (regulatory provisioning + active trading/backtesting orchestration) is internally inconsistent: an IFRS9 provisioning engine should not need trading-execution semantics or locked MACD trading parameters. This looks like either mixing two different blueprints or an over-broad scope.

Instruction Scope

Runtime instructions require re-reading seed.yaml, enforcing many fatal 'semantic locks' (e.g., order of sell/buy execution, next-bar execution) and running precondition python commands that check/import zvt, check writable directories, and possibly run recorders. Those steps can cause the agent to execute host commands, interact with data providers, or prepare to execute trades. The SKILL.md effectively grants the agent discretion to run environment checks and pipeline actions beyond a narrow ECL calculation, which is scope creep relative to the stated purpose.

Install Mechanism

No install spec is provided (instruction-only), which is low risk from an installation download standpoint. However SKILL.md declares a runtime requirement (Python 3.12+ and the 'uv' package manager) but does not supply an install recipe. The missing install spec is a practical inconsistency—users/hosts will be asked to meet runtime dependencies with no guidance.

Credentials

The skill declares no required environment variables or credentials, yet the text asks about data sources (joinquant, qmt/broker, eastmoney) and references recorders and trading execution. Those data providers and brokers normally require API keys/credentials. The absence of declared env vars is an incoherence: either the skill will prompt for or attempt to use secrets at runtime, or callers will need to supply credentials ad-hoc. This mismatch increases risk because credential access is not made explicit or constrained.

Persistence & Privilege

The skill does not request 'always: true' and is user-invocable only. There is no install spec that would write arbitrary binaries to disk, and no declared behavior that it will persistently modify other skills or system-wide settings. Persistence/privilege requests appear proportionate.

What to consider before installing

This skill contains mixed signals: it claims to be an IFRS 9 ECL engine but also embeds quant-trading/backtest semantics and host-level preconditions. Before installing or running it: 1) Ask the author to clarify scope — is this purely an IFRS9/regtech model or also a trading/backtest toolkit? 2) Request an explicit install spec (how to install Python/uv deps) and a list of exact environment variables/credentials the skill will need (and where/how they'll be used). 3) If you plan to run it, do so in an isolated environment (sandbox/container) and inspect seed.yaml and referenced scripts first. 4) Ensure the skill cannot autonomously execute trades or access broker APIs without explicit, auditable user approval — semantic locks implying order execution are a red flag. 5) If you require regulatory-grade results, validate outputs against trusted implementations and review the referenced anti-patterns and preconditions (seed.yaml) for correctness. Refuse to provide credentials until the skill author explains why they are needed and how they are used/stored.

IFRS 9 损失引擎 (ifrs9-loss-engine)

计算IFRS 9预期信用损失（ECL），支持Vasicek单因子前瞻性调整、Kaplan-Meier生存分析计算PD及贷款摊销计划生成，满足Basel III减值合规要求。

Pipeline

data_collection -> data_storage -> factor_computation -> target_selection -> trading_execution -> visualization

Top Use Cases (42 total)

ECL Limit Level Truncation Analysis (UC-101)

Calculates Expected Credit Loss (ECL) at the limit/tranche level by computing remaining tenor and projecting loan balances with interest, supporting I Triggers: ECL, Expected Credit Loss, limit level

Loan Amortization Schedule Calculator (UC-102)

Computes loan amortization schedules by iteratively calculating interest amounts and remaining balances after each payment, determining total repaymen Triggers: amortization, loan, payment schedule

Amortization Schedule with NumPy Financial (UC-103)

Generates amortization schedules using numpy-financial library functions (PMT, PPMT, IPMT) for calculating periodic payments, principal, and interest Triggers: amortization, numpy-financial, PMT

For all 42 use cases, see references/USE_CASES.md.

Execute trigger: When user intent matches intent_router.uc_entries[].positive_terms AND user uses action verb (run/execute/跑/执行/backtest/fetch/collect)

What I'll Ask You

• Target market: A-share (default), HK, or crypto? (US stocks in ZVT are half-baked — stockus_nasdaq_AAPL exists but coverage is thin)
• Data source / provider: eastmoney (free, no account), joinquant (account+paid), baostock (free, good history), akshare, or qmt (broker)?
• Strategy type: MACD golden-cross, MA crossover, volume breakout, fundamental screen, or custom factor?
• Time range: start_timestamp and end_timestamp for backtest period
• Target entity IDs: specific stocks (stock_sh_600000) or index components (SZ1000)?

Semantic Locks (Fatal)

ID	Rule	On Violation
SL-01	Execute sell orders before buy orders in every trading cycle	halt
SL-02	Trading signals MUST use next-bar execution (no look-ahead)	halt
SL-03	Entity IDs MUST follow format entity_type_exchange_code	halt
SL-04	DataFrame index MUST be MultiIndex (entity_id, timestamp)	halt
SL-05	TradingSignal MUST have EXACTLY ONE of: position_pct, order_money, order_amount	halt
SL-06	filter_result column semantics: True=BUY, False=SELL, None/NaN=NO ACTION	halt
SL-07	Transformer MUST run BEFORE Accumulator in factor pipeline	halt
SL-08	MACD parameters locked: fast=12, slow=26, signal=9	halt

Full lock definitions: references/LOCKS.md

Top Anti-Patterns (15 total)

• AP-REGTECH-001: Missing attribute initialization on data structures
• AP-REGTECH-002: Self-loops in transaction graphs violate domain rules
• AP-REGTECH-003: Unvalidated floating-point inputs cause runtime crashes

All 15 anti-patterns: references/ANTI_PATTERNS.md

Evidence Quality Notice

[QUALITY NOTICE] This crystal was compiled from blueprint finance-bp-062. Evidence verify ratio = 80.0% and audit fail total = 15. Generated results may have uncaptured requirement gaps. Verify critical decisions against source files (LATEST.yaml / LATEST.jsonl).

Reference Files

File	Contents	When to Load
references/seed.yaml	V6+ 全量权威 (source-of-truth)	有行为/决策争议时必读
references/ANTI_PATTERNS.md	15 条跨项目反模式	开始实现前
references/WISDOM.md	跨项目精华借鉴	架构决策时
references/CONSTRAINTS.md	domain + fatal 约束	规则冲突时
references/USE_CASES.md	全量 KUC-* 业务场景	需要完整示例时
references/LOCKS.md	SL-* + preconditions + hints	生成回测/交易代码前
references/COMPONENTS.md	AST 组件地图（按 module 拆分）	查 API 时

---

Compiled by Doramagic crystal-compilation-v6.1 from finance-bp-062 blueprint at 2026-04-22T13:00:19.657886+00:00. See human_summary.md for non-technical overview.