Skip to content

tywinlu1988/Credence-China / version/v0.12.3-release/.claude/skills/fixed-income-credit-analysis

fixed-income-credit-analysis

Verified

Use when analyzing industries or companies for credit decisions in Chinese fixed income markets, building industry analysis frameworks for lending or bond investment, evaluating credit quality via dual-track methodology, constructing investment dashboards from public data, validating frameworks against historical defaults, or assessing cross-industry contagion, portfolio concentration, and systemic risk via the system-intelligence layer. Route vague needs to the credit-analysis-router skill.

references/mosaic-engine-architecture.md

View on GitHub
# Mosaic Engine Architecture (Mode A — Primary)

**版本**: v0.12.3-release

> 本文自 `fixed-income-credit-analysis` SKILL.md 下沉而来,组织马赛克引擎(Mode A)的详述参考。阈值/规则的单一事实源为 `engine/mosaic-engine.md`;本文件仅作导航与组织,不引入新数值。

## Signal Confidence Levels

| Level | Type | Meaning |
|---|---|---|
| L5 | Multi-source cross-validated | Same fact confirmed by >=2 independent sources |
| L4 | Single-source direct | One reliable source explicitly states |
| L3 | Derived/inferred | Assembled from multiple fragments through reasoning |
| L2 | Directional weak | Direction known, magnitude unknown |
| L1 | Missing (gap) | Data that should exist but is absent — risk signal |

## Signal Density Assessment

For each analysis dimension, compute signal density = obtained signals / expected signals.

| Density | Meaning | Impact on Scoring |
|---|---|---|
| >80% | Sufficient | High confidence, direct scoring |
| 50-80% | Moderate | Medium confidence, score +/-1 interval |
| 20-50% | Insufficient | Low confidence, directional only, mark "needs supplement" |
| <20% | Severely lacking | Cannot score, mark "insufficient data" |

## Mandatory Density Rules

- If **any critical dimension** (L1 for the industry type, or any dimension the user explicitly asks about) has signal density **<20%**, you MUST NOT output a numeric score for that dimension. State `信息不足无法评估` and list the missing signals.
- If the **weighted-average signal density across all scored dimensions** is **<50%**, you MUST NOT output a final letter rating. Output a qualitative directional assessment plus a prioritized gap list instead.
- If density is 50-80%, you MAY output a rating but MUST label it as `中置信度` and widen the implied interval by ±1 notch.
- The completeness report is mandatory for every analysis; omitting it is a protocol violation.

Before finalizing any numeric rating, verify:
1. Every dimension used in the score has a documented density.
2. No critical dimension is below 20% density.
3. The final rating maps to the official 18-notch table.
4. Every veto condition has been explicitly checked.
If any check fails, downgrade the rating or replace it with a directional statement.

## Data Gap-to-Risk Mapping

Common gaps and their risk implications:

| Gap Type | Typical Missing Data | Information Risk | Substitute Signal |
|---|---|---|---|
| Competitive data | Exact cost comparison across firms | Over/under-estimate cost advantage | SOE bidding prices (market-accepted premium) |
| Financial data | Non-listed company financials | Cannot assess financial health | Enforcement records, judicial disputes, hiring trends |
| Market pricing | No stock/bond data for non-listed firms | Track B completely unavailable | Public financing events, equity transfer prices |
| Terms data | Duration/Z-spread (no Wind terminal) | Cannot do precise rate risk analysis | YTM approximation, same-rating spread comparison |
| Liquidity data | Bid-ask spread (not disclosed in China) | Cannot assess true trading cost | Daily volume, turnover rate, abnormal volume events |

## Completeness Report Output

Every analysis MUST include a completeness report structured as:
1. Per-dimension signal density bar
2. Gaps list (prioritized by impact on conclusion)
3. Gap-to-risk mapping
4. Confidence interval on scores
5. User-actionable suggestion (e.g., "If you need precise relative value ranking, connect Wind/Choice via Mode B")
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 323kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 323kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 323kUpdated Claude CodeCodexVerifiedInstall