Is stress-test safe to install?

SaferSkills independently audited stress-test (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.

CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

100

35.0 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 0 flagged

Securityscore 100 · 0 findings

✓—All security checks passedNo findings in this category for the latest scan.pass

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

.claude/skills/stress-test/SKILL.md · 1 file

.claude/skills/stress-test/SKILL.md5.8 KB · Markdown

name: stress-test

description: "Portfolio stress test. Receives a list of holdings and identifies portfolio weaknesses through shock sensitivity, scenario analysis, and causal chain analysis."

argument-hint: "[symbol list] [--scenario SCENARIO] e.g.: 7203.T,AAPL,D05.SI --scenario triple-meltdown"

allowed-tools: Bash(python3 *)

Portfolio Stress Test Skill

Parse $ARGUMENTS to determine the portfolio symbol list and scenario, then run the following command.

Execution Command

python3 /Users/kikuchihiroyuki/stock-skills/.claude/skills/stress-test/scripts/run_stress_test.py --portfolio <symbols> [--scenario <scenario>] [--weights <weights>]

Natural Language Routing

For natural language → skill selection, see .claude/rules/intent-routing.md.

Argument Parsing Rules

portfolio (symbol list — required)

Extract a comma-separated symbol list from user input. Convert space-separated input to comma-separated.

User Input Example	--portfolio Value
`7203.T,AAPL,D05.SI`	`7203.T,AAPL,D05.SI`
`7203.T AAPL D05.SI`	`7203.T,AAPL,D05.SI`
`Toyota Apple`	Convert to corresponding tickers first

weights (holding ratios — optional)

Comma-separated ratio list matching the number of symbols. Defaults to equal weights (1/N per symbol).

User Input Example	--weights Value
`0.5,0.3,0.2`	`0.5,0.3,0.2`
`50%,30%,20%`	`0.5,0.3,0.2` (convert percentages to decimals)
omitted	Equal weight 1/N per symbol

Scenario List

Scenario	Description
triple-meltdown	Equities, FX, and bonds fall simultaneously. Stress across all asset classes
usd-jpy-surge	Sharp JPY depreciation. Rising import costs; overseas assets appreciate in JPY terms
us-recession	US economic downturn. Global demand contraction, risk-off
boj-rate-hike	Rising Japanese interest rates. Banks up, growth stocks and REITs down
us-china-conflict	Escalating trade friction. Supply chain disruption; hit to semiconductors and manufacturing
inflation-resurgence	Rising prices again. Real purchasing power declines; rate hike expectations
tech-crash	NASDAQ -30%. AI expectations collapse; tech stocks hit hard, flight to quality
jpy-appreciation	USD/JPY -20 yen. Foreign asset JPY-denominated values fall; hit to exporters
custom	User-specified scenario interpreted from natural language

Output Format (10-Step Pipeline)

Present results in the following structured steps.

Step 1: Portfolio Overview

Symbol list (symbol, name, sector, weight)
Total market cap (estimated)

Step 2: Concentration Analysis

Sector HHI / Region HHI / Currency HHI
Identify highest concentration axis
Risk level judgment (diversified / somewhat concentrated / dangerously concentrated)

Step 3: Shock Sensitivity Scores

Assessment of each stock's beta, financial health, and valuation resilience
Per-symbol shock sensitivity score (0–100)

Step 4: Scenario Definition

Applied scenario name and description
Macro variable changes (assumed movements in interest rates, FX, equities)

Step 5: Per-Symbol Impact Estimates

Estimated loss rate per symbol
Application of concentration multiplier
Portfolio-weighted impact

Step 6: Portfolio-Wide Impact

Estimated portfolio-wide loss rate
Largest loss contributor

Step 6b: Correlation Analysis (KIK-352)

Cross-symbol correlation matrix (Pearson correlation, 1-year daily returns)
High-correlation pairs detection (|r| >= 0.7)
Factor decomposition: regress each symbol against macro variables (USD/JPY, Nikkei 225, S&P 500, crude oil, US 10-year yield)
LLM interpretation: For residual correlations not explained by factor regression, use domain knowledge to infer causes (e.g., supply chain dependencies, shared customer base). Clearly separate "confirmed factors (statistical)" from "inferred factors (estimated)"

Step 6c: VaR (Historical Data-Based Risk Metrics) (KIK-352)

Calculate portfolio weighted returns from 1-year daily return history
95% VaR / 99% VaR (daily and monthly)
Explain the difference from the scenario analysis (tail risk) in the stress test

Step 7: Causal Chain Analysis

Explanation of cascading effects when the scenario occurs
Cross-sector propagation paths

Step 8: Overall Assessment + Recommended Actions (KIK-352)

Specific risk mitigation proposals (rule-based auto-generated + Claude supplement)
Recommendations integrating concentration, correlation, VaR, and stress test results
Hedge candidates (symbols and sectors)
LLM supplement: In addition to rule-based recommendations, Claude should suggest sectors not in the portfolio and propose diversification targets informed by qualitative correlation causes

Execution Examples

# Basic stress test (scenario auto-detected)
python3 .../run_stress_test.py --portfolio 7203.T,AAPL,D05.SI

# Triple meltdown scenario
python3 .../run_stress_test.py --portfolio 7203.T,9984.T,6758.T --scenario triple-meltdown

# With weight specification
python3 .../run_stress_test.py --portfolio 7203.T,AAPL,D05.SI --weights 0.5,0.3,0.2

# Custom scenario
python3 .../run_stress_test.py --portfolio 7203.T,AAPL --scenario "semiconductor supply chain collapse"

Knowledge Integration Rules (KIK-466)

When get_context.py output contains the following, integrate with stress test results:

Previous stress test (StressTest): Compare with previous scenario and result. "Previous tech crash scenario: -18% → Current: -15% (improved: lower tech weighting)"
Concern notes: If a flagged symbol takes heavy damage in the stress test, "concern note matches → consider countermeasures"
Investment notes: Reference hedge strategy notes if available. "Previous lesson note: insufficient JPY hedge → still weak in USD surge scenario"

stress-test— agent skill

Is stress-test safe to install?

Portfolio Stress Test Skill

Execution Command

Natural Language Routing

Argument Parsing Rules

portfolio (symbol list — required)

weights (holding ratios — optional)

Scenario List

Output Format (10-Step Pipeline)

Step 1: Portfolio Overview

Step 2: Concentration Analysis

Step 3: Shock Sensitivity Scores

Step 4: Scenario Definition

Step 5: Per-Symbol Impact Estimates

Step 6: Portfolio-Wide Impact

Step 6b: Correlation Analysis (KIK-352)

Step 6c: VaR (Historical Data-Based Risk Metrics) (KIK-352)

Step 7: Causal Chain Analysis

Step 8: Overall Assessment + Recommended Actions (KIK-352)

Execution Examples

Knowledge Integration Rules (KIK-466)

Audit the pieces. Scan the whole. Decide.