Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add austin-starks/Public-Portfolio-Challenge --skill bug-protocolgit clone --depth 1 https://github.com/austin-starks/Public-Portfolio-ChallengeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/austin-starks/public-portfolio-challenge/bug-protocol)<a href="https://agentmods.dev/skills/austin-starks/public-portfolio-challenge/bug-protocol"><img src="https://agentmods.dev/badge/skills/austin-starks/public-portfolio-challenge/bug-protocol/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/austin-starks/public-portfolio-challenge/bug-protocol"><img src="https://agentmods.dev/badge/skills/austin-starks/public-portfolio-challenge/bug-protocol.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00103 | $0.00890 |
| Opus 5 | $0.00051 | $0.00445 |
| Sonnet 5 | $0.00021 | $0.00178 |
| Haiku 4.5 | $0.00010 | $0.00089 |
Grade A, and why
bug-protocol scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Bug Protocol — "Loudly Declare"
The compute + engine stack is new and moves. Every failure gets this protocol. Never paper over an error to keep a campaign moving — a swallowed bug is worse than a blocked stage. A confabulated mechanism (inventing why the engine did something) is the worst failure mode: the campaign's worst moments have been confabulated reconcile events, pricing behavior, and budget behavior. Verify, don't assert.
Precedent: a totalBudget over-deployment bug (a book with a 40% budget deployed ~96% of NAV at cold
start) was caught exactly this way and invalidated weeks of results. The general lesson is to
quarantine every result affected by an engine defect and rerun only after the behavior is corrected.
The five steps (apply on every failure)
- Stop and characterize — is it config (my fault), data (source-side), or engine (NexusTrade)? Reproduce minimally — the smallest job/backtest that still shows it.
- Log it immediately in the run's
CAMPAIGN_LOG.mdunder a⚠️ BUG/ISSUEheading: symptom, repro IDs, expected-vs-observed, current hypothesis (labeled hypothesis), what it blocks. - If it's a real engine/data bug, write a hand-off doc
<NAME>_BUG.mdin the run directory, modeled on the template inreferences/BUG_TEMPLATE.md(one-line summary, expected vs observed, reproduction with IDs, leading hypothesis, alternatives to rule out, where to look, impact on results). - Quarantine affected results — mark every backtest/study that ran before the fix as suspect/invalid. Re-run after the fix. If the bug inflated leverage, both returns AND drawdowns are inflated and any sweep gene it touched may be non-differentiating — re-validate all of it.
- Tell the human, prominently, in the chat. Bugs are a co-equal deliverable of the episode with the trading result — not an embarrassment to route around.
What is NOT a bug (honest-signal failures — don't file these)
- A fold returning
NoEligibleCandidate= an honest signal, not a bug. Widen validation; don't drop the certification floors. - A dead / not-yet-listed name (e.g. SNDK in 2022) being absent or in a reject list with no fabricated values = correct behavior. The engine fabricating a price/indicator value for a name that didn't trade IS a bug (see engine-sanity).
- A single-tick reconcile target = the live strategy accretes the rest over daily rebalances — say so, don't re-diagnose it as a bug (see deploy-gate).
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 60 lines · 103 tokens per session scan A 32c8e5400a1f
bug-protocol is a skill published in the GitHub repository austin-starks/Public-Portfolio-Challenge (43 stars, last pushed 2d ago), licensed MIT. It adds 103 tokens to every session and 890 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
backtest-diagnose
Diagnose failed or underperforming backtests, locate the root cause, and fix the issue.
tushare
A Python interface for Tushare, a financial data service that provides market and company information for stocks, funds, futures, and digital assets. It returns queried data as pandas tables.
social-media-intelligence
Social media intelligence: financial signal extraction from Twitter/X, Telegram, Discord, and Reddit for sentiment-driven trading strategies.
ashare-pre-st-filter
An A-share China stock risk checker that forecasts whether a company may receive an ST or *ST warning in the next financial year. ST labels are Chinese exchange warnings for companies facing specified financial or regulatory problems.
correlation-analysis
Correlation and cointegration analysis — co-movement discovery, deep return-correlation analysis, sector clustering, realized correlation, Engle-Granger / Johansen cointegration, half-life, Kalman dynamic hedge ratio, cross-market linkage analysis, and pair-trading signal generation.
credit-analysis
A guide to analysing bonds and other fixed-income investments, including issuer credit quality, interest payments, default risk, credit spreads, and convertible bonds. It also covers Chinese fixed-income markets and local-government financing bonds.