Risk-Aware Trading Agents
PYPI · RISK-AWARE-TRADING-AGENTS · 3 COMPONENTS
AI trading agents for US stocks: point-in-time news backtests, a frozen 2026 market, paper trading
Last scanned (12 hours ago)
Scores higher than 76% of PyPI servers
Category Finance and Payments
How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →
Supply Chain Security100
- No malware found by supply-chain analysis.Pass
- No known CVEs affecting this package version or its production dependencies.Pass
- Runs hatchling.build at install time, a recognised build step with no custom scripting around it. View diagnostics → Pass
- 5 of 68 dependencies flagged as unhealthy. View diagnostics → Partial
Provenance & Transparency100
- Source repository is publicly reachable at the declared URL. View diagnostics → Pass
- Cryptographically verified build provenance (signed, bound to jackson-cmd/risk-aware-trading-agents). View diagnostics → Pass
- Clear OSI-approved license (MIT).Pass
- Actively maintained (last published 6 days ago).Pass
- Publishes a security disclosure policy (SECURITY.md).Pass
Schema Quality & AI Usability55
- AI-judged instruction clarity (fair).Partial
- Context-footprint check failed: tool/resource definitions use about 2995 tokens (~136/item across 22 items; 22 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management27
- Stability observed for 8 of 30 days with no destabilising changes; credit accrues until the full window elapses.Partial
Tool Coverage71
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 0% of tool parameters carry a description.Fail
- Structured output schemas are declared (100% of tools); any adoption earns full credit.Pass
Tool Safety100
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- All 1 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.Pass
- An AI judge read all 23 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
- Implements a current MCP spec version (2026-07-28).Pass
Checks we ran (16)
- Malware scan
- Known CVEs
- Install scripts
- Dependency health
- Source repository
- Provenance
- License
- Maintenance
- Security disclosure
- Schema quality
- Stability
- Tool coverage
- Injection markers
- Destructive annotations
- Judged manipulation
- MCP protocol
Does not apply to this package (2)
- Endpoint SecurityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
- Transport & ReachabilityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
The score does not measure everything. What the score does not measure →
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 21 Nov 26 +15
- Malware scan: unverified → pass ▲ security
- 20 Nov 26 −14
- Malware scan: pass → unverified ▼ security
- 19 Nov 26 +15
- Malware scan: unverified → pass ▲ security
- 18 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 13 to 17. That category is still filling its 30-day observation window: 4 days of observed history at the previous scan, 5 at this one. The score rises as the window fills, whether or not the server changes.
- 17 Nov 26 −15
- Malware scan: pass → unverified ▼ security
- 15 Nov 26 +16
- Malware scan: unverified → pass ▲ security
- 14 Nov 26 0
- Malware scan: unverified → pass ▲ security
- Schema quality: 111 → 136 ▼ functional
- Stability: unverified → 0.03 ▲ functional
- Package version: 0.7.2 → 0.8.0 functional
- 13 Nov 26 61
First indexed and scored.
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 21 Nov 2026 · Analysed pypi/risk-aware-trading-agents@0.8.0
Provenance Verified
A signed build attestation was found and verified, binding this exact artifact to the source repository it claims to come from.
| Result | Verified |
|---|---|
| Ecosystem | pypi |
| Reason | Verified |
| Discovered via | Registry attestation endpoint |
| Source repo | jackson-cmd/risk-aware-trading-agents |
| Certificate issuer | https://token.actions.githubusercontent.com |
| Certificate SAN | https://github.com/jackson-cmd/risk-aware-trading-agents/.github/workflows/release.yml@refs/heads/main |
| Rekor log index | 3882047406 |
| Predicate type | PyPI publish attestation https://docs.pypi.org/attestations/publish/v1 |
| Subject digest | sha256:43a7635a988d59dfe28c47947a79da070e3aee731adf181e60d1a722b5849c13 |
Background: How many MCP packages publish verified provenance →
Install scripts 1 script
| Hook | Tier | Command |
|---|---|---|
| build_backend | Known build tool | hatchling.build |
Background: Why install scripts are a supply-chain risk →
Dependencies 68 packages
| Packages resolved | 68 |
|---|---|
| Stale | 4 |
| No linked repository | 1 |
| Tree resolution | Complete |
Background: SBOMs and build attestations, explained →
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
agent_decisions ~97
How an agent made its recent decisions: the setups and reasons, whether the risk manager let each one through, and the LLM's bull case, bear case and verdict when the agent has an LLM. include_prompts adds the exact prompt sent to the model and its raw reply.
| Name | Type | Req | Description |
|---|---|---|---|
| include_prompts | boolean | – | – |
| limit | integer | – | – |
| name | string | yes | – |
| symbol | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
agent_detail ~35
One agent: its settings, cash and positions at cost, equity by day and its latest orders.
| Name | Type | Req | Description |
|---|---|---|---|
| name | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
agents_create ~145
Create a paper-trading agent: a strategy with its own simulated account that trades every trading day. strategy is a built-in name or the path of a strategy saved by backtest_code (list_strategies shows both). llm optionally adds an LLM review board, e.g. "openai:gpt-5.6-luna", "ollama:qwen2.5:3b" or "off".
| Name | Type | Req | Description |
|---|---|---|---|
| cash | number | – | – |
| llm | – | – | – |
| name | string | yes | – |
| note | string | – | – |
| profile | string | – | – |
| strategy | string | – | – |
| symbols | – | – | – |
| universe | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
agents_leaderboard ~31
Every paper-trading agent ranked by return since it started, compared with SPY over the same days.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
agents_run ~39
Run agents once now on their simulated paper accounts (all active agents when names is empty).
| Name | Type | Req | Description |
|---|---|---|---|
| dry_run | boolean | – | – |
| names | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
analyze_stock ~35
What every agent thinks about one stock: technical setups, earnings quality, verified news, the decision.
| Name | Type | Req | Description |
|---|---|---|---|
| symbol | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
backtest ~151
Backtest a strategy on daily data (signals at the close, fills the next day) against SPY buy and hold. news=True replays the news archive: each day the strategy sees only items published before that day's close. The research headline files (2026 headlines and SEC filings for the built-in watchlists) are licensed and do not ship with the package; news comes from your local archive and from those files if you have a copy.
| Name | Type | Req | Description |
|---|---|---|---|
| cash | number | – | – |
| end | – | – | – |
| news | boolean | – | – |
| profile | string | – | – |
| start | string | – | – |
| strategy | string | – | – |
| symbols | – | – | – |
| universe | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
backtest_code ~243
Backtest a strategy written as Python code (one Strategy subclass; see strategy_guide). The stocks come from symbols or universe (megacap, tech, defensive, core, etf, sectors), never from the code. The code is checked first: dates or stock tickers written in the code, files, network and private attributes are refused, because they let a backtest see the future or touch the machine. It then runs in a separate process, where each day sees only that day's data and news published before the close. Returns metrics, SPY and buy-and-hold for comparison, trades and the risk manager's refusals; or the problems to fix, with line numbers. save_as='name' keeps the code in ~/.riskaware/strategies/ai/ for backtest and agents_create.
| Name | Type | Req | Description |
|---|---|---|---|
| cash | number | – | – |
| code | string | yes | – |
| compare | – | – | – |
| end | – | – | – |
| news | boolean | – | – |
| params | – | – | – |
| profile | string | – | – |
| save_as | – | – | – |
| start | string | – | – |
| symbols | – | – | – |
| universe | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
experiment_analysis ~206
What a saved experiment or a finished trading session made its money on. Decisions: the excess return over SPY 1, 5 and 20 trading days after every buy and sell (a sell counts the excess it avoided), with hit rates and t-statistics that allow for same-day decisions and overlapping windows, and the rank IC of each decision day against the next five days. Factors: daily returns regressed on the market, size, growth and momentum with Newey-West t-statistics, the alpha left after them and what each exposure contributed. P&L: by holding period, sector, market regime, news-triggered or not, and stock; the best and worst positions; how much of the profit rests on one or two trades. Writes analysis.json, analysis.md and analysis.png into the record. run_id is a run id or folder from list_experiments, or a trading session id.
| Name | Type | Req | Description |
|---|---|---|---|
| run_id | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
experiment_report ~34
The report (markdown, ready to paste), manifest and file list of one saved experiment.
| Name | Type | Req | Description |
|---|---|---|---|
| run_id | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
list_experiments ~53
Saved research experiments and trading sessions, newest first: strategy, period, return, Sharpe, SPY and the look-ahead verdict, with the folder of each record.
| Name | Type | Req | Description |
|---|---|---|---|
| limit | integer | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
list_strategies ~39
The built-in strategies that backtest and the agents can use, and the strategies saved from backtest_code (pass their path as the strategy).
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
market_day ~245
Everything known at the close of one day of the bundled periods (2018 to 2020, 2020 to 2022, 2024 to 2026-11-11; news in 2020, 2022 and 2026 when you have a copy of the headline files, which are not distributed): the market regime, VIX, VIX9D, VIX3M, VVIX and the term structure, fear & greed, breadth, credit stress, the macro events of the next two weeks; and per stock its prices and returns, indicators (EMAs, RSI, MACD, ATR, ADX, Bollinger, 52-week high), technical setups and the news published before the close (score, verified status, red flags, headlines, last earnings release). Nothing after that close is visible. Without symbols it lists all 64 stocks in a compact table; name up to 10 symbols for the full detail with headlines. A weekend or holiday shows the close before it.
| Name | Type | Req | Description |
|---|---|---|---|
| day | string | yes | – |
| headlines | integer | – | – |
| news | boolean | – | – |
| symbols | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
market_regime ~43
The market regime the agents see today (risk-on, neutral, uncertain, risk-off), the readings behind it, and the macro events of the next 30 days.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
research_guide ~131
Read this first for experiments on the frozen markets: what data ships with the package (three periods: 2026 to date, the 2022 bear market and the 2020 crash, each with daily bars and the VIX term structure; each year's headline file is licensed and not distributed, so news needs your own copy), how a backtest decides and fills (after the close, at the next open: swing trading), the look-ahead rules and check, how to write a strategy, the five baseline swing strategies, what a saved record holds, and advice for papers.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
run_experiment ~605
Run an experiment on a frozen period and save a record (see research_guide). period is 2026 (default, a rising year), 2022 (the rate-hike bear market) or 2020 (the COVID crash and rebound); start and end default to the period's test window. costs='fixed' (default) pays 5 bps on every fill at the open; costs='realistic' adds 10% of the overnight gap and square-root market impact and fills each order for at most 1% of its 20-day average volume. mode (plain execution) is long_only (default), long_short (short positions with a signed target_weight, e.g. Signal(sym, Action.SELL, target_weight=-0.1), long plus short at most 100% of equity, each short at most 20%, 0.5% a year borrow fee) or margin (Reg T: up to 200% gross, maintenance margin calls, borrowed cash pays BIL + 1.5%); mode_params overrides those rules. The rule-based long-short anchors are momentum_ls and trend_ls. Give either strategy (a built-in name such as swing_rsi2, swing_trend, swing_breakout, swing_pullback, swing_news, risk_aware, rsi2, momentum_rotation; or the path of a saved strategy) or code (one Strategy class, as in strategy_guide; it is checked and runs in a separate process). Signals at each close, market orders at the next open. execution='plain' trades exactly what the strategy says (equal-size positions, at most max_positions); 'risk_aware' adds the package's risk manager. split='2026-07-01' also reports the parts before and after. The leak check re-runs it with a random future after three cut days and compares the decisions. extra_news adds headlines ({ts, symbol, headline, source}) for counterfactual or robustness tests. Returns the metrics, SPY and the equal-weight universe, beta, alpha, t-statistic and bootstrap intervals, the leak-check verdict, the best and worst trades, and the folder of the record (manifest with data hashes, metrics, equity, trades, decisions, report.md, chart).
| Name | Type | Req | Description |
|---|---|---|---|
| code | – | – | – |
| costs | string | – | – |
| end | – | – | – |
| execution | string | – | – |
| extra_news | – | – | – |
| label | – | – | – |
| leak_check | boolean | – | – |
| max_positions | integer | – | – |
| mode | string | – | – |
| mode_params | – | – | – |
| news | boolean | – | – |
| params | – | – | – |
| period | string | – | – |
| position_size | – | – | – |
| split | – | – | – |
| start | – | – | – |
| strategy | – | – | – |
| symbols | – | – | – |
| universe | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
scan ~46
Today's buy and sell signals with their reasons. Scans the configured watchlist unless symbols are given.
| Name | Type | Req | Description |
|---|---|---|---|
| profile | string | – | – |
| strategy | string | – | – |
| symbols | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
stock_news ~82
Cross-verified news for one stock: a story counts when two independent outlets, the company's own press release or its SEC filing carry it. Only a severe flag the company itself backs (its 8-K, or its release confirming earlier reports) counts as a thesis break; denials clear the stories they refute.
| Name | Type | Req | Description |
|---|---|---|---|
| symbol | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
strategy_guide ~64
How to write a strategy for backtest_code: the Strategy class, what ctx holds (indicators, price history, positions, market regime, point-in-time news), the rules the code must follow, a template and a news example. Read it before writing code.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
trading_session_start ~264
Trade the frozen market yourself, one decision at a time (an LLM-agent experiment). Returns the rules and the first observation: the market, your portfolio, and per stock the main indicators and the news published before that close. Answer with trading_session_step. decide_every=5 decides once a week (about 44 steps for 2026); 1 decides every day. blind=True hides tickers, company names and dates so that you cannot recall what happened. Orders fill at the next open; long only; at most max_positions stocks. period is 2026 (default), 2022 or 2020. mode='long_short' also allows short positions (no leverage) and mode='margin' a Reg T margin account (up to 2x gross, margin calls); the returned rules say what each allows and costs.
| Name | Type | Req | Description |
|---|---|---|---|
| blind | boolean | – | – |
| cash | number | – | – |
| decide_every | integer | – | – |
| end | – | – | – |
| max_news | integer | – | – |
| max_positions | integer | – | – |
| mode | string | – | – |
| news | boolean | – | – |
| period | string | – | – |
| start | – | – | – |
| symbols | – | – | – |
| universe | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
trading_session_status ~36
Settings, steps sent so far and the record folder (when finished) of a trading session.
| Name | Type | Req | Description |
|---|---|---|---|
| session_id | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
trading_session_step ~151
Send the orders for the current step and get the next observation (or, after the last step, the result with its saved record). Each order: {"symbol", "action": "buy" | "sell" | "target", "weight" (share of equity for buy and target), "fraction" (share of the position for sell, 1 = all), "reason"}. An empty list holds. In a long_short or margin session also "short" (weight = size of the short) and "cover" (fraction of a short), and target takes a negative weight.
| Name | Type | Req | Description |
|---|---|---|---|
| note | – | – | – |
| orders | – | – | – |
| session_id | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
How do I install the Risk-Aware Trading Agents MCP server?
Risk-Aware Trading Agents runs locally as a PyPI package, launched with uvx risk-aware-trading-agents mcp. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.
pypi · risk-aware-trading-agents
claude mcp add jackson-cmd-risk-aware-trading-agents -- uvx risk-aware-trading-agents mcp
What is the Risk-Aware Trading Agents MCP server?
Risk-Aware Trading Agents is an MCP server listed in the public MCP registry as io.github.jackson-cmd/risk-aware-trading-agents. AI trading agents for US stocks: point-in-time news backtests, a frozen 2026 market, paper trading. This page covers its PyPI package (risk-aware-trading-agents).
Is the Risk-Aware Trading Agents MCP server safe to use?
Risk-Aware Trading Agents scores 79 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 21 November 2026. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the Risk-Aware Trading Agents MCP server expose?
Risk-Aware Trading Agents exposes 22 tools: market_regime, scan, analyze_stock, stock_news, list_strategies, and 17 more. Their descriptions and schemas cost roughly 2,775 tokens of context every time the server is loaded.
Is the Risk-Aware Trading Agents MCP server still maintained?
Risk-Aware Trading Agents is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.
What licence is the Risk-Aware Trading Agents MCP server under?
Risk-Aware Trading Agents declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.
What did VerifyMCP test on the Risk-Aware Trading Agents MCP server?
VerifyMCP ran 16 checks on Risk-Aware Trading Agents, covering Malware scan, Known CVEs, Install scripts, Dependency health and 12 more. Endpoint Security and Transport & Reachability do not apply to a PyPI package, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.