What does this skill do, and when should you use it?
News Narrative Miner is one of 14 OpenClaw skills bundled in olaxbt/ai-market-maker (AIMM, an open-source agentic AI hedge fund OS), located at src/agents/operator/1.2_news_narrative_miner/. It takes a single ticker (e.g. BTC/USDT), pulls news via the Nexus News API, analyzes whether breaking news impacts the asset, identifies the current narrative phase, and maintains an in-memory news decay timeline. Queries go through /news_narrative_miner?ticker=BTC/USDT and return an impact score, event type, decay, and reasoning. When Nexus is disabled it falls back to neutral stubs rather than failing. The host repo is AGPL-3.0 licensed and targets Python 3.11+ with uv-managed dependencies.
Reads news data for a given ticker through the Nexus News API (via the NexusDataClient); analyzes breaking events (e.g. "Any breaking news impacting BTC?") and produces an impact score and event type; identifies the current narrative phase ("What's the current narrative phase?"); maintains an in-memory news decay timeline tracking how event influence fades over time; exposes a Tier-0 contract analyze(ticker, market_data, nexus_context) and serves the query endpoint /news_narrative_miner?ticker=BTC/USDT, which returns impact score + event type + decay + reasoning. When Nexus is disabled (NEXUS_DISABLE) it returns neutral stub data so the pipeline keeps running.
- A crypto fund researcher with an open BTC position wants a quick, scored answer on whether breaking news threatens the trade
- A quant team running a narrative-trading strategy needs a programmatic signal for which phase of the narrative cycle an asset is in, to time entries and exits
- An AIMM pipeline operator needs an upstream news-signal node feeding narrative context into downstream alpha synthesis and risk checks
- A developer running backtests or demos offline (Nexus disabled, no network) needs the news node to degrade gracefully to neutral stubs instead of blocking the workflow
- An engineer evaluating the AI trading stack before deployment wants to inspect this single skill's data dependencies and interface contract in isolation
How do you install this skill?
- Core data depends on the overseas Nexus News API; mainland-China reachability is unverified. Confirm network access and quota (Nexus credits settle on BNB Chain) before use.
- Not executed in this review: impact scoring, decay modeling and black-swan triggering are static inferences; do not use directly for live trading decisions.
- Publisher identity is unverified by the FollowSkills registry, and the skill lacks versioning/changelog; diff behavior manually on upgrades.
- When Nexus is disabled the skill silently falls back to neutral stubs, which may return uninformative results; explicitly detect the fallback state downstream.
- Shell / CLI
- Network access
- Local filesystem
NexusDataClient news endpoints (Nexus Skills API access, optional — falls back to neutral stubs when Nexus is disabled)
The skill ships inside the monorepo: git clone https://github.com/olaxbt/ai-market-maker.git and cd ai-market-maker; install uv (pip install uv); run uv sync --extra dev (the full platform stack additionally needs TA-Lib and Docker, see the repo's Quick Start — the minimal dependency set for running this skill alone is not documented); alternatively install via OpenClaw with claw install https://github.com/olaxbt/ai-market-maker. The skill file lives at src/agents/operator/1.2_news_narrative_miner/SKILL.md. Note: the source does not document how to install a single skill folder from this monorepo into an arbitrary Agent Skills client, nor any skill-specific environment variables.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/olaxbt/ai-market-maker.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/src/agents/operator/1.2_news_narrative_miner" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Any breaking news impacting BTC?
- What's the current narrative phase?
Query it in natural language, e.g. "Any breaking news impacting BTC?" or "What's the current narrative phase?". Programmatically, hit the query interface /news_narrative_miner?ticker=BTC/USDT to get impact score, event type, decay, and reasoning. You can also call it in code via the Tier-0 contract analyze(ticker, market_data, nexus_context). Configure Nexus Skills API access in .env (optional); if it is missing or NEXUS_DISABLE=1 is set, the skill returns neutral stub results. The recommended full-stack run is docker compose -f docker-compose.prod.yml up --build -d with the dashboard at http://localhost:3000.
What are this skill's strengths and limitations?
- Clear interface contract: the Tier-0 analyze signature and the query response fields (impact score, event type, decay, reasoning) are documented
- Graceful degradation: falls back to neutral stubs when Nexus is unavailable, so the pipeline never breaks
- Built-in news decay timeline tracks how event influence fades, rather than producing a static score
- Part of a project with standardized agent interfaces, full traceability, and AGPL-3.0 auditable source
- Single data source: depends on the Nexus News API; without Nexus credentials you only get stub results, not real news analysis
- The SKILL.md does not explain how the impact score is computed, what news coverage exists, or how the decay model works, making output quality hard to assess independently
- No skill-specific tests or benchmarks are cited; the repo-level backtest figures describe the whole system and cannot be attributed to this skill
- The interface is oriented to AIMM's internal pipeline (BTC/USDT-style tickers, nexus_context), so standalone reuse requires adaptation
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| News Narrative Miner this page | 46 · Use with care | ★ 2.1k | 2mo ago | AGPL-3.0 |
| DeepEar | 48 · Use with care | ★ 285 | 2mo ago | MIT |
| Pattern Recognition Bot (2.1) | 42 · Not recommended | ★ 2.1k | 2mo ago | AGPL-3.0 |
| Lighter Agent Kit | 57 · Use with care | ★ 67 | 2mo ago | MIT |
| Minara — Personal Crypto AI CFO | 51 · Use with care | ★ 362 | 1mo ago | — |
Within AIMM's architecture this skill corresponds to the narrative side of the README's "Sentiment & Narrative" research desk, complementing the Market Scan, Technical TA, and Statistical Alpha desks; the source names no external competitor, so no external comparison is offered.
How did FollowSkills review this skill?
Read-only news/narrative analysis; no malware, credential theft or covert exfiltration found. SKILL.md discloses the Nexus News API dependency and a neutral-stub fallback, giving partial data-flow transparency. Deducted for: undisclosed scope/frequency of external requests, no user-confirmation or rollback mechanisms, unverified publisher identity, and incomplete attribution — well short of the 25 anchor.
Repo-level CI (ruff + pytest) and deterministic Nexus fixtures covering news endpoints plus context-building tests make the happy path plausible. Deducted for: no executed verification, thin skill-level failure feedback, edge-case handling and fallback ('neutral stub') behavior not specified in the skill files; capped at 10 by static calibration.
Clear scenario (crypto news impact and narrative-phase analysis) with an explicit query interface and tier0/v1 output contract. Deducted for: no declared capability boundaries or non-fit ranges; core dependency on the overseas Nexus News API with a credit/settlement system raises mainland-China reachability concerns; no Chinese-language support noted.
SKILL.md is concise and readable; persona.md supplies reasoning guidelines, output contract and few-shot examples; repo carries AGPL-3.0 and layered docs. Deducted for: no per-skill versioning/changelog/known-limitations/FAQ, hidden NexusDataClient configuration assumptions, and no skill-level maintenance/update responsibility.
Output format (impact score, event type, decay, reasoning) is explicit and marginal value within the quant framework is discernible (decay modeling, black-swan grading). Deducted for: static review cannot confirm directly usable outputs; few-shot numbers are author claims with no representative output samples or comparison against manual alternatives.
Auditable primary material exists: CI workflow, test fixtures and news-context unit tests partially support claims. Deducted for: no third-party execution evidence; skill-level key paths (score formula, decay model, black-swan trigger) lack independently reproducible coverage; fact/inference separation unmarked — below the 5 cap.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →