See changes
Dates, units and verification stay with every value.
Dated macro data. Public reports. Author reasoning. One small skill for your AI assistant.
26 KB class bundle · Node 22+ · No API key · Public reads only
Dates, units and verification stay with every value.
Combine evidence and viewpoints into a decision, with action and invalidation conditions.
RSS for new publications. Confirmed weekly email for your theme.
Use $argus-research to find the latest AI infrastructure daily report, then explain its debt reasoning with sources.PROJECT / FILES.agents/skills/argus-research/SKILL.md.agents/skills/argus-research/npx --yes skills@1.7.2 add wp3xpp/argus-skill --skill argus-research --agent codex --copy --yesInvoke $argus-research. The shipped helper has been tested against production.
.claude/skills/argus-research/npx --yes skills@1.7.2 add wp3xpp/argus-skill --skill argus-research --agent claude-code --copy --yesInvoke /argus-research. Same folder and read-only helper; no provider SDK.
npx skills@1.7.2 add wp3xpp/argus-skill --skill argus-researchInstall where your client loads Agent Skills, or read the API directly with HTTP tools.
OpenAPI 3.1 ↗skills.sh installation guide · Client installation & network guide →
Codex and Claude Code install commands verified in isolated projects with skills CLI 1.7.2; installed helpers read live ARGUS data. Requires Node 22+. Copying does not run the command. No automatic subscriptions or schedules. SHA-256 · Bundle manifest
Install first, then copy a prompt into your assistant. In Codex you can prefix it with $argus-research; in Claude Code use /argus-research. Replace dates and themes as needed.
| Purpose | You can say | Copy |
|---|---|---|
| Form an investment judgment | Use ARGUS data, recent reports and contrasting author views to assess increasing AI infrastructure exposure over the next 3–6 months. Give a clear decision, evidence table, causal reasoning, strongest counterargument, action conditions and falsification. I have not supplied holdings or current valuation; keep entry prices and sizing conditional. | |
| Catch up on research | Use ARGUS to read the latest global liquidity weekly report. Highlight three takeaways and include the report date and reading link. | |
| Check an indicator | Use ARGUS to check China PMI new orders, including the observation month, value, unit, source and verification status. | |
| Explore historical trends | Use ARGUS to examine PMI new orders over the past year. List available months, explain the trend and turning points, and flag missing observations. | |
| Read a report in depth | Use ARGUS to find the latest AI infrastructure report. Expand its main judgment, premises, causal links and intermediate steps. Explain supporting evidence and invalidation conditions, with sources. | |
| Understand an author’s view | Use ARGUS to find the latest indexed Arthur Hayes guide. Trace his reasoning, separate his argument from ARGUS interpretation, and include falsification conditions and the original link. | |
| Track what changed | Use ARGUS to find recorded China economic updates and corrections since October 1, 2026. Start with five, including observation dates, correction reasons and the coverage of the records. | |
| Get a subscription link | Use ARGUS to give me the English AI infrastructure research-digest subscription link and explain email confirmation and unsubscribe. | |
| Get an RSS feed | Use ARGUS to give me the English global liquidity daily-report RSS URL and explain how to add it to a reader. |
ARGUS preserves dates, sources and verification. Missing history and incomplete reports are disclosed. Author guides are editorial interpretations, with links to originals. Email subscriptions require website confirmation and send at most one weekly digest.
Public retrieval uses no model API. Your assistant may consume tokens to explain results; metadata and section reads keep context small.
A frozen example from ARGUS records retrieved October 13, 2026. Research horizon: 3–6 months. Holdings and current valuation were not supplied; recompute before using as a current view.
Microsoft quarterly OCF minus cash CapEx: +$19.639bn; Amazon: −$8.821bn, improving from −$18.171bn. Period: June 30, 2026; consolidated businesses, not AI-only returns. ARGUS marks these observations primary-verified.
Causal path: builders’ spending becomes suppliers’ revenue, but it does not prove a return above capital costs. The Damodaran guide supplies this distinction; the Richter guide highlights equity-financing dependence. This synthesis is an inference, not an author’s trade instruction.
Counterargument: Amazon’s deficit narrowed, and the October 12 report does not yet see financing stress passing into hardware orders. This does not support a blanket sell call either.
Verified order conversion, improving comparable cash conversion and valuation supported by explicit capital-cost assumptions.
Financing failures cause actual project delays and subsequent procurement cuts; or margins and collections deteriorate together.
Moderate confidence in operating differentiation; entry valuation is unassessed. No portfolio sizing or leverage instruction without investor constraints.
Try the same public API used by the skill. Literal title, summary and author search; no model call.
Sunday 18:00 Beijing. At most one email per week, when new research exists. Confirm your address; unsubscribe anytime.
Choose a theme & subscribe →Daily reports are available to read; daily email is not enabled.New reports and published author guides, with stable IDs and original dates. No email address needed.
Read-only JSON, no credentials, paginated search and section reads. Reuses the existing cloud infrastructure.
ARGUS guides are editorial paraphrases, not authorized full translations. Source and report coverage remain explicit. Helper code is MIT; retrieved content retains its own rights.
One digest each Sunday at 18:00 Beijing time, only when your selected theme has new research. Confirm to join; unsubscribe anytime.