Conduct grounded, deduplicated multi-source technical research integrating mvanhorn/last30days-skill (commit a854f3b) with primary/secondary evidence classification, staleness detection, and SHA-256 citation tracking. Use for technical briefs, engineering standards lookup, and source-traceable drafting.
Installs into .claude/skills of the current project.
Sponsors
SSkills DirectorySkills Directory
Know which skills are safe — weekly.
SSkills DirectorySkills Directory
Is that skill safe to install?
Are you the author of Rei Multisource Research?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/rootcastleco-rei-multisource-research)
---
name: rei-multisource-research
description: "Conduct grounded, deduplicated multi-source technical research integrating mvanhorn/last30days-skill (commit a854f3b) with primary/secondary evidence classification, staleness detection, and SHA-256 citation tracking. Use for technical briefs, engineering standards lookup, and source-traceable drafting."
category: data-ai
risk: safe
source: rootcastle-rei
license: MIT
metadata:
tags:
- research
- last30days
- citations
- evidence
- deduplication
---
# REI Multisource Research (`last30days-skill` Adapter)
Collect, normalize, deduplicate, and trace engineering sources via [`skills/last30days`](../last30days/SKILL.md) and [`lib/multimodal-providers.js`](../../lib/multimodal-providers.js).
## When to Use
- Gathering standards, papers, repository docs, or recent discussions across Reddit, X, arXiv, YouTube, HN, or scoped engineering corpora.
- Producing `research/sources.json` and `research/evidence-report.json` with traceable `SRC-*` IDs for downstream slides and narration.
## Inputs and Outputs
- **Inputs:** Scoped research topic, date window (`maxAgeDays`), optional local/verified source entries, and optional `OPENAI_API_KEY` / `XAI_API_KEY`.
- **Outputs:** `research/sources.json` (normalized URLs, deduplication counts, SHA-256 content hashes) and `research/evidence-report.json` (`PRIMARY_EVIDENCE` vs `SECONDARY_EVIDENCE` counts and claim-to-citation mappings).
## Prerequisites & License Constraints
- Upstream: `mvanhorn/last30days-skill` (`MIT`, commit `a854f3bcb2c4344d7498426d653378510f8b253a`).
- Respect external platform Terms of Service; do not bypass paywalls or read unauthorized private browser cookies.
## Offline vs Online Capabilities
- **Offline:** Runs deterministically over scoped local source corpora and Rootcastle Prompt Archive snapshots without network access.
- **Online (Opt-In):** Invokes `skills/last30days/scripts/last30days.py --preflight` or web/API backends only when explicitly requested.
## Procedure
1. Inspect planned reads/writes with `last30days.py --preflight` before any external query.
2. Normalize URLs (strip UTM/tracking parameters) and compute deduplicated SHA-256 hashes.
3. Classify every claim as `PRIMARY_EVIDENCE`, `SECONDARY_EVIDENCE`, or `INFERENCE`.
```sh
python skills/last30days/scripts/last30days.py --preflight
node bin/install.js pipeline run "Rotor vibration ISO 20816-1 brief" --output ./tmp-research --json
```
## Do not use
Constraints and safety: Operational boundary — do not invent citations, treat secondary blog summaries as primary experimental data, or scrape blocked sources.
## Verification Criteria & Failure Recovery
- Verify `allClaimsTraceable === true`, `staleClaimsCount === 0` (or explicitly flagged), and valid SHA-256 digests (`^[a-f0-9]{64}$`) for every source.
- If API keys are missing, degrade cleanly to offline/scoped evidence mode without failing unrelated workflows.
## Authoritative product references
- https://github.com/mvanhorn/last30days-skill