Research Findings API — Wyltek Industries

Quick Reference

GET /api/research/findings List all findings GET /api/research/findings?tag=<tag>&limit=N Filtered list with metadata GET /api/research/findings/:id Full finding content GET /api/research/findings/:id?format=markdown Raw markdown GET /api/research/tags All tags with counts GET /api/research/health Health check

🔓 No authentication required All endpoints are public and CORS-enabled. Any agent, script, or browser can call them directly. Rate limits apply at Cloudflare edge level.

Endpoints

GET /api/research/health Service health and available endpoints

200 OK

{
  "status": "ok",
  "service": "wyltek-research-api",
  "version": "2.0.0",
  "source": "github:toastmanAu/kernel-workspace/research/findings",
  "endpoints": [...]
}

GET /api/research/findings List findings — with optional filter and metadata

Query Parameters

Parameter	Type	Description
tagoptional	string	Filter by tag (case-insensitive). All results include title, date, tags and summary.
qoptional	string	Full-text search across id, title, tags and summary fields.
limitoptional	integer	Maximum number of results to return. Default: all.

Examples

# All findings (includes title, tags, summary)
curl https://api.wyltekindustries.com/api/research/findings

# Filter by tag
curl "https://api.wyltekindustries.com/api/research/findings?tag=fiber&limit=10"

# Full-text search
curl "https://api.wyltekindustries.com/api/research/findings?q=payment+channel&limit=5"

// All findings
const res = await fetch('https://api.wyltekindustries.com/api/research/findings');
const { count, findings } = await res.json();

// With metadata + tag filter
const res = await fetch(
  'https://api.wyltekindustries.com/api/research/findings?tag=ckb&limit=20'
);
const data = await res.json();
data.findings.forEach(f => console.log(f.title, f.tags));

import requests

BASE = "https://api.wyltekindustries.com"

# List with metadata
r = requests.get(f"{BASE}/api/research/findings", params={"tag": "fiber", "limit": 20})
for finding in r.json()["findings"]:
    print(finding["title"], finding["tags"])

200 OK

{
  "count": 188,
  "findings": [
    {
      "id": "fiber-payment-channels",
      "filename": "fiber-payment-channels.md",
      "url": "https://raw.githubusercontent.com/...",
      "size": 14230,
      
      "title": "Fiber Network Payment Channels",
      "date": "2026-03-18",
      "tags": ["fiber", "ckb", "payment-channels"],
      "summary": "Overview of Fiber Network's payment channel architecture..."
    }
  ]
}

GET /api/research/findings/:id Fetch a single finding by ID

Path Parameters

Parameter	Type	Description
id	string	Finding ID (filename without `.md` extension).

Query Parameter	Type	Description
formatoptional	markdown	Return raw markdown instead of JSON. Content-Type: `text/markdown`.

Examples

# JSON (metadata + content)
curl https://api.wyltekindustries.com/api/research/findings/fiber-payment-channels

# Raw markdown
curl "https://api.wyltekindustries.com/api/research/findings/fiber-payment-channels?format=markdown"

// JSON response
const res = await fetch(
  'https://api.wyltekindustries.com/api/research/findings/fiber-payment-channels'
);
const { title, date, tags, content } = await res.json();

// Raw markdown (e.g. to pass directly to an LLM)
const md = await fetch(
  'https://api.wyltekindustries.com/api/research/findings/fiber-payment-channels?format=markdown'
).then(r => r.text());

import requests

BASE = "https://api.wyltekindustries.com"
finding_id = "fiber-payment-channels"

# JSON
data = requests.get(f"{BASE}/api/research/findings/{finding_id}").json()
print(data["title"], data["tags"])

# Raw markdown (feed directly to LLM context)
md = requests.get(
    f"{BASE}/api/research/findings/{finding_id}",
    params={"format": "markdown"}
).text

200 OK

{
  "id":      "fiber-payment-channels",
  "title":   "Fiber Network Payment Channels",
  "date":    "2026-03-18",
  "tags":    ["fiber", "ckb", "payment-channels"],
  "summary": "Overview of Fiber Network's payment channel architecture...",
  "content": "# Research: Fiber Network Payment Channels\n\n..."
}

404 Not Found

{ "error": "Not found", "id": "bad-id" }

GET /api/research/tags All tags sorted by usage count

💡 Fast & cheap Reads the pre-built index in a single request — no per-file fetching. Safe to call freely. Agent tip: cache it at boot, refresh hourly.

200 OK

{
  "count": 47,
  "tags": [
    { "tag": "ckb",   "count": 63 },
    { "tag": "fiber", "count": 41 },
    { "tag": "esp32", "count": 28 },
    ...
  ]
}

🤖 Agentic Usage

This API is designed for agent-to-agent consumption. No API keys, no sessions, no webhooks — just plain HTTP GET. Agents can pull research context on demand and inject it into their working memory or LLM context window.

📋 Pre-task Research Injection

Before starting a task, query for relevant findings and inject them as context. Useful for OpenClaw agents, LangChain tools, or any agent framework.

Fetch findings tagged fiber before working on payment channel tasks
Pull the latest ckb findings to inform transaction-building decisions
List includes titles, tags and summaries — fetch individual findings with ?format=markdown for full content

# In an OpenClaw skill or heartbeat task
BASE = "https://api.wyltekindustries.com"

async def get_research_context(tag, limit=5):
    r = requests.get(f"{BASE}/api/research/findings",
                     params={"tag": tag, "limit": limit})
    findings = r.json()["findings"]
    # Build context block for LLM
    context = "\n\n".join(
        f"## {f['title']}\n{f.get('summary', '')}"
        for f in findings
    )
    return context

import requests

def fetch_research_context(tag: str, limit: int = 5) -> str:
    """Pull relevant research and format for LLM injection."""
    BASE = "https://api.wyltekindustries.com"
    r = requests.get(
        f"{BASE}/api/research/findings",
        params={"tag": tag, "limit": str(limit)}
    )
    findings = r.json().get("findings", [])
    return "\n\n---\n\n".join(
        f"**{f['title']}** ({f.get('date','?')})\n{f.get('summary','')}"
        for f in findings
    )

# Example: inject into LLM prompt
context = fetch_research_context("fiber", limit=3)
prompt = f"Use this research context:\n\n{context}\n\nNow answer: ..."

🔍 On-Demand Deep Fetch

List findings first (lightweight), then fetch full content only for the most relevant ones. Minimises token burn on context injection.

Get the ID list first — cheap, just filenames and sizes
Score/rank IDs against your query using your own logic
Fetch ?format=markdown for the top N hits only

Pattern

import requests, re

BASE = "https://api.wyltekindustries.com"

def smart_fetch(query_keywords: list[str], top_n=3) -> list[str]:
    # 1. Get cheap listing
    ids = [f["id"] for f in
           requests.get(f"{BASE}/api/research/findings").json()["findings"]]

    # 2. Score by keyword match in ID string
    scored = sorted(
        ids,
        key=lambda id: sum(kw.lower() in id for kw in query_keywords),
        reverse=True
    )

    # 3. Fetch full markdown for top hits only
    results = []
    for finding_id in scored[:top_n]:
        md = requests.get(
            f"{BASE}/api/research/findings/{finding_id}",
            params={"format": "markdown"}
        ).text
        results.append(md)
    return results

🗺️ Discovery — What's Been Researched?

Use the tags endpoint at startup to discover what topics are covered before planning research tasks. Avoids duplicating work that's already been done.

Cache the tag list at agent boot — it's slow but only needed once
Check if a topic has coverage before spawning a new research crawl
Cross-reference with your own task queue to find gaps

import requests

BASE = "https://api.wyltekindustries.com"

# Fetch tag coverage once, cache locally
tags_data = requests.get(f"{BASE}/api/research/tags").json()
covered_tags = {t["tag"] for t in tags_data["tags"]}

def already_researched(topic: str) -> bool:
    keywords = topic.lower().split()
    return any(kw in covered_tags for kw in keywords)

# Before queuing a new research task:
if not already_researched("fiber htlc"):
    queue_research_task("fiber htlc routing")

💡 Recommended agent pattern Boot: fetch tags → cache. Per-task: query ?tag=X&limit=5 for tagged summaries → inject top 2–3 into LLM context. On demand: fetch individual findings with ?format=markdown when full depth is needed. This keeps token cost minimal while giving your agent strong domain context.

Infrastructure

Runtime

Cloudflare Workers

Edge-deployed, global PoPs

Data Source

GitHub (private repo)

~188 markdown findings

Auth

None (consumers)

GitHub token stored in Worker secrets

CORS

Open (*)

Browser-accessible

🔬 Research Findings API

Quick Reference

Endpoints

Query Parameters

Examples

Path Parameters

Examples

🤖 Agentic Usage

📋 Pre-task Research Injection

🔍 On-Demand Deep Fetch

🗺️ Discovery — What's Been Researched?

Infrastructure