Skip to main content

Talaria. Seven-league messengers — parallel sub-agents for deep multi-source research.

Project description

Talaria Logo

SvarogForge

⚡ Talaria — Parallel Messengers for Hermes Agent

Talaria (Ταλάρια) — the winged sandals, forged in Svarog's smithy, swift as seven-league boots of legend. One stride carries your AI messengers across every source at once, bringing back intelligence.

Talaria is a Hermes Agent skill that dispatches 3-5 parallel sub-agents for deep multi-source research, then verifies and synthesizes results into a single report.

EN RU

Stars Last commit License CI Hermes Skill

Python JSON Markdown GitHub Actions


💡 Why Talaria?

Problem: Deep research takes forever when you query one source at a time. Each search → read → repeat cycle costs minutes.

Solution: Talaria dispatches 3–5 parallel sub-agents — each with its own research axis, tools, and context. They run simultaneously, then verify and synthesize results into one report. What took 15 minutes now takes 2.

Talaria vs. Raw delegate_task

Feature delegate_task (bare) ⚡ Talaria
🛡️ Anti-hallucination ❌ You build it verify-report.py with 4 profiles
🔄 Cascade search ❌ Manual fallback ✅ 4-tier auto (GitHub→raw→API→browser)
🧠 Smart synthesis ❌ You merge manually ✅ Intersections + contradictions + insights
🔒 Safe sub-agents ❌ No guardrails ✅ 6 blocked tools
📊 Benchmark data ❌ None ✅ Preliminary: 2.5 min flat, 7.2 min nested
💾 Output structure ❌ Ad-hoc reports/gonets*.md + synthesis*.md

🎯 Quick Start

# Install in 30 seconds
git clone https://github.com/SvarogForge/talaria.git
hermes skill install ./talaria

Then load in your session:

skill_view(name='talaria')
# Verify it works
python talaria/verify-report.py --list-profiles
# → default, strict, tech, quick

🏛️ How It Works

flowchart TB
    A["💬 User asks research question"] --> B["⚡ Talaria dispatches 3-5 messengers"]
    B --> C1["📡 Messenger 1<br/>GitHub API + Browser"]
    B --> C2["📡 Messenger 2<br/>HuggingFace + Direct APIs"]
    B --> C3["📡 Messenger 3<br/>Browser + Curl"]
    C1 --> D["🔍 verify-report.py<br/>checks for hallucinations"]
    C2 --> D
    C3 --> D
    D --> E["🧠 Synthesis → One report<br/>with intersections & insights"]
    E --> F["📬 Executive summary → you<br/>Full report → reports/*.md"]
Step What Happens
1️⃣ DISPATCH — 3-5 messengers delegate_task in parallel
2️⃣ RESEARCH — each uses GitHub API, browser, or direct APIs
3️⃣ VERIFY — cross-check against trusted domains
4️⃣ SYNTHESIZE — merge findings, find contradictions
5️⃣ DELIVER — Full report + executive summary

🎬 In Action

Talaria Terminal Demo
3 messengers dispatched → verified → synthesized in 2.3 minutes

✨ Key Features

Feature Description
Parallel Dispatch 3-5 messengers run simultaneously
🛡️ Anti-Hallucination Cross-verification, fake detection, URL validation
🔄 Cascade Search Falls through GitHub → browser → direct APIs
📊 Smart Synthesis Finds intersections, contradictions, insights
🧹 Clean Research All examples are technical research
🔒 Safe Sub-Agents 6 blocked tools prevent harm

📦 What's Included

talaria/
├── SKILL.md              — Full skill documentation
├── verify-report.py      — Anti-hallucination checker (4 profiles)
├── references/
│   ├── architecture.md       — Dispatch & 6 unique features
│   ├── working-apis.md       — API endpoints (no proxy)
│   └── anti-hallucination.md — Detection guide
├── examples/
│   ├── 01-market-research.py
│   ├── 02-tech-audit.py
│   ├── 03-pricing.py
│   └── 04-trends.py
├── README.md              — This file
├── README.ru.md           — Russian version
└── LICENSE                — MIT

📋 Example Output

When Talaria finishes, you get:

🔍  TALARIA VERIFICATION — messenger report validation
======================================================================
reports/gonets1_market.md:
  Status: ✅ REAL DATA  (profile: tech)
  URLs total: 12, real: 8
  Sources: github.com, huggingface.co, pypi.org
  Code: ✅ | Tables: ✅ | Lists: ✅

✅ All messengers returned REAL data!

📊 SYNTHESIS — Key findings:
  • Top 3 AI frameworks by GitHub stars (ranked)
  • 2 contradictions flagged (data age mismatch)
  • Pricing table: DeepSeek $0.14 vs GPT-4o $2.50 per 1M tokens

🔥 From the forge-fire

Talaria is part of SvarogForge — a family of tools forged for Hermes Agent.

Project Description
Talaria Seven-league messengers for market research (you are here)
🔥 Crucible Touchstone for quality & benchmarks
⚒️ Forge The smithy itself — AI-powered project forge

📄 License

MIT — see LICENSE


Star on GitHub · 🐦 Follow @iMonstra · 💬 Join Discussions

Built for Hermes Agent · MIT · Contributions welcome

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

svarog_talaria-0.1.0.tar.gz (161.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

svarog_talaria-0.1.0-py2.py3-none-any.whl (167.6 kB view details)

Uploaded Python 2Python 3

File details

Details for the file svarog_talaria-0.1.0.tar.gz.

File metadata

  • Download URL: svarog_talaria-0.1.0.tar.gz
  • Upload date:
  • Size: 161.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.11.27 {"installer":{"name":"uv","version":"0.11.27","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":null,"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for svarog_talaria-0.1.0.tar.gz
Algorithm Hash digest
SHA256 dd4296f711ed03cade08496cb0cfaa35c12d27d4eb8286fb614934485babc42e
MD5 c98e1f4c388debb5ddff4140fbc5816c
BLAKE2b-256 cfd297b492a757c7dfa337f53e1700fe63736cbeecdfb0a0da47f0cc0df38594

See more details on using hashes here.

File details

Details for the file svarog_talaria-0.1.0-py2.py3-none-any.whl.

File metadata

  • Download URL: svarog_talaria-0.1.0-py2.py3-none-any.whl
  • Upload date:
  • Size: 167.6 kB
  • Tags: Python 2, Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.11.27 {"installer":{"name":"uv","version":"0.11.27","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":null,"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for svarog_talaria-0.1.0-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 deaa62d22fca61b0d20db6489a44f48b699cf23035c2583d0f1185db0d6bb1f3
MD5 0c8ecefae66e5db98d7e664cbbe08edd
BLAKE2b-256 c63705e6af3775632ba9a39401d8116cf2ecc327dcfdf06e6f3de68f64ce0b4d

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page