- Home
- Skills
- Social Media
- Veille Agent – Tech Watch & LinkedIn Posts
Works with the AI Agents you already use
Veille Agent – Tech Watch & LinkedIn Posts
Save links and notes to a local, tagged library: your agent summarizes each source, checks how reliable it is, files it, then drafts LinkedIn or blog posts in your own voice, with optional SEO.
$12
Veille Agent – Tech Watch & LinkedIn Posts
Example session with this skill installed
I just read this: https://www.anthropic.com/research/building-effective-agents
Save it to my watch, then draft a LinkedIn post about it.
- Read your context and instructions
- Compiled the veille agent
This link wasn't in your library yet, so I read the article, qualified it and saved it.
Saved to your watch
| Field | Value |
|---|---|
| ID | 1 |
| Title | Building effective agents (Anthropic) |
| Folder | AI |
| Tags | agents, ai, architecture, llm |
| Nature | Article |
| Summary | Anthropic shares lessons from dozens of teams building LLM agents: the most successful ones used simple, composable patterns rather than complex frameworks. The post separates workflows (predefined code paths) from agents (the model directs its own steps), describes five workflow patterns, and advises starting with plain LLM API calls. |
| Reliability | High: written by Anthropic's engineering team from real deployments. Published December 2024, and the page itself notes the tooling has changed since, so framework names may be dated while the patterns still hold. |
| Why it matters | It gives you a shared vocabulary to decide how much autonomy a feature really needs before reaching for an agent framework. |
LinkedIn draft (in your brand voice)
Before you add an agent framework to your project, read this.
Anthropic looked at dozens of teams building LLM agents. The ones that worked best weren't using complex frameworks. They used simple patterns they could understand and debug.
The useful part is the line they draw between workflows and agents. A workflow follows code paths you define. An agent decides its own steps. For most features, the first one is enough.
They name five workflow patterns: prompt chaining, routing, parallelization, orchestrator-workers and evaluator-optimizer. Having names for them makes it much easier to pick the simplest one that does the job.
My favourite detail: on their coding agent, the team spent more time improving the tools than the main prompt. Writing a clear tool definition is prompt engineering too.
The post dates from December 2024, so some framework names have aged. The patterns haven't.
Worth reading if you build with LLMs: https://www.anthropic.com/research/building-effective-agents
Next steps
- Tell me what to change in the post, or approve it and I'll save it with the article.
- Want it SEO-ready for a blog? I can look up real search keywords (paid per call, I'll show the price first).
- Ask anytime: "what did I save about agents?" or "show my AI folder".
Connects securely to your tools. The creator never sees your data.
About this skill
Read-later apps often become places where links go to die. This skill turns what you read into a searchable tech-watch library, and helps you share the best of it with your own angle.
What it does
- Qualifies each source: summary, reliability check, why it matters and what you learn from it.
- Files and tags it: picks a folder and reuses your existing tags, so your library stays consistent over months.
- Avoids duplicates: checks if a link is already saved before processing it.
- Lets you search: filter by folder, tag, category or keyword, and get stats on your library.
- Drafts posts: writes LinkedIn or blog posts that comment on a saved source in your brand voice. You review before anything is saved, and nothing is posted for you.
- Optional SEO: with search data tools, suggests real keywords, an SEO title and description.
How it works
- Share a link or paste a text and ask your agent to save it to your watch.
- The agent reads it, qualifies it, files it and saves it to a local SQLite database (
~/.veille-agent/veille.db). - Ask for a post about any saved article. Describe your voice once in
~/.veille-agent/brand.mdand every post will follow it.
Requirements
Node.js 22.13+ (uses Node's built-in SQLite, nothing to install) and an agent that can run shell commands and fetch web pages. Your data stays on your machine.
Known limitations
The agent can only qualify pages it can read: for paywalled pages, paste the text instead. The skill does not publish to social networks.
Requires Node.js 22.13+ for SQLite. Cannot bypass auth/paywalls for URL fetching. Does not include an automated web fetcher; relies on your agent's available tools.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 days ago
Requires
Permissions
File Scopes
Shell runs the local storage script. Files: the SQLite database and brand file in ~/.veille-agent. Network: reading the web pages you ask to save.