recon

Records external-source research as canonical evidence bundles in a recon vault.

Updated Aug 2, 2026
One-click install
npx skills add https://github.com/leonardoacosta/agents --skill recon-leonardoacosta
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: recon
Source: https://github.com/leonardoacosta/agents/tree/main/.agents/skills/recon
Command: npx skills add https://github.com/leonardoacosta/agents --skill recon-leonardoacosta

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Web research, site captures, and source analysis often produce scattered, unreproducible notes. This Skill routes external-source analysis through the canonical recon CLI so every report, source URL, and supporting asset is recorded as one durable, queryable evidence record instead of a one-off lookup. ## Core Features & Use Cases - Canonical evidence recording: Resolves the installed recon runtime, prepares scrubbed query and finding ledgers, and writes a single source-addressed record via recon record. - Routed acquisition capabilities: Maps evidence needs to recon modes such as recon:research, recon:design, recon:papers, recon:social, and recon:agents for web research, site crawling, scholarly papers, social metadata, and agent/workflow analysis. - Design-system mining: Captures a site's tokens, layout, components, and assets into a validated, versioned design bundle that can feed UI-library generation. - Use Case: Research a competitor's product site, capture its design system into a bounded evidence bundle, and record the report with full provenance so future work reuses the same canonical record. ## Quick Start Ask the agent to research a competitor's website with recon and record the findings as a canonical evidence record in the recon vault.

Frequently Asked Questions about recon

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I record web research as reusable evidence with recon?

Resolve the recon CLI, run recon query to check prior coverage, produce the report in a temporary directory, prepare query and finding ledgers with recon prepare-ledgers, then write the canonical record with recon record using a source URL or inquiry key.

How do I extract a design system from a website?

Route the request to recon:design, capture the target pages, and validate the bundle with recon prepare-design. The bundle records tokens, layout, components, and assets with anchors, confidence, and classification labels separating observed values from inference.

When should I not use recon for a web lookup?

Do not use recon for one-off page fetches or transient lookups with no report, reuse, or memory goal. Those requests are declined with a bounded routing explanation and create no vault record or query ledger.

What happens if the recon CLI is not installed or configured?

The skill checks command -v recon, recon --version, and recon doctor --json, then falls back to an explicit RECON_HOME checkout. If neither works it returns a bounded unavailable result and never installs software, creates a vault, or writes configuration.

Can recon authenticate, log in, or submit forms during acquisition?

No. Recon acquisition is read-only against the public web. Installation, authentication, form submission, purchases, telemetry, and export of private data are outside its authority and require separate explicit authorization under host policy.