guide_skill

This skill automatically detects repeating workflow patterns, generates reusable SKILL.md files, and loads them mid-session for faster automation.
  • TypeScript

6

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill matrixy/auto-skill --skill guide

  • install-cli.sh1.9 KB
  • SKILL.md6.6 KB

Overview

This skill provides automatic detection of repeated workflow patterns, generates reusable skills from those patterns, and supports dynamic mid-session loading. It streamlines repetitive multi-step tasks by converting observed tool sequences into loadable, constrained routines with confidence scoring and execution context.

How this skill works

The plugin records every tool call and analyzes session histories to find repeated sequences (default: 3+ occurrences within a lookback window). Detected patterns are turned into generated skills with metadata, allowed-tool restrictions, and execution context, then indexed in a lightweight registry for on-demand loading. A CLI and in-session commands let you review, approve, and load skills without restarting the agent.

When to use it

  • You repeatedly perform the same multi-step tool sequences across sessions.
  • You want to formalize common workflows into safe, reusable routines.
  • You need to load new capabilities mid-session without restarting the agent.
  • You want guided, confidence-ranked suggestions for automating tasks.
  • You need isolation for side-effectful flows (e.g., builds, deployments).

Best practices

  • Review detected patterns before approving to ensure intent and safety.
  • Set conservative detection thresholds (min occurrences and lookback) during early use.
  • Assign minimal allowed-tools to generated skills to reduce scope creep.
  • Use fork context for workflows with side effects to isolate effects from the main session.
  • Periodically run discovery and stats commands to monitor adoption and quality.

Example use cases

  • Search-read-edit loop: detect Grep → Read → Edit patterns and convert them into a single loadable skill.
  • File creation flow: detect Glob → Read → Write sequences for quick scaffolding tasks.
  • Test-and-run cycle: Read → Edit → Bash patterns become a forked skill for running tests safely.
  • Proactive assistance: request a multi-step task and let the agent propose loading a matching skill.
  • On-demand automation: load a generated skill mid-session to handle a repetitive maintenance operation.

FAQ

By default a sequence must appear at least three times within the configured lookback period to be considered.

Can generated skills run without restarting the session?

Yes. Generated skills are indexed in a registry and can be loaded dynamically mid-session so they become active immediately.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational