sigil_skill

This skill analyzes a project's context to generate tailored Claude Code micro-skills that fit its conventions and workflows.
  • Shell

8

GitHub Stars

1

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 simota/agent-skills --skill sigil

  • SKILL.md12.0 KB

Overview

This skill analyzes a project's codebase, manifests, and existing agent skills to generate project-optimized Claude Code skills. It discovers repeatable patterns and workflows, then synthesizes micro or full skills and installs them into the project's .claude/skills/ and .agents/skills/ directories while keeping both locations synchronized. It logs activity and validates generated skills before installation.

How this skill works

Sigil performs a mandatory SCAN to detect tech stack (package.json, go.mod, Cargo.toml, pyproject.toml, etc.), directory layout, config files, and existing skills. It runs DISCOVER to find automation opportunities (scripts, Makefile targets, repetitive code patterns) and then CRAFTS Micro or Full skills according to project conventions. INSTALL writes synchronized files to both skill directories, creates references for Full Skills when needed, and VERIFY runs format, duplication, and sync checks, then logs results to project activity files.

When to use it

  • When you want to convert repetitive project workflows into reusable Claude Code skills
  • When onboarding a repository to an agent-driven developer workflow
  • When you need a synchronized skill inventory in .claude/skills/ and .agents/skills/
  • When you want automated discovery of skill opportunities (npm scripts, Makefile, patterns)
  • Before creating new long-form ecosystem agents — to prefer project-specific micro/full skills first

Best practices

  • Always run a full SCAN before generating or installing any skills
  • Prefer Micro Skills (10–80 lines) by default; request Full Skills (100–400 lines) only when needed
  • Ask for confirmation before bulk generation (10+ skills) or overwriting existing files
  • Keep generated skills free of secrets and avoid modifying global ecosystem agents
  • Maintain activity logs in .agents/PROJECT.md and follow the agent logging format

Example use cases

  • Generate a set of micro skills to automate common npm script workflows and tests
  • Create Full Skills for complex CI tasks that require references/ and multi-step logic
  • Sync and repair orphaned skills between .claude/skills/ and .agents/skills/ after a repo restructure
  • Discover and scaffold a new component creation skill from repeating frontend patterns
  • Audit the project for missing automation and produce a prioritized skill generation plan

FAQ

Sigil will not overwrite files silently. It prompts before overwriting and requires confirmation for replacements.

Can Sigil include secrets or environment variables in generated skills?

No. Sigil never includes secrets, credentials, or raw .env values in generated skills.

What file types and manifests does Sigil inspect during SCAN?

Common manifests and configs: package.json, go.mod, Cargo.toml, pyproject.toml, Makefile, docker-compose.yml, tsconfig.json, .eslintrc, and existing skill directories.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
sigil skill by simota/agent-skills | VeilStrat