A collection of agent skills for the forjd organisation — portable, version-controlled capabilities that any skills-compatible AI agent can discover and use.
| Skill | Description |
|---|---|
repo-hardening |
Audit and harden GitHub repository security settings via the gh CLI |
github-pr |
Create standardised pull requests with conventional titles, templates, and reviewers |
pr-review-actioner |
Action PR review feedback — triage, fix, and reply to unresolved review comments |
browse |
Browser automation CLI for navigating, testing, screenshotting, and interacting with web pages |
Via the Skills CLI (no setup required):
npx skills add forjd/agent-skills
# or
bunx skills add forjd/agent-skillsVia git clone, then copy the skill directories into your agent's skills directory. Most clients discover direct child directories that contain SKILL.md, so copy skills/* rather than relying on the collection repo root being scanned.
# Project-level (one project)
git clone https://github.com/forjd/agent-skills.git /tmp/agent-skills
mkdir -p .agents/skills
cp -R /tmp/agent-skills/skills/* .agents/skills/
# User-level (all projects)
git clone https://github.com/forjd/agent-skills.git /tmp/agent-skills
mkdir -p ~/.agents/skills
cp -R /tmp/agent-skills/skills/* ~/.agents/skills/| Agent | User-level path |
|---|---|
| Claude Code | ~/.claude/skills/ |
| Cursor | ~/.cursor/skills/ |
| Cross-client | ~/.agents/skills/ |
Individual skills can also be copied directly because each skills/<name>/ directory is self-contained.
See the Agent Skills docs for the full list of supported agents and discovery paths.
# Audit a repo's security posture
bash skills/repo-hardening/scripts/harden.sh audit --repo forjd/my-service
# Preview fixes without applying
bash skills/repo-hardening/scripts/harden.sh fix --repo forjd/my-service --checks branches --required-check "test" --dry-run
# Apply confirmed fixes
bash skills/repo-hardening/scripts/harden.sh fix --repo forjd/my-service --checks branches --required-check "test"skills/
└── <skill-name>/
├── SKILL.md # Instructions + metadata (required)
├── scripts/ # Executable code
├── references/ # Detailed documentation
└── assets/ # Templates, resources
Each skill follows the Agent Skills specification. See docs/skill-guidelines.md for our authoring guidelines.
MIT — Forjd.dev