leogrleogrCommunityยท1 Agent Skills Included

coding-agents-kit

Real-time guardrails and audit trails for coding agents

Intercepts every shell command, file write, and web request a coding agent makes before it runs. Evaluates each action against customizable Falco rules and returns allow, deny, or ask verdicts in real time. Eliminates blind trust in autonomous coding sessions with a full audit trail and ready-made security policies. Works across Linux, macOS, and Windows with Claude Code and Codex.
npx skills add leogr/coding-agents-kit --all -g -y
Available:

Explains the interception pipeline, verdict system, and rule structure so the agent can author, validate, and troubleshoot Falco policies correctly.

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install coding-agents-kit?โ–ผ

Run `npx skills add leogr/coding-agents-kit --all -g -y` in your terminal to install all skills in this suite globally.

How to control what a coding agent can do?โ–ผ

Prempti intercepts every tool call before execution and applies your Falco rules to allow, block, or ask for confirmation. You can write custom YAML rules to set boundaries like blocking access to .env files or preventing pushes to remote.

How to write custom Falco rules for coding agents?โ–ผ

Use the included prempti-falco-rules skill, which guides your agent through drafting, validating, and installing rules with the correct fields, tags, and output format.

Does it work with Claude Code and Codex?โ–ผ

Yes. Claude Code is the primary integration on Linux, macOS, and Windows, with experimental support for Codex CLI.

Can I monitor agent activity without blocking anything?โ–ผ

Yes. Switch to monitor mode with `premptictl mode monitor` to log and evaluate every tool call without enforcing verdicts.

Related Repositories in Software Engineering

View All in Software Engineeringโ†’