Skillget

edikt

agent
Security Audit
Fail
Health Warn
  • License — License: NOASSERTION
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Fail
  • rm -rf — Recursive force deletion command in .coderabbit.yaml
  • rm -rf — Recursive force deletion command in experiments/exp-001-rule-compliance/run.sh
  • rm -rf — Recursive force deletion command in experiments/exp-001-rule-compliance/setup.sh
  • rm -rf — Recursive force deletion command in experiments/exp-002-extended-compliance/run.sh
  • rm -rf — Recursive force deletion command in experiments/exp-002-extended-compliance/setup.sh
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This tool acts as a governance and compliance layer for agentic software development. It captures engineering decisions and compiles them into automated rules to ensure AI assistants consistently follow your exact architectural standards and development lifecycle.

Security Assessment
The tool poses a High security risk due to its reliance on unsafe shell execution patterns. The automated scan failed after discovering multiple `rm -rf` (recursive force deletion) commands hardcoded throughout its scripts, including the setup, run, and configuration files. Additionally, the installation method requires piping a remote script directly to bash (`curl | bash`), which means your machine will blindly execute whatever code the repository serves. While no hardcoded secrets were found, automatically running destructive file deletion commands via an unvetted remote script is a significant concern.

Quality Assessment
The project is highly active, with its most recent code push occurring today. However, it suffers from extremely low community visibility, having only 5 GitHub stars. This means the codebase has not been broadly peer-reviewed or battle-tested by the open-source community. Furthermore, the repository lacks a clearly defined open-source license (listed as NOASSERTION), which creates potential legal ambiguies regarding how the software can be used or modified in commercial projects.

Verdict
Use with caution — while the concept is valuable and actively developed, the destructive shell commands, lack of licensing, and unvetted `curl | bash` install method make it too risky for sensitive or production environments without a manual code review first.
SUMMARY

The governance layer for agentic engineering — governs your architecture and compiles your engineering decisions into automatic enforcement.

README.md

edikt

The governance layer for agentic engineering.

edikt governs your architecture and compiles your engineering decisions into automatic enforcement. It governs the Agentic SDLC from requirements to verification.

curl -fsSL https://raw.githubusercontent.com/diktahq/edikt/main/install.sh | bash

Then open any project in Claude Code and run /edikt:init.

What it does

Without edikt, every Claude Code session starts from scratch. Standards live in your head. Decisions get forgotten between sessions. Each engineer's Claude drifts differently.

edikt fixes this with two systems that reinforce each other:

Architecture governance & compliance. Capture architecture decisions (ADRs), constraints (invariants), and conventions (guidelines). /edikt:compile reads all three, checks for contradictions, and produces a governance file Claude reads automatically — every session, before writing code. Rule packs add correctness guardrails to the same enforcement surface.

Agentic SDLC governance. PRD → spec → artifacts → plan → execute → drift detection. Status-gated transitions. Specialist agents review at every critical step. Drift detection verifies what was built matches what was decided.

The lifecycle produces new engineering decisions. Compiled decisions govern the lifecycle. Decisions compound rather than decay.

The full cycle

/edikt:prd             → requirements and acceptance criteria
/edikt:spec            → technical specification
/edikt:spec-artifacts  → data model, API contracts, test strategy
/edikt:plan            → phased execution with specialist review
  execute             → Claude builds with enforced standards
/edikt:drift           → verify implementation matches the spec

What edikt installs

  • 20 rule packs — path-conditional standards (Go, TypeScript, Python, Next.js, Django, and more)
  • 18 specialist agents — architect, dba, security, api, qa, sre, and others
  • 9 lifecycle hooks — auto-format, plan injection, compaction recovery, quality gates
  • Compiled governance — engineering decisions (ADRs, invariants, guidelines) compile into directives Claude follows automatically every session
  • 24 commands — from init through drift detection

Research

123 eval runs across 2 experiments prove the enforcement mechanism works. Rules in .claude/rules/ drive 100% compliance on conventions Claude has never seen in training data (15/15 with rules, 0/15 without). Fully reproducible — see experiments/.

Documentation

Full documentation, guides, and examples at edikt.dev.

Claude Code only

edikt uses Claude Code's platform primitives — path-conditional rules, lifecycle hooks, slash commands, specialist agents, quality gates. Other tools don't have them. The knowledge base (project-context.md, ADRs, specs) is plain markdown that works anywhere. The governance loop only works in Claude Code.

No build step. No runtime. No magic.

Every file is .md or .yaml you can read, edit, and version-control. Plain markdown, no dependencies.

Reviews (0)

No results found