Sovereign Skills for Coding Workflows

What does this agent do, and when should you use it?

Sovereign Skills contains 22 standalone skill directories, each described by a SKILL.md workflow and, where needed, bundled Python helpers. It covers project scaffolding, scope definition, execution discipline, pre-push checks, code review, session handoffs, and audits of documentation and evaluations. Users invoke skills as slash commands in Claude Code, install them under Codex's .agents/skills/ path, or copy the Markdown instructions into Cursor and other agents that read Markdown. Skills read project files and optional session records, run checks or guide an agent through reviews, then produce lists, scores, audit findings, or next-action suggestions. This is a skill collection installed into an agent environment, not a standalone service; some workflows refer to optional agents and document skip behavior or fallback steps when they are absent.

For a new project, /project-init generates CLAUDE.md, ROADMAP, .gitignore, and .env.example from an interview, then /setup configures rules, hooks, memory, routing, and agents. In the daily flow, /scope defines boundaries and exit criteria, /freeze declares the editable zone, and /goal-lock enforces a PLAN→DO→VERIFY loop; /pre-push scans outgoing changes for secrets, runs project build/tests and lint, and performs AI code review. /code-autopsy produces a 12-question review with severity scoring and a deployment verdict, while /session-start and /session-checkpoint read or save handoffs and lessons. Governance skills scan projects, audit loaded documentation, check for circular evaluation, trace duplicate copies of a fact, or inspect skill routing; outputs include gap lists, coverage maps, scores, findings, and recommendations. Skills install as local directories and read project files; no network service or remote API call is documented.

1. A developer maintaining a Claude Code project who wants scope, an editable boundary, and a verifiable completion loop before implementation.
2. An engineer preparing a push who wants secrets scanning, build/tests, lint, and AI review in one workflow.
3. A solo developer working across sessions who needs to restore handoffs, extract lessons, and save state for the next session.
4. A maintainer of custom skills who wants usage tracking, snapshots and rollback, or checks for trigger-description collisions.
5. An engineer responsible for documentation or evaluation quality who needs to audit context drift, duplicated facts, or evaluation leakage.

How do you install or deploy this agent?

For Claude Code, the simple installation method is to clone the repository and copy the skill directories you want. To copy every directory containing SKILL.md:

git clone https://github.com/AlexZio00/sovereign-skills.git
cd sovereign-skills
for d in */; do [ -f "$d/SKILL.md" ] && cp -r "$d" ~/.claude/skills/; done

To install one skill, for example:

cp -r goal-lock ~/.claude/skills/

Codex user-level skills go under ~/.agents/skills/; repository-level skills go under .agents/skills/:

cp -r goal-lock ~/.agents/skills/goal-lock/

The repository also documents a Claude Code marketplace route. For Cursor and other agents, copy SKILL.md into the agent's instruction path. Skills with bundled scripts require Python 3.9+; pre-push includes a Perl fallback for secret scanning. No API key is listed as required.

How do you use this agent?

In Claude Code, invoke a skill by typing its slash command, for example:

/project-check

A new-project sequence can start with:

/project-init
/setup
/session-start

Codex discovers skills under .agents/skills/<skill-name>/SKILL.md; user-level skills can be placed at ~/.agents/skills/<skill-name>/. For another agent that reads Markdown instructions, place SKILL.md in its instruction path and use that agent's supported invocation method. Individual skills read project files; session skills read or write handoff, lesson, and state records when those shared files are present.

What are this agent's strengths and limitations?

How does this agent compare with similar options?

Key facts side by side with the most closely related agents.

How does FollowAgents rate this agent?

FAQ