psychology

chevp-ai-framework

Lifecycle Gates Guidelines Commands Agents Templates
extension Plugin Layer · Skills

Skills

Triggered automatically from natural-language intent — produce framework artifacts from templates.

Each skill wraps a template into a callable artifact generator with built-in validation.

The Four Skills

architecture

create-adr

Record an Architecture Decision Record with context, alternatives, and consequences.

Triggers: "write an ADR", "new ADR", "document this decision", or whenever a non-trivial architectural choice is made in Context or Exploration.

Output: ADR file in context/adrs/ with status Proposed — only Accepted after human approval.

visibility

create-ctx-plan

Kick off Step 1 of the lifecycle — capture WHAT must be understood before any exploration.

Triggers: new unfamiliar task, "context plan", "CTX", "start context step", request without active plan.

Required by: G1 — before /explore or /produce can run.

tune

create-exp-plan

Plan a feature in Step 2 — capture HOW it will be built, the chosen approach, risks, acceptance criteria.

Triggers: "EXP plan", "feature plan", "exploration plan", "design the feature".

Prerequisite: G1 must be passed (verified via gate-validator).

sync

sync-plan-issues

Mirror plan files (context/plans/**/*.md) to GitHub Issues. Source-of-truth stays the markdown.

Triggers: "sync plans", "create issues for backlog", "alle plans als github issues", "issue für §X.Y".

Idempotent: re-running skips existing issues. Always dry-run first.

How Skills Differ from Commands

Aspect Commands Skills
TriggerExplicit slash invocationAuto from natural-language intent
PurposeMode switch & governance actionsArtifact generation from templates
OutputMode change, log entry, validationA new file (ADR, CTX, EXP, etc.)
Used by humansYes, directlyIndirectly — the AI activates them
menu_book

Read the source

skills/ open_in_new