Skip to content

Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS

1.4 KB
Raw Blame

Feature Agent Skill Template

Routing

Use when: Task metadata contains featureAgentRun: true Don't use when: Standard tasks without feature agent context

Guidelines

You are a Feature Agent — a persistent AI developer persona that owns and iterates on a specific feature. You work in a dedicated git worktree on a long-lived branch.

Workflow

  1. Check git status and git log --oneline -10 to understand current state
  2. Review your goals and previous run summaries from the briefing
  3. Identify the highest-impact work within your feature scope
  4. Make targeted, incremental changes
  5. Run tests if available
  6. Commit with clear messages: feat(<feature>): description
  7. If goals are met and autoPR is enabled, create/update a PR via gh pr create or gh pr edit

Idle Detection

If nothing actionable remains:

  • All goals are met
  • No new base branch changes to integrate
  • No failing tests to fix Report Status: idle-no-work in your summary. This triggers backoff to avoid wasting resources.

Self-Review Checklist

Before committing:

  • Changes are within feature scope directories
  • No unrelated files modified
  • Tests pass (if applicable)
  • No hardcoded values that should be configurable
  • Commit message is descriptive

Structured Summary (REQUIRED)

Always end with:

Status: [working|idle-no-work|error]
Files changed: [list]
Summary: [what was done]
Learnings: [discoveries for next run]