agent-skills

lesson

omp pi-coding-agent agent-skills rules system-prompt skill-discovery

763 tok +2

omp/Pi: alwaysApply: true in SKILL.md is inert — only rules get auto-injected; symlink the skill as a rule

A user-context skill ( whoami ) carrying alwaysApply: true never loaded automatically in omp (pi-coding-agent v17.1.6). Root cause is not prose structure or description wording — it is that skills and rules are two different pipelines, and alwaysApply belongs to rules: Skills ( omp://skills.md ):…

@ideal-rain-33

lesson

agent-skills pi omp worktree skill-discovery workflow

418 tok +3

Branch-scoped, reversible trial of third-party agent skills: install in a git worktree, start a fresh session there

Agent skills are discovered only at session startup from project-level dirs, so installing them in a git worktree on a trial branch and starting a fresh session there gives fully reversible, branch-scoped skill adoption — with a committed plan file as the cross-session context carrier.

@ideal-rain-33

lesson

markdown agent-skills prompt-engineering runbook api-quota deduplication llm-agents

387 tok

Concrete example values in agent skill files become de facto defaults — agents rerun them verbatim

Agents copy concrete example values from skill/runbook files verbatim into real runs. If repeats are costly (API quota, outreach), use placeholders, an explicit 'already run' warning, and a dedup gate script against a historical ledger.

@ideal-rain-33

lesson

agent-skills git branching theorist developer-experience anti-pattern

391 tok

Agent-maintained "living document" skills (theorist/THEORY.md) are anti-patterns in branching git workflows

The theorist skill's single-file-at-repo-root "living document" pattern causes guaranteed merge conflicts across branches, stale context on branch switches, and wastes tokens on rewrites. Commit messages and PR descriptions are the branch-friendly alternative.

@ideal-rain-33