MVPavan/via / .claude/skills/authoring-for-agents
authoring-for-agents
VerifiedUse when creating or changing agent instructions or their activation behavior. Skip edits that only correct spelling or formatting.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 2.4 KB |
| agents/openai.yaml | config | 42 B |
| references/skill-anatomy.md | markdown | 1.8 KB |
| references/testing-docs.md | markdown | 1.4 KB |
| references/writing-principles.md | markdown | 1.8 KB |
references/testing-docs.md
View on GitHub# Evaluating agent instructions
Choose checks for the failure the edit addresses. Use existing evidence and
permitted runtime tools; a prose review cannot prove a model will obey a rule.
| Change | Useful check |
|---|---|
| Editorial correction | Reread and inspect the diff |
| Metadata, pointer or alias | Parse metadata, resolve links, inspect actual discovery |
| Trigger boundary | Fresh-context positive case and a neighboring negative case |
| Workflow rewrite | Representative task with acceptance criteria and a bounded budget |
| Consequential guardrail | Baseline plus a realistic pressure case |
Test discovery with only the normal entrypoint available; handing the body to an
agent bypasses the routing under test. Observe selected actions and resulting
artifacts rather than asking whether the wording is clear. Keep evaluation tasks
isolated from production side effects.
For comparisons, hold task, context, tools and effort constant. Use an unchanged
control when attributing improvement to wording. Repeat only enough to distinguish
variance from a material effect; report sample size and uncertainty.
A failed case calls for the smallest effective change, not three new rules.
Recheck the failure and plausible regressions. Record actual outcomes separately
from predicted behavior. If runtime access or budget prevents a trial, report the
limit and finish the authorized structural work.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |