skills.rest
sumitake
sumitakeCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

agent-readiness

Evaluates whether an agent-runtime combination is ready for a proposed role.

Updated Jul 12, 2026
Search Tags:#multi-agent#governance#agent readiness#capability evaluation#runtime health#role promotion#review independence
One-click install
npx skills add https://github.com/sumitake/agent-collab --skill agent-readiness-sumitake
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-readiness
Source: https://github.com/sumitake/agent-collab/tree/main/plugins/agent-collab/skills/agent-readiness
Command: npx skills add https://github.com/sumitake/agent-collab --skill agent-readiness-sumitake

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Assigning an AI agent to a role like primary author, reviewer, or release agent based on a single success or raw model capability leads to failed handoffs, unreviewed merges, and silent fallback to weaker agents. This Skill provides a structured review process to decide whether a specific agent-runtime-role combination is genuinely ready for a given responsibility. ## Core Features & Use Cases - Multi-dimension readiness review: Assesses role fit, capability evidence, runtime health, tool boundaries, governance, reliability, and cost posture before promotion. - Structured verdicts: Produces APPROVE-ROLE, APPROVE-WITH-CONDITIONS, DO-NOT-PROMOTE, or NEEDS-OPERATOR decisions with explicit guardrails. - Use Case: Before letting Codex act as primary author on a repository, run this review to check CLI auth state, non-interactive behavior, prior task evidence, and review independence, then decide whether to promote it, require a pilot, or keep the role elsewhere. ## Quick Start Ask the assistant to evaluate whether Codex is ready to be the primary author for this repository using the agent readiness review.

Frequently Asked Questions about agent-readiness

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I evaluate if an AI agent is ready for a role?▼

Run a readiness review that names the exact role and scope, collects evidence from runtime checks, CI, and prior tasks, then compares capability needs against known agent strengths. The output is a verdict: APPROVE-ROLE, APPROVE-WITH-CONDITIONS, DO-NOT-PROMOTE, or NEEDS-OPERATOR.

What dimensions should an agent readiness review cover?▼

Cover role fit, capability evidence, runtime health (auth, quotas, non-interactive behavior), tool boundaries, governance and review independence, reliability and fallback behavior, and cost posture. Model capability alone is not sufficient evidence of readiness.

Can an agent review code it or its family authored?▼

No. Review authority requires independence from the author and artifact. Before dispatch, establish the primary and artifact-author families, check the proposed reviewer, and verify the observed reviewer family afterward before accepting independent evidence.

When should an agent not be promoted to a role?▼

Do not promote when evidence comes from tasks outside the target role, when runtime readiness (auth, tools, non-interactive behavior) is unverified, when review independence is missing, or when authority and risk acceptance exceed agent scope and require an operator.

What is the difference between model capability and runtime readiness?▼

Model capability is what the underlying model can do in principle; runtime readiness covers CLI availability, auth state, version currency, quotas, timeout handling, and non-interactive behavior. A capable model with a broken or unauthenticated runtime is not ready for the role.

Related Skills

Explore More Software Engineering Skills→
UKGovernmentBEISUKGovernmentBEIS
657

ci-maintenance-workflow

Diagnose and fix failing CI tests, smoke tests, and slow test markers in GitHub Actions workflows.

Official
Advanced
Jin9Jin9

model-selection

Assigns models and reasoning effort per agent role using privacy gates and cost-per-task criteria.

Community
Intermediate
ditchallawayditchallaway
1

subagent-driven-development

Dispatch independent subagents to execute plan tasks with two-stage reviews.

Community
Advanced
gfiresgfires

devex-review

Audit developer onboarding and CLI help text to produce a DX scorecard.

Community
Advanced