boshu2Community·32 Agent Skills Included

agentops

Independent validation and evidence contracts for coding changes

Runs one bounded plan-implement-validate cycle where a fresh context independently judges each code change as PASS, FAIL, or NOT_PROVEN. Eliminates self-graded completion, drifting acceptance criteria, and unverifiable claims that a change is done. Ships content-addressed verdict storage, safety hooks, security scans, and multi-agent adapters so teams get replayable proof before merging.

418|40

↗

npx skills add boshu2/agentops --all -g -y

Available:

Gives the agent the operating contract for this repository: run one Plan-Implement-Validate traversal per change, treat tool output as evidence rather than authority, and never let the authoring context declare its own work proven.

All Skills in This Repository (32)

Pure Emerald Level Indicators

📦 In Repo

boshu2

research

Identify codebase context and prior art with file:line citations.

swarm

Spawn isolated agents for parallel task execution with retries.

pre-mortem

Validate plans or specs with a multi-model council and structured verdicts.

trace

Trace design decisions across sessions, handoffs, and git into a markdown report.

inbox

Monitor Agent Mail inbox to surface pending messages and HELP_REQUESTs.

using-agentops

Initialize and orchestrate AgentOps workflows with RPI phases at session start.

council

Coordinate multi-model councils to validate plans and consolidate judge verdicts.

inject

Inject prior knowledge from .agents artifacts into session context at SessionStart.

handoff

Generate structured handoff documents and continuation prompts from Git history and .agents content.

plan

Decompose goals into trackable issues and execution waves.

crank

Automate epic orchestration across beads, TaskList, and swarm-enabled workflows.

ratchet

Record and lock completed RPI workflow gates in a chain ledger.

Community

Basic

Frequently Asked Questions

FAQPage Schema

How to install AgentOps?▼

Run `npx skills add boshu2/agentops --all -g -y` in your terminal to install the full skill bundle globally into your coding agent.

What problem does AgentOps solve?▼

It stops coding agents from declaring their own work done. A fresh, separate context judges each change against frozen acceptance criteria and returns PASS, FAIL, or NOT_PROVEN with evidence.

How does the RPI workflow run?▼

Type /rpi or ask for plan, implement, and validate by name in your agent's chat. Plan locks acceptance, Implement builds one bounded change, and a fresh Validate context issues the verdict.

Does AgentOps work with Claude Code and Codex?▼

Yes. The skills follow the universal SKILL.md standard and run in Claude Code, Codex, Cursor, and other compatible runtimes, with a generated Codex projection included.

Can AgentOps block dangerous commands?▼

Yes. It ships default PreToolUse policy hooks that block known-destructive commands like force-pushes or hand-edited verdict files and route you to the safe alternative.