aops-cli-core

Guides agents through AOPS CLI discovery, guard flags, sync, and archive lifecycle operations.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/eeemzs/aops --skill aops-cli-core-eeemzs
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: aops-cli-core
Source: https://github.com/eeemzs/aops/tree/main/assets/skills/aops-cli-core
Command: npx skills add https://github.com/eeemzs/aops --skill aops-cli-core-eeemzs

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Agents operating the AOPS CLI often guess flags, misuse caches as sources of truth, or skip required write guards, causing failed commands and unsafe destructive operations. This Skill provides a thin operator playbook that routes agents to the correct discovery ladders, guard patterns, and canonical sources. ## Core Features & Use Cases - Help-First Discovery: Enforces a decision chain from --help through compact tool catalogs (agent tools --summary) and schema inspection before any raw hosted invoke. - Guard Flag Discipline: Standardizes use of --preview, --apply, --confirm, --idempotency-key, --json, and --yes so writes and destructive actions are always explicitly gated. - Server-Canonical Routing: Clarifies that PM, memory, discuss, and experience authoring go through the hosted server, while .aops-cache/** directories are read-only mirrors refreshed by partitioned sync pull. - Project Registry & Archive Lifecycle: Covers multi-project linking via project link, scoped sync with --project-slug or --all-projects, and the guarded archive create/verify/delete/decommission-check flow. - Use Case: An agent needs to refresh one project's local cache and then archive its PM graph. It runs sync pull --project-slug aops --apply --json, then archive create, archive verify --apply, and only proceeds to archive delete --apply --confirm after verification passes. ## Quick Start Ask the agent to use the aops-cli-core discipline to discover the right aops-cli command with --help and run it with the proper guard flags for your project slug.

Frequently Asked Questions about aops-cli-core

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

FAQPage Schema
How do I discover the right aops-cli command for a task?▼

Start with `aops-cli --help`, then `aops-cli <family> --help` and nested subcommand help. If no sugar command exists, inspect the hosted catalog with `aops-cli agent tools --domain <domain> --summary --json`, then check the compact schema with `agent schema --summary` before invoking.

What guard flags does aops-cli require for write operations?▼

Reads need no guard, writes require `--apply`, and destructive writes require `--apply --confirm`. Use `--preview` for planning output, `--idempotency-key` for deterministic retries, and `--json` for scriptable output. `--yes` only makes commands non-interactive and never substitutes for `--apply` or `--confirm`.

Can I edit files in .aops-cache to change project state?▼

No. Directories under `.aops-cache/` (projectman, agentspace, hosted, docman) are read-only mirrors of server state. Author changes through the hosted CLI commands like `pm`, `mem`, or `discuss` with `--project-slug`, then refresh the cache with `sync pull`.

How do I sync multiple linked projects in one repository?▼

Link each project with `aops-cli project link --slug <slug>`, then use `sync pull --project-slug <slug> --apply --json` for one project or `sync pull --all-projects --apply --json` for all. Use `--only projectman|agentspace|hosted-skills|hosted-prompts` to scope the refresh to one partition.

Why does archive delete refuse to run on my manifest?▼

Archive delete requires a verified manifest, so run `archive verify --apply` first to persist verification status. Destructive hosted deletion also needs `--apply --confirm`, and `decommission-check` stays blocked while `pendingDomains` is non-empty.

When should I use raw agent invoke instead of sugar commands?▼

Use `agent invoke` only after confirming the contract via `agent schema --summary` or full schema/OpenAPI when no sugar exists or sugar returns validation errors. Stop guessing flag variants on 400 errors; compare the sugar payload against the schema first, and treat raw invoke as a temporary bypass.