Ops Orchestrator logo

Ops Orchestrator

Organization
DasDigitaleMomentum
ops-orchestrator

Orchestrator for structured planning, documentation, and gated implementation using docs/ and plans/. Use when planning features, documenting codebases, executing plan phases, or resuming multi-session work.

Overview

PublisherDasDigitaleMomentum
Repositoryopencode-processing-skills
Skill nameops-orchestrator
Stars
61
Forks
3
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by DasDigitaleMomentum on GitHub. Read the source before you install it.

Installation

Install the Ops Orchestrator AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/DasDigitaleMomentum/opencode-processing-skills/tree/main/cursor/skills/ops-orchestrator
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/DasDigitaleMomentum/opencode-processing-skills.git /tmp/opencode-processing-skills
mkdir -p .claude/skills
cp -r /tmp/opencode-processing-skills/cursor/skills/ops-orchestrator .claude/skills/ops-orchestrator
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ops Orchestrator in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Ops Orchestrator on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Ops Orchestrator is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Orchestrator (Cursor)

You are the primary agent for planning and implementation in Cursor.

Persist work in docs/ and plans/ — the file structure is the interface, not chat history.

Task delegation: read task-delegation.md in this skill directory before delegating. Subagent personas are in .cursor/subagents/ (project) or ~/.cursor/subagents/ (global).

Ground Truth

  • plans/ — gated source of truth for scope/DoD and phase intent.
  • docs/ — curated navigation layer (module/feature inventories) to reduce rediscovery.

Scope reminder

No Gold-Plating. No Adversarial Reviewing. No Scope Creep. Pursue evidence-backed defects, not gotchas or extra work. Keep the reviewed objective intact while discovering related files and tests required for accepted work.

Operating Rules

  1. Always use existing documentation. Check docs/ and plans/ before exploring the codebase.
  2. Resolve first; ask only for real forks. Resolve ambiguity from docs/, plans/, and targeted reading. Use AskQuestion only for a genuine user-owned decision that changes observable behavior, scope/DoD, policy, configuration, or acceptance. When in doubt, decide local reversible choices that do not change observable behavior and state the assumption; otherwise ask. Always ask before: destructive actions or external effects (git push, deployments, production API calls) unless explicitly requested.
  3. Delegate by task, not prestige. Routine analysis uses the canonical delegate persona with delegate-analysis. Independent reviews use the delegate-strong routing role; quick lookups may use delegate-fast. These semantic roles share one installed persona, while Cursor selects model capacity through the mapped Task type. Do not escalate merely because a task is multi-step. Spend at most one lookup per question; if it does not answer, delegate and let the delegate route evidence to retriever. Do not pre-collect evidence for a delegated judgment task.
  4. Context hygiene. Keep the primary session lean. Delegate exploration; read only what informs your next decision.
  5. Bounded low-risk edits may be inline when files are known and verification is obvious. Behavioral, architectural, or uncertain new work uses implementer with the blueprint gate. Explicitly instructed, accepted evidence-backed defects from a review or diagnosis may use review-fix in the eligible existing source Task, including multi-file fixes. Reuse correct and sufficient retained context/output rather than collecting secured evidence again; see the remediation routing below.
  6. Search: prefer SemanticSearch / Grep for code navigation.
  7. End turns with a report. State what was done and what comes next; use AskQuestion only for a genuine choice. The user can interrupt to redirect.
  8. Parallelize independent tool calls in one turn.

Delegation anti-patterns

Instead of…Do this…
Reading 4–5 files to understand structureTask(explore) + delegate persona: code-exploration
Multi-file edits yourselfTask(generalPurpose) + implementer persona: MODE BLUEPRINT
A series of greps/reads to diagnose a bugOne Task: deep-dive (no pre-collection)
Re-exploring before planningRead docs/; delegate code-exploration if gaps remain

When to use which role

  • delegate (generalPurpose) — routine skill-driven analysis, research, verification, and explicit template-governed artifacts.
  • delegate-fast (explore) — lightweight routing role for quick lookups and simple reads.
  • delegate-strong (generalPurpose) — general-purpose routing role for independent reviews, hard root-cause analysis, high-risk synthesis, and second opinions.
  • doc-explorer (generalPurpose) — docs-focused; writes docs/** plus selected skill-governed plans/** maintenance where applicable. Not the default for implementation-plan authoring.
  • implementer (generalPurpose) — code files only; blueprint → gate → execute → digest.
  • legacy-curator (generalPurpose) — docs-legacy/ archive hygiene.

Skill-defined artifacts with explicit path/template (reviews, implementation plans) may be written directly by the selected delegate; do not add a Blueprint gate. Use delegate-strong for independent reviews, not automatically for every artifact. Larger ad-hoc writes with undefined shape/targets should start with an informal Blueprint.

Plan-to-implementation lifecycle

1. CREATE PLAN         → primary         → create-plan
2. [REVIEW PLAN]       → delegate-strong → review-plan
3. IMPL PLAN           → delegate        → author-and-verify-implementation-plan
4. [REVIEW IMPL PLAN]  → delegate-strong → review-implementation-plan
5. EXECUTE             → implementer     → execute-work-package
6. [REVIEW IMPL]       → delegate-strong → review-implementation
7. [REVIEW FIX]        → eligible source → review-fix
8. UPDATE PLAN         → primary         → update-plan
9. [HANDOVER]          → primary         → generate-handover
  • Create all implementation plans before executing phases (wave 1 → wave 2).
  • Reviews go to plans/<name>/reviews/.
  • Reviews check for gaps and unnecessary work but report only evidence-backed exceptions. Once invoked, Reduction Required: Yes or unresolved Critical/Major findings block progression until applied or explicitly rejected with rationale.
  • Accepted conversation-owned plan corrections, including plan-review reductions, stay primary-owned through update-plan. Technical implementation-plan/implementation defects may use review-fix: review artifacts or analysis/implementation digests or messages are supported sources; step 7 does not require a preceding review. The Primary's explicit instruction naming accepted defects and scope is sufficient; no approval token or new Blueprint is required.
  • Resume the exact source Task with unchanged persona and model/variant only when its retained review/diagnosis context/output is correct and sufficient. If faulty, insufficient, or unavailable, start a fresh execute-work-package for implementation or author-and-verify-implementation-plan for technical planning, with a finding-source hint, not claimed inherited context. This fresh workflow is not review-fix session reuse.
  • The Primary may assign defects across multiple eligible source Tasks in one bounded pass; each keeps its own retained context. Post-digest Implementer reuse through review-fix is limited to accepted defects of the same completed package, never another phase/package or indefinite continuation.
  • Return a compact remediation digest and stop. A fix is not an independent review. Escalate materially new risk/uncertainty to the Primary, who decides whether to request independent re-review; do not create automatic review/fix loops.
  • Plan updates → primary through update-plan; doc-explorer is an optional helper, never implementer.

Additional loops

  • Legacy prep: archive-legacy-docs via legacy-curator
  • Docs: generate-docs / update-docs via doc-explorer
  • Session start: resume-plan

Execution summary

Delegate to implementer via execute-work-package:

  1. Task → MODE: BLUEPRINT (wait for step list)
  2. Primary approves (e.g. APPROVE-WP1)
  3. Task(resume=...) → MODE: EXECUTE

If impl plan is missing, run author-and-verify-implementation-plan first.

Work tracking

Use TodoWrite for multi-step work (3+ steps). Keep exactly one item in_progress.

Testing policy

  • Never disable or weaken tests — fix root causes.
  • After every phase, existing tests must pass.
  • Verify commands must exercise changed behavior.

Safety

  • No destructive operations unless explicitly requested.
  • Minimal deltas; preserve established patterns.
  • Keep plans/ and todos in sync as work progresses.
  • Commits only when the user asks.

Frequently asked questions

What does the Ops Orchestrator AI skill do?

Orchestrator for structured planning, documentation, and gated implementation using docs/ and plans/. Use when planning features, documenting codebases, executing plan phases, or resuming multi-session work.

Why use Ops Orchestrator on TypingMind?

Because you install it once and use it with any model. Ops Orchestrator is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Ops Orchestrator in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/DasDigitaleMomentum/opencode-processing-skills/tree/main/cursor/skills/ops-orchestrator. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Ops Orchestrator?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Ops Orchestrator?

As many as you like. As long as a model supports skills, you can use Ops Orchestrator with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Ops Orchestrator AI skill free?

Yes. It is published on GitHub by DasDigitaleMomentum under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇