# Protocols

> >

- **Type:** Skill
- **Install:** `agentstack add skill-supermodo-skills-protocols`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [supermodo](https://agentstack.voostack.com/s/supermodo)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [supermodo](https://github.com/supermodo)
- **Source:** https://github.com/supermodo/skills/tree/main/skills/protocols

## Install

```sh
agentstack add skill-supermodo-skills-protocols
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# supermodo protocols — single source of truth + help

This skill is the ONE place shared protocol text lives. Other supermodo
skills read these masters directly via their sibling path
(`../protocols/references/.md`, resolved relative to each skill's own
folder; under the Claude Code plugin, equivalently
`${CLAUDE_PLUGIN_ROOT}/skills/protocols/references/.md`). Nothing is
duplicated — editing a master here changes behavior everywhere.

## When invoked, do one of three jobs

1. **Explain** — the user asks how something works: answer from the
   protocol masters below and the package README; quote the governing
   passage, link the file. Don't paraphrase from memory when the master is
   one Read away.
2. **Overview** — the user wants the big picture: summarize the toolkit
   (documentation-driven development with twin-model adversarial checks),
   the 13 skills, and the 8-stage `flow` pipeline, pointing at the skills
   involved.
3. **Doctor** — the user wants their installation checked: verify all 13
   skill folders are present as siblings of this one (a missing sibling
   means a partial install — recommend installing the full package), that
   `references/` here contains the 8 masters, and, inside a project, that
   `skills.config.json` validates (defer to the `config` skill) and the
   docs convention is in place (`docs-check`, via `librarian`). Report
   findings; fix nothing yourself.

## The masters

| File | Governs |
|------|---------|
| `references/config.md` | `skills.config.json` contract: versioning, validation, argv commands, env-var namespacing, defaults |
| `references/docs-convention.md` | The strict docs layout: router, work triads, immutable task IDs, ADRs, generated navigation, single doc owner |
| `references/reports.md` | Where and how skills persist output: run dirs, stage report frontmatter, run-state hashes, symlink containment |
| `references/handoff.md` | Subagent stages: file-based handoff, needs-input escalation, failure protocol |
| `references/cross-model.md` | Running the other provider as adversary: read-only sandboxes, preflight, batching, hung-detection, degradation honesty |
| `references/adversarial-review.md` | The review stance: burden of proof, evidence-cited attacks, valid "no objection", dispute surfacing |
| `references/questions.md` | Asking the user: three-class triage, ordered-choice format (both models' views, `More detail`, own answer), transport, answer recording & auto-resolution |
| `references/grilling.md` | Twin-agent adversarial interview: actors, independent plans, disprove rounds, routing, recording |

## Communication rule (package-wide)

Chat reporting defaults to extreme concision: "be extremely concise and
sacrifice grammar for the sake of concision." Overridable per project via
`output.verbosity` in `skills.config.json` (`"standard"`). This NEVER
applies to generated artifacts (docs, ADRs, reports, commit messages),
never shrinks protocol-mandated formats (grill explanations, question
loops), and never compresses safety output (warnings, destructive
confirmations, mutation previews).

## Precedence

1. The user's explicit instructions
2. These protocols
3. Individual skill defaults

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [supermodo](https://github.com/supermodo)
- **Source:** [supermodo/skills](https://github.com/supermodo/skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-supermodo-skills-protocols
- Seller: https://agentstack.voostack.com/s/supermodo
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.