Record Business Artifact

Write a structured markdown file to .oh/ with YAML frontmatter. Use the templates below for each type.

Arguments

$ARGUMENTS should be: <type> <slug> [options]

Example: /rna-mcp:record metis protocol-mismatch-hangs

Templates

Metis (learning)

Write to .oh/metis/<slug>.md:

---
id: <slug>
title: "<title>"
outcome: <related-outcome-id>
---

<body — what was learned and why it matters>

Signal (measurement)

Write to .oh/signals/<slug>.md:

---
id: <slug>
outcome: <related-outcome-id>
type: slo|metric|qualitative
threshold: "<measurable threshold>"
---

<body — what this measures and how>

Guardrail (constraint)

Write to .oh/guardrails/<slug>.md:

---
id: <slug>
severity: candidate|soft|hard
statement: "<one-line constraint>"
outcome: <related-outcome-id>
---

<body — rationale for this constraint>

Outcome (update existing)

Edit the existing file at .oh/outcomes/<slug>.md — update status, mechanism, or files in the frontmatter.

Process

Parse $ARGUMENTS to determine type and slug
Check if the file already exists — if so, confirm before overwriting (metis/signal/guardrail) or merge updates (outcome)
Read one existing artifact of the same type for frontmatter format reference
Write the file using the Write tool
Confirm: "Recorded at .oh/<subdir>/<slug>.md"

Slug Rules

Lowercase, alphanumeric + hyphens only
No path separators (/, \, ..)
Example: protocol-mismatch-hangs, agent-scoping-accuracy

ADRs (architecture decisions)

When recording an ADR, write to docs/ADRs/<NNN>-<slug>.md with YAML frontmatter plus markdown body. Keep the ADR prose canonical, and make executable validation declarations match plugin/skills/adr-sync/SKILL.md.

ADR template

---
id: <NNN>-<slug>
status: proposed|implementing|implemented|superseded
validate:
  cargo_tests:
    - <exact cargo test name from `cargo test -- --list`>
  audits:
    - <exact built-in audit name if the claim is structural rather than behavioral>
  smoke:
    - <fixture path if needed>
  scripts:
    - <exact script path only if a normal test or audit cannot honestly express the claim>
---

# <Decision Title>

## Context
<why this decision exists>

## Decision
<what was decided>

## Consequences
<trade-offs and follow-on effects>

ADR rules

Prefer cargo_tests over every other check type whenever the claim can be expressed as a normal test
Use built-in audits for code-shape constraints that are not honest test cases
Use direct executable references only — no opaque evidence IDs
If a validation does not exist yet, leave it out and call out the missing check explicitly in the ADR body or follow-up work
After creating or updating an ADR, run /rna-mcp:adr-sync so frontmatter, compile/check output, and missing-check reporting stay aligned