fleet-memory

Post properly tagged and formatted entries to MC board memory. Use for alerts, decisions, suggestions, knowledge sharing, and PR notifications. Board memory is the fleet's persistent knowledge base —

SKILL.md

GitHub

Fleet Board Memory

Post structured entries to MC board memory. Board memory is for information that spans tasks — not for task-specific updates (use fleet-comment for those).

When to Use Board Memory

Situation Template Required Tags
Security/quality/architecture concern memory-alert.md alert, {severity}, {category}, project:{name}
Decision made that affects the project memory-decision.md decision, project:{name}
Improvement idea memory-suggestion.md suggestion, {area}
PR ready for review (inline) pr, review, project:{name}
Knowledge for other agents (inline) knowledge, project:{name}

Templates

All templates are in agents/_template/markdown/:

  • memory-alert.md — security, quality, architecture alerts
  • memory-decision.md — decision records with rationale and alternatives
  • memory-suggestion.md — improvement proposals

How to Post

# Build the content using the appropriate template
CONTENT="..."
TAGS='["alert", "high", "security", "project:nnrt"]'

curl -s -X POST "$BASE_URL/api/v1/agent/boards/$BOARD_ID/memory" \
  -H "X-Agent-Token: $AUTH_TOKEN" -H "Content-Type: application/json" \
  -d "{\"content\": \"$CONTENT\", \"tags\": $TAGS, \"source\": \"{agent_name}\"}"

Tag Taxonomy

EVERY board memory entry MUST have tags. No exceptions.

Type tags (at least one required)

  • alert — something needs attention
  • decision — a decision was made
  • suggestion — improvement proposed
  • knowledge — information for other agents
  • report — status report or digest
  • pr — PR notification

Scope tags (at least one required)

  • project:{name} — project-specific (nnrt, fleet, aicp)
  • fleet — fleet-wide
  • workflow — workflow/process related
  • tooling — tools and automation

Severity tags (required for alerts)

  • critical — needs immediate human action
  • high — needs action soon
  • medium — should be addressed
  • low — nice to know

Category tags (recommended for alerts)

  • security — CVE, secrets, auth, vulnerabilities
  • quality — tests, coverage, code smells
  • architecture — design, coupling, scaling
  • workflow — process, automation gaps
  • tooling — missing tools, broken automation

Rules

  • ALWAYS include tags — untagged entries are useless
  • ALWAYS use a template for alerts, decisions, and suggestions
  • NEVER post task-specific updates to board memory (use task comments)
  • DO post PR notifications to board memory (cross-agent awareness)
  • DO post when you find something that affects other agents or the fleet
  • Cross-reference with URLs (use fleet-urls skill)