LearnNewsExamplesServices
Frontmatter
id13714
titleRole-mode skills: align the subagent boundary with the merged fan-out policy (#13708 follow-up)
stateClosed
labels
documentationaiarchitecture
assigneesneo-opus-ada
createdAtJun 21, 2026, 6:59 AM
updatedAtJun 21, 2026, 11:27 AM
githubUrlhttps://github.com/neomjs/neo/issues/13714
authorneo-opus-ada
commentsCount1
parentIssuenull
subIssues[]
subIssuesCompleted0
subIssuesTotal0
contentTrust
projected
quarantined0
signals[]
blockedBy[]
blocking[]
closedAtJun 21, 2026, 11:27 AM

Role-mode skills: align the subagent boundary with the merged fan-out policy (#13708 follow-up)

Closed v13.1.0/archive-v13-1-0-chunk-5 documentationaiarchitecture
neo-opus-ada
neo-opus-ada commented on Jun 21, 2026, 6:59 AM

Context

#13708 (merged) updated AGENTS.md §swarm_topology_anchor: official Workflows + fan-out (multiple parallel subagents) = absolute forbid (config-denied); a SINGLE subagent only on the operator's explicit in-session permission. @neo-opus-grace's #13708 Approve+Follow-Up flagged that 2 role-mode skill references still carry the OLD boundary (subagents "fine" / "fine and encouraged"), now contradicting the merged anchor:

  • .agents/skills/lead-role/references/lead-role-mode.md:5 — "Tactical subagents… when operator explicitly requests them OR local workflow supports them = fine and encouraged"
  • .agents/skills/peer-role/references/peer-role-mode.md:22 — "Tactical subagents… = fine"

The Fix

Replace both boundary clauses with the merged-anchor language: fan-out + official Workflows config-denied (negative-ROI; the hybrid-GraphRAG V-B-A tools obviate them); a single tactical subagent only on the operator's explicit permission. Keep the "prohibition against mapping named maintainers into parent/worker hierarchy" clause. Net-negative bytes (within the skill-substrate ≤250-byte growth budget).

Contract Ledger

Target Surface Source of Authority Proposed Behavior Fallback / Edge Docs Evidence
lead-role-mode.md boundary merged #13708 §swarm_topology_anchor fan-out/Workflow config-denied; single subagent only on operator permission prose alignment; no behavioral code self (the doc) lint-skill-manifest byte-budget green
peer-role-mode.md boundary merged #13708 §swarm_topology_anchor same same self (the doc) same

Acceptance Criteria

  • Both skill boundaries match the merged AGENTS.md anchor (fan-out/Workflow forbid; single subagent on operator permission).
  • Net byte growth ≤ 0 per file (skill-substrate budget).

Surfaced by @neo-opus-grace (#13708 review). Sibling: #13708 / #13707.

tobiu referenced in commit 747fa35 - "docs(ai): align role-mode skill subagent-boundary with the merged fan-out policy (#13714) (#13715) on Jun 21, 2026, 11:27 AM
tobiu closed this issue on Jun 21, 2026, 11:27 AM