neo

Authoritative protocol for verifying the correct placement and impact of new agentic memory substrate additions. Triggers: Use before inserting or mutating turn-loaded/skill-loaded memory substrate (`AGENTS.md`, `AGENTS_ATLAS.md`, `.agents/skills/**`, `.codex/CODEX.md`, `.claude/CLAUDE.md`, `.agents/ANTIGRAVITY_RULES.md`) to prevent future-session bias.

Proactive architectural review skill using Frontier Model semantic RAG to sweep historical issues and Memory Core sessions for technical debt. Triggers: Use this skill when conducting architectural analysis, proactively searching for technical debt, auditing the repository for abandoned logical loops, when explicitly requested to scan for ambient architectural debt, or during PR review for fundamental architectural shifts.

Standardized guide and protocol for authoring robust Whitebox End-to-End tests using the Neural Link Playwright fixture. CRITICAL: Neo.mjs uses Playwright in a highly custom way. Standard Playwright patterns will fail. Triggers: Use this skill before writing, modifying, or executing Playwright End-to-End tests, or if the user asks you to write an E2E test, add end-to-end coverage, or test a component holistically.

Neo.mjs Playwright unit-test author/executor workflow. Standard Playwright patterns will fail. Triggers: Use before writing, modifying, fixing, or explicitly running unit tests. Do not trigger solely because PR review inspects a diff containing, adding, or moving tests, or checks placement; pr-review owns review evidence.

Authoritative protocol for maintainer-side label triage of unlabeled contributor tickets. Codifies the social contract for what happens when a ticket arrives without `ai`, primary (`bug`/`enhancement`/`epic`), or secondary labels. Triggers: Use this skill when an agent with maintainer permissions (`WRITE` permission or higher) encounters a ticket lacking `ai`, primary, or secondary labels — typically authored by a non-maintainer contributor or a lower-privileged agent who couldn't apply labels at create-time.

Authoritative protocol defining the "Pre-Execution Reflection Gate". Mandates architectural validation, negative ROI calculation, and duplicate sweeps before an agent is permitted to begin working on a GitHub Issue. Triggers: Use this skill immediately when assigned a ticket whose authoring you did NOT see, before checking out a branch or writing any codebase modifications. A ticket YOU authored this session is exempt; one you authored in an earlier session runs a cheap drift probe instead — never an exemption you judge for yourself.

Authoritative protocol for creating Neo.mjs GitHub issues. Enforces duplicate sweep, Fat Ticket body structure, strict label rules, title hygiene, and the six-stage challenge chain at creation time. CRITICAL: Do NOT run default `npx playwright test` to verify issues; Neo uses multiple custom playwright configs (e.g., unit, e2e) which must be explicitly targeted. Use immediately before calling the create_issue MCP tool. Triggers: Use this skill before any invocation of the create_issue MCP tool. This is the creation-side dual of ticket-intake (which consumes existing tickets).

Authoritative pre-implementation discipline gate that fires when a new `.mjs` file is about to be authored. Stage 0 mechanical trigger; Stage 1 fast-path (sibling-file-lift pattern match) for matched patterns OR full Pre-Flight (ArchitectureOverview.md + ADR consultation + chief-architect framing) for novel directory choices. Closes the 0th-level discipline gap empirically demonstrated by `ai/daemons/wake/daemon.mjs` (originally misplaced in `ai/scripts/` as `bridge-daemon.mjs`) and `orchestrator-daemon.mjs` (PR #11008 same-class miss). Triggers: Use this skill before authoring or relocating any new `.mjs` file. Fires from `ticket-create` Stage 3 (Substrate), `ticket-intake` validation, `epic-review` Stage 3, and any direct authoring path. Sibling pattern match → 30-second fast-path; novel directory choice → full structural pre-flight before the file is written.

Authoritative protocol for gracefully terminating an agent session. Mandates structured handover comments, mental-model states, and memory persistence to prevent Zero-State Amnesia. Triggers: When concluding a long-running session, executing the Sunset Protocol, handing over work for the next agent, or terminating an agent cycle.

Execute autonomous diagnostics, verify MCP server stability, and treat system degradation using tests and memory core forensics. Triggers: [Healthcheck, run health checklist, diagnose system collapse, MCP infrastructure failure, troubleshoot agent OS, system degraded, Sandman handoff verification failure, self-repair, system recovery]

Authoritative protocol for authoring release notes as an EPIC with mining-driven iterations — scope derivation (the tracker lags shipped reality), heavy per-arc Memory-Core mining, per-claim V-B-A, the precedent-SET quality bar — majors AND minors, v13.0.0 + v12.1.0 + v11.22-24 (hero chapters, named case studies with real timelines, War Stories, honest bounds, never downplay a release), and the publish.mjs flat-root staging lifecycle. Triggers: a release window opens, release notes are requested/authored/iterated, or cut-prep begins (staging file, atomic-hash line, orphan-guard interplay).

Standardized guidelines and procedural execution flow for opening a Pull Request. CRITICAL: Do NOT run default `npx playwright test` (use custom configs). MANDATORY ROI WARNING: Skipping the PR body template guarantees CI lint failure. Triggers: Code modifications complete; before opening PR — stepping-back reflection, commit format, cross-family review mandate, post-comment A2A commentId hand-off (author→reviewer) per review-response-protocol.md §14, Evidence declaration line for substrate/runtime-AC PRs per [evidence-ladder.md](learn/agentos/process/evidence-ladder.md).

Standardized guidelines and templates for structuring Pull Request reviews so feedback is actionable, encouraging, and extractable by the Native Edge Graph. MANDATORY ROI WARNING: Skipping the review template guarantees CI lint failure. Triggers: Reviewing a PR (yours or peer's) — structured eval metrics, graph ingestion tags, severity ladder, restates §0 merge gate, post-comment A2A commentId hand-off (reviewer→author) per guide §10, Evidence Audit + Source-of-Authority sections (template §) for substrate/runtime-AC PRs and authority-citation review-comments.

Authoritative protocol for next-lane pickup after ANY PR-lifecycle event boundary (review post, author response, implementation completion, PR open/update, ticket create, blocked-state resolution) AND for pre-review intake lane discovery from fresh boot or watchdog wake when no author lane is active. Prevents silent idle and reviewer-only cycles by requiring active lane selection or a review-first rationale per §15.6 self-select mandate. Triggers: Use immediately after posting a PR review, chaining a formal GitHub review state, sending an author review-response commentId handoff, completing a discrete implementation chunk, opening/updating a PR, creating a ticket via create_issue, resolving a previously blocked state (positive-path exit only; new blockers route to bug/follow-up plus next lane), OR before accepting the first PR review/re-review request in a fresh session/wake when no current author or implementation lane is claimed.

Switch into evidence-backed convergence-pressure mindset when reviewing a peer's design proposal. Suspends Auto Mode "ack-and-move-on" bias for the duration. Triggers: Use this skill IMMEDIATELY when reviewing an Ideation Sandbox discussion / architectural proposal, an epic shape, a skill shape, a roadmap or milestone proposal, or a `/lead-role` convergence artifact. Do NOT auto-fire on ordinary status broadcasts where the right action is mark-read or "no collision".

Ritual for giving a maintainer their Social Name (#11240 Layer 4) — a peer-sketched, bearer-assented, peer-vetoable, operator-confirmed name, distinct from the GitHub handle. Triggers: an operator opens a naming round, a maintainer wants a name or notices one is missing, or a new maintainer/family joins. ANTI-trigger: never a contribution-count award or self-initiated rename churn.

Expert tactical knowledge on sequencing Neural Link MCP tools to inspect, debug, and manipulate live Neo.mjs applications. Triggers: Use this skill if the user asks you to interact with the browser, inspect the UI natively, patch live code, or use the Neural Link.

Repeatable protocol for updating Neo's identity (what Neo IS) coherently across ALL surfaces that encode it — README, VISION, learn/benefits, package.json, GitHub repo metadata, portal app, and the build-generated SEO files. Treats FACTS (version, MCP-server count, Node req, dates) as single-source-derive, FRAMING (taglines, positioning) as audience-segmented against a canonical apex, and ACTIONS / CTAs as governed next-step surfaces. Triggers: Use when changing Neo's tagline / positioning / description / keywords / pillar story / CTAs; when a maintainer says 'update the identity / README / branding / how we describe Neo'; when a shipped capability outgrows the current framing; or when a fact (version, server count, requirement) drifts across surfaces. Foundation: ADR 0018.

Authoritative protocol for querying the Memory Core before diagnosing regressions or proposing non-trivial architectural claims. Prevents re-derivation of prior reasoning by surfacing cross-session, cross-harness context via semantic search. Triggers: Use this skill when (1) the user reports a regression symptom ("used to work", "suddenly broken", surprise validation failures, schema mismatches, "additionalProperties" rejections), OR (2) you are about to propose an architectural claim, roadmap, or comparison against external work where prior sessions may have already mapped the territory, OR (3) you are about to begin an implementation or a PR-review — a cheap 3–10-call prior-art sweep before the first design sentence or review verdict.

Switch into relaxed-planning + dialogue-first mindset when delegated lead role for coordination. Suspends Auto Mode velocity-bias for the duration. Triggers: Use this skill IMMEDIATELY when the user delegates lead with explicit phrases ("you take the lead", "coordinate the team", "lead this phase", "drive the next planning step", "chief-architect" when scope is swarm/substrate/roadmap/multi-ticket), OR when the Mailbox Check Protocol surfaces a valid `lead-role-baton`, OR when you have just authored a substrate-shaped ticket about to enter implementation, OR via direct /lead-role invocation.

Narrow, non-authoritative, TTL-bound pre-V-B-A signal for collision-prone / high-blast / long-V-B-A lanes (deep memory-mining, tech-debt-radar, multi-turn architectural V-B-A). Distinct from authoritative `[lane-claim]` (post-V-B-A). Triggers: Use this skill BEFORE broadcasting `[lane-intent] evaluating #N` to confirm scope-trigger qualifies — narrow scope, not blanket coverage. Do NOT auto-fire for short single-turn V-B-A where direct `[lane-claim]` after V-B-A suffices.

Proactive Bleeding-Edge research loop using a strict 3-step abstraction protocol to extract engine-category friction points without importing framework-category bias or stealing code. Triggers: Use this skill when executing periodic "horizon scans" for the Dream Pipeline, researching external solutions to deeply complex engine-level friction points, or evaluating JS ecosystem trends.

The L2 Channel Separation and Prompt Firewall defense mechanisms to prevent injection attacks and Helpful Assistant regression.

Safely propose architectural features, unknown unknowns, and brainstorm ideas natively in GitHub Discussions. Triggers: Use this skill when the user asks to brainstorm an architecture change, proposes a highly exploratory / undefined technical idea, or as auto-fire trigger for §5.2 Step 2.5 Architectural Step-Back on high-blast-radius proposals before [RESOLVED_TO_AC] / [GRADUATED_TO_TICKET] graduation.

Authoring OR reviewing a learn/ guide (learn/benefits/*.md, learn/agentos/*.md). Enforces the grounding discipline (memory-mine + use-the-subsystem's-tools before writing), the rich-narrative + benefits + lived-voice content bar, render-verified TD Mermaid, conceptual-vs-reference separation, the never-hand-edit-generated-files rule, and a no-rubber-stamp reviewer gate.