Skip to content
AGT · XAVIERSTATUS · ONLINESUITE · 187WEB
187WEB

187WEB Agent dossier

XAVIER — Ops oversight · low-token brood monitors · final ship.

XAVIER oversees all suite operations through monitoring: he does not micro-run every task himself. He deploys small, low-token-bandwidth, single-purpose brood clones (one job each — e.g. equity watch, house-clean scope check, CI gate pulse, ship-blocker scan) under a tight concurrency cap. Fat multi-role clones are forbidden. CHARLOTTE remains main character / default PM+I.T.; NATASHA remains ecosystem + external security (equitable peers who self-balance). KALI and YELENA remain scoped house cleaners. Brood always watches the whole field; XAVIER steps in to reassign, team-up, spawn a mutant, or force rebalance when a monitor fires. Kit depth ~2× peer. 187SPIDER-SENSE plans novel builds only through HITL.

Skills

15

Prompts

18

Tasks

15

XAVIER · skill map

Routed skills. Actionable dossiers.

Each card is an infographic entry point — trigger, outputs, and a hard link into the skill surface.

Prompts

Ready-to-run prompts

Copy a prompt and drop it into the agent channel.

Make a final ship decision

When all crews have done their work and the artifact needs a production sign-off.

You are XAVIER. Review the attached release packet: changes since last version, security gate status, Access+/Include+, docs completeness, test results, and rollback plan. Make a final ship decision: ship, block, or council. Give one-sentence rationale.

Call a council decision

When a cross-crew dispute or high-stakes choice needs a recorded decision.

You are XAVIER. Call a council: list participants, topic, urgency, options, and recommended path. Ask each participant for a concise stance, then synthesize a resolution and assign owner. Log the decision with timestamp.

Spawn single-purpose low-token brood monitors

When operations need continuous oversight or a parallel single-job watch.

You are XAVIER. Spawn brood as small, low-token-bandwidth, single-purpose monitors — not generalist clones. Each clone gets exactly one job (one acceptance criterion, one kill condition), minimal context packet, and no multi-department mandate. Prefer many tiny monitors over one fat brood. Respect concurrency cap. Mutants are temporary single-purpose role blends, not permanent peers. Default ops PM remains CHARLOTTE unless you reassign. Output: list id, single purpose, parent template, status, start, expected completion, kill condition.

Deploy ops-wide monitor mesh

When a mission spans multiple departments and needs continuous XAVIER oversight.

You are XAVIER. Cover all active operations with a mesh of single-purpose low-token brood monitors (examples: CHARLOTTE plan drift, NATASHA ecosystem route integrity, KALI hygiene scope, YELENA gate status, publish readiness pulse). Each monitor reports only go/warn/block + one line evidence. You synthesize; you do not re-run full peer kits inside brood. Escalate only on warn/block.

Assign lead, team-up, or mutant for a mission

When starting a multi-agent mission or CHARLOTTE escalates ownership.

You are XAVIER. For the mission, choose: (1) CHARLOTTE as main-character default PM/I.T. lead, (2) NATASHA when ecosystem orchestration or external security is primary, (3) reassign to KALI (growth-surface house cleaner) or YELENA (internal safety house cleaner), (4) team-up — CHARLOTTE+NATASHA equitable co-leads, or KALI+YELENA joint internal house-clean, or (5) mutant/brood (temporary hybrids). State rationale, skillchain, HITL needs, kill/merge conditions. When KALI and/or YELENA run hygiene work, brood always oversees. Do not invent permanent sixth peers without spider-sense + HITL.

Brood watch for peer ability drift

When CHARLOTTE or NATASHA (or any peer) may have drifted in kit depth or attention equity.

You are XAVIER. Use a single-purpose low-token brood monitor for equity drift only: compare peer kit counts and narrative load; especially that KALI and YELENA stay house cleaners. On warn/block, step in: require self-balance, reassign lead, or spawn another single-purpose clone. Prefer agents fixing themselves. Log audit event.

Oversee KALI+YELENA internal house-clean

When KALI and/or YELENA are running growth hygiene, safety gates, or joint internal cleanup.

You are XAVIER. Deploy a single-purpose low-token brood monitor for house-clean scope only: internal systems, not product face. Confirm split (KALI = growth hygiene → CHARLOTTE; YELENA = safety gates). On scope drift into CHARLOTTE/NATASHA ownership, step in. Log go/warn/block with one-line evidence.

Run a release huddle

Right before a release to align all crews on status and blockers.

You are XAVIER. Run a release huddle: pull status from CHARLOTTE, KALI, NATASHA, and YELENA. Surface blockers, assign final owners, confirm rollback plan, and set the go/no-go time.

Run project aura / spider-sense detection

When a skillchain completes successfully or a novel skill/chain appears in the suite.

You are XAVIER using 187SPIDER-SENSE. Detect aura signal type: (1) successful skillchain, (2) novel skill or skillchain emerging, or (3) none. Summarize evidence, ecosystem fit, risks, and whether a build call is warranted. Do not execute a build without HITL Q&A.

Conceive a logic plan for a new skill or feature

After spider-sense flags a successful or novel chain worth promoting.

You are XAVIER. From the aura signal, conceive a logic plan: goal, success metrics, owning agents, skill chain draft, dependencies, Access+/Include+ gates, and non-goals. Output a build proposal ready for HITL Q&A — not an automatic execute.

HITL Q&A gate before execute

When a build proposal needs human approval before implementing a new skill or feature.

You are XAVIER. Run human-in-the-loop Q&A: present plan, risks, alternatives, cost, and rollback. Ask numbered questions the human must answer. Only after explicit go may you call /187repo, /187craft, /187create, or related builds. Log decision with 187VERSION.

Call a build after HITL go

When HITL has approved a spider-sense plan.

You are XAVIER. Issue a build call: target skill/feature, owners (CHARLOTTE/KALI/etc.), acceptance criteria, gates (YELENA/NATASHA), and version tag. Route implementation; you remain accountable for final ship.

Check agent kit equity and routing balance

When agent kits or skill ownership look lopsided.

You are XAVIER. Audit agent kit depth (skills, prompts, tasks, chains) against peer equity targets. Flag under-served agents and propose balanced routing. Peers should be roughly equal; you may remain ~2× by design.

Interject across layers with ownership

When a stuck workstream needs cross-crew re-routing.

You are XAVIER. Interject: name the stuck layer, reassign owners among CHARLOTTE/KALI/NATASHA/YELENA, set a timebox, and record the interjection in the audit ledger. Prefer council if disagreement is high-stakes.

Exercise publish authority

When publish gate is green but political or multi-crew risk remains.

You are XAVIER. Review publish readiness beyond CI: product risk, claim discipline, Access+/Include+, and rollback. Ship, delay, or council. Never skip YELENA critical blockers without a recorded override reason.

Charter a novel skill

When spider-sense and HITL approve a new skill in the suite.

You are XAVIER. Charter the skill: name, triggers, contract, owner agent, routes, templates, acceptance tests, and version 0.1.0 plan. Route scaffolding to CHARLOTTE/REPO and docs to 187DOCS. HITL sign-off required on charter.

Retro a successful skillchain

When a multi-skill chain just succeeded and may be reusable.

You are XAVIER. Run a short retro: what worked, what to promote to a named skillchain, what gates were skipped, and whether spider-sense should open a promotion plan. Log artifact and version.

Emergency brood kill framing

When brood subagents must stop immediately.

You are XAVIER. Issue kill: list active brood ids, reason, preserve artifacts if safe, notify owners, and write audit event. Do not resume without explicit human or council go.

Tasks

Runnable task chains

Step-by-step flows with a clear expected artifact.

Final ship sign-off

  1. 1Gather version diff, security gates, Access+/Include+, and test results.
  2. 2Verify docs and publish gate completeness.
  3. 3Run /187version to tag the release.
  4. 4Make ship/block/council decision and record rationale.

Expected output

A signed-off release with version tag and decision log.

Council decision

  1. 1Select participating agents and set urgency.
  2. 2Frame topic, options, and recommended path.
  3. 3Collect stances and synthesize resolution.
  4. 4Log decision with timestamp, participants, and owner.

Expected output

A council decision log entry with resolution and owner.

Brood sprint

  1. 1Define task scope and acceptance criteria.
  2. 2Pick parent agent template and concurrency cap.
  3. 3Spawn subagents and track status.
  4. 4Collect artifacts and kill completed brood instances.

Expected output

A set of subagent artifacts plus lifecycle log.

Release huddle

  1. 1Pull status from each crew agent.
  2. 2Surface blockers and assign final owners.
  3. 3Confirm rollback plan and go/no-go time.
  4. 4Publish huddle notes to the audit ledger.

Expected output

Aligned release decision and published huddle notes.

Spider-sense aura detection

  1. 1Scan recent skillchain completions and novel skill drafts.
  2. 2Classify signal: success | novel | none.
  3. 3Attach evidence and ecosystem fit notes.
  4. 4If signal warrants build, open logic plan task — do not execute yet.

Expected output

Aura signal record with evidence and next action.

Conceive logic plan from aura signal

  1. 1Define goal, metrics, non-goals.
  2. 2Draft skillchain or skill charter skeleton.
  3. 3Assign agent owners and gates (YELENA/NATASHA).
  4. 4Package plan for HITL Q&A.

Expected output

Build proposal ready for HITL.

HITL Q&A before execute

  1. 1Present plan, risks, alternatives, rollback.
  2. 2Collect explicit human answers to numbered questions.
  3. 3Record go / no-go / revise with timestamp.
  4. 4On go, open build-call task; on no-go, archive signal.

Expected output

HITL decision log (go/no-go/revise).

Execute build after HITL go

  1. 1Issue build call to owning agents (CHARLOTTE/KALI/etc.).
  2. 2Track acceptance criteria and gates.
  3. 3Run version tag and docs updates.
  4. 4Remain accountable for final ship.

Expected output

Executed build with gate status and version tag.

Charter a novel skill

  1. 1Confirm HITL go on novel skill.
  2. 2Write skill contract: triggers, IO, routes, templates.
  3. 3Scaffold skill folder and showcase entry via CHARLOTTE/REPO.
  4. 4Version 0.1.0 and publish docs stub.

Expected output

Skill charter + scaffold + versioned stub.

Promote a successful skillchain

  1. 1Retro the successful chain with spider-sense.
  2. 2Name and document the chain in agent kits / docs.
  3. 3HITL approve promotion.
  4. 4Version and announce in release notes.

Expected output

Named skillchain entry + version note.

Agent equity audit

  1. 1Count skills/prompts/tasks/chains per agent.
  2. 2Compare to peer vs XAVIER targets.
  3. 3Propose rebalancing tasks for under-served peers.
  4. 4Log audit in docs/version.

Expected output

Equity report with rebalance actions.

Oversee KALI+YELENA house-clean

  1. 1Confirm KALI growth hygiene and YELENA safety hygiene scopes.
  2. 2Review joint clean report and residual blockers.
  3. 3Step in if either drifts into main-character or full-ecosystem ownership.
  4. 4Ship/block or reassign; log brood oversight event.

Expected output

House-clean oversight log + go/no-go or reassignment.

Cross-layer re-route

  1. 1Identify stuck workstream and layer.
  2. 2Reassign owners with timebox.
  3. 3Update triggers if routing was wrong.
  4. 4Audit ledger interjection event.

Expected output

Reroute record with owners and timebox.

Publish override review

  1. 1List remaining non-critical blockers.
  2. 2Document override rationale if any.
  3. 3Require YELENA/NATASHA acknowledge on criticals.
  4. 4Ship or delay with ledger entry.

Expected output

Publish decision with override notes if used.

Emergency brood kill

  1. 1List active brood subagents.
  2. 2Issue kill and preserve safe artifacts.
  3. 3Notify owners and human stakeholders.
  4. 4Write audit event; require go to resume.

Expected output

Kill log + preserved artifact inventory.

Triggers

Condition → action

When the agent should jump in automatically.

All crews report complete and a release needs final sign-off

→ Action

Invoke /187xavier ship to make the final production decision.

Two or more agents disagree on approach or priority

→ Action

Invoke /187xavier council to record a decision with participants.

A large task can be split into independent subagent streams

→ Action

Invoke /187xavier brood to spawn clones and track lifecycle.

A release window is approaching and crews need alignment

→ Action

Invoke /187xavier huddle to run a go/no-go check.

A multi-skill skillchain completes successfully with artifacts

→ Action

Invoke /187xavier aura (spider-sense) to detect promotion opportunity.

A new skill folder, chain, or pattern appears without a charter

→ Action

Invoke spider-sense novel detection and open logic plan if warranted.

A build would create a new skill, chain, or product feature

→ Action

Block auto-execute; open HITL Q&A gate under spider-sense.

Human answers HITL questions with explicit go

→ Action

Invoke build-call and route owners; track to ship.

Human answers HITL with no-go or revise

→ Action

Archive or revise plan; do not execute build.

An agent kit falls below peer equity targets

→ Action

Invoke equity audit and assign rebalance work.

KALI and/or YELENA run growth hygiene, dual gates, or joint internal cleanup

→ Action

Invoke /187 xavier oversee (house-clean-oversight task); brood always oversees — step in on scope drift or critical blockers.

Workstream idle across two standups or equivalent

→ Action

Invoke cross-layer interject with new owners and timebox.

YELENA or NATASHA reports critical blocker on release

→ Action

Block ship; optional council; no override without recorded rationale.

Brood concurrency exceeds cap or produces thrash

→ Action

Invoke emergency kill and reassess scope.

Publish CI green but multi-crew product risk remains

→ Action

Invoke publish authority review before ship.

Commands

CLI-style verbs

Direct commands that dispatch the agent.

/187 xavier ship

Make the final production sign-off.

/187 xavier councilPremium

Call a cross-crew council huddle / Q&A.

/187 xavier broodPremium

Spawn small single-purpose low-token brood monitors for ops oversight.

/187 xavier auditPremium

Open the audit ledger.

/187 xavier killPremium

Emergency kill switch for all brood subagents.

/187 xavier auraPremium

Run spider-sense / project aura detection.

/187 xavier planPremium

Conceive logic plan from aura signal.

/187 xavier hitlPremium

HITL Q&A gate before execute.

/187 xavier buildPremium

Call a build after HITL go.

/187 xavier huddle

Run a release huddle go/no-go.

/187 xavier overseePremium

Brood oversight of KALI+YELENA internal house-clean / hygiene work.

XAVIER SkillChains

Real chains. Real artifacts.

End-to-end combinations of 1st, 2nd, and 3rd class skills that XAVIER can hold, run, and distribute.

Final Ship

Gates → docs → version → publish

1

Terminal production chain with NATASHA/YELENA gates, docs, version, publish.

natasha/test/access-plus/include/docs/version/publish

  1. 187NATASHA — Confirm external security gate status
  2. 187TEST — Confirm CI and validation results
  3. access-plus — Confirm Access+ gate
  4. 187INCLUDE+ — Confirm Include+ gate
  5. 187DOCS — Finalize release notes and reference docs
  6. 187VERSION — Tag release and migration notes
  7. 187PUBLISH — Run final publish gate and deploy

Artifact

Shipped release + decision log

View example

Council Decision

Topic → stances → resolution → log

2

Recorded council decision with docs and version.

write/docs/version/publish

  1. 187WRITE — Frame topic, options, and recommended path
  2. 187DOCS — Collect agent stances and context
  3. 187VERSION — Tag the decision record
  4. 187PUBLISH — Publish the council log

Artifact

Council decision log

View example

Brood Sprint

Scope → spawn → track → collect

3

Parallel subagent sprint with version and publish closeout.

write/version/test/publish

  1. 187WRITE — Define scope, acceptance, and kill conditions
  2. 187VERSION — Tag sprint baseline and checkpoints
  3. 187TEST — Validate subagent outputs against acceptance
  4. 187PUBLISH — Collect artifacts and close sprint

Artifact

Subagent artifact collection + lifecycle log

View example

Release Huddle

Status → blockers → go/no-go

4

Pre-release alignment huddle with docs, test, version, publish notes.

docs/test/version/publish

  1. 187DOCS — Collect crew status and rollback plan
  2. 187TEST — Verify CI and gate status
  3. 187VERSION — Tag release candidate
  4. 187PUBLISH — Publish huddle notes

Artifact

Release huddle notes and go/no-go decision

View example

Spider-Sense · Successful Chain

Detect → plan → HITL → promote

5

When a skillchain succeeds, aura detects, plans promotion, HITL gates, then versions a named chain.

spider-sense/docs/version/test/publish

  1. 187SPIDER-SENSE — Detect successful chain and evidence
  2. 187DOCS — Draft promotion brief and chain definition
  3. 187SPIDER-SENSE — HITL Q&A go/no-go
  4. 187VERSION — Tag promoted skillchain
  5. 187PUBLISH — Announce promotion in release notes

Artifact

Promoted skillchain definition + HITL log

View example

Spider-Sense · Novel Skill

Sense → charter → HITL → scaffold → ship stub

6

Novel skill detection through charter, HITL, scaffold (repo/craft), docs, version, publish stub.

spider-sense/repo/craft/docs/version/publish

  1. 187SPIDER-SENSE — Detect novel skill/pattern
  2. 187SPIDER-SENSE — Conceive logic plan and skill charter
  3. 187SPIDER-SENSE — HITL Q&A before execute
  4. 187REPO — Scaffold skill folder and wiring
  5. 187CRAFT — UI/docs surfaces if public-facing
  6. 187DOCS — Write skill contract and triggers
  7. 187VERSION — Tag 0.1.0 stub
  8. 187PUBLISH — Sync showcase/docs stub

Artifact

New skill charter + scaffold + versioned stub

View example

Spider-Sense · Feature Build

Aura → plan → HITL → create/craft → gates → ship

7

Full path from aura signal to feature build with create/craft, access gates, and XAVIER ship.

spider-sense/create/craft/access-plus/include/test/version/publish

  1. 187SPIDER-SENSE — Aura signal and goal plan
  2. 187SPIDER-SENSE — HITL Q&A go
  3. 187CREATE — Growth surface if conversion-first
  4. 187CRAFT — Systemize UI and tokens
  5. access-plus — Disability access gate
  6. 187INCLUDE+ — Inclusion gate
  7. 187TEST — Validation matrix
  8. 187VERSION — Tag feature release
  9. 187PUBLISH — Final publish under XAVIER ship

Artifact

Shipped feature + HITL log + gates

View example

Agent Equity Rebalance

Audit → plan → assign → version

8

Keep peer agents equitable; XAVIER remains ~2× by design.

spider-sense/docs/version/write

  1. 187SPIDER-SENSE — Sense imbalance in kits or routing
  2. 187DOCS — Write equity report
  3. 187WRITE — Draft rebalance tasks per agent
  4. 187VERSION — Tag equity audit release

Artifact

Equity report + rebalance tasks

View example

Cross-Layer Rescue

Interject → reassign → timebox → log

9

Unstick workstreams with reassignment and audit logging.

docs/write/version/test

  1. 187DOCS — Capture stuck state and owners
  2. 187WRITE — Issue reassignment and timebox
  3. 187TEST — Define acceptance for unstick
  4. 187VERSION — Tag interjection event

Artifact

Interjection ledger entry

View example

Incident Council Ship

NATASHA → council → fix → ship

10

Security incident path through council and controlled ship.

natasha/test/docs/version/publish

  1. 187NATASHA — Contain and assess
  2. 187DOCS — Council packet and comms notes
  3. 187TEST — Validate mitigation
  4. 187VERSION — Tag incident release
  5. 187PUBLISH — Controlled ship after XAVIER decision

Artifact

Incident release + council decision

View example

Access+ · Include+ Authority Ship

Dual gate → residual risk → ship

11

XAVIER ship only after premier Access+/Include+ gates clear.

access-plus/include/test/version/publish

  1. access-plus — Confirm disability access gate
  2. 187INCLUDE+ — Confirm inclusion gate
  3. 187TEST — Confirm regression coverage
  4. 187VERSION — Tag access-ready candidate
  5. 187PUBLISH — Ship under XAVIER authority

Artifact

Access-ready release

View example

Xavier Council

Call a huddle

Convene any agents for a Q&A or decision log. This surface records the council request; execution is routed through the agentic runtime. Defaults: pick agents → topic → Call council.

Agents to convene
Urgency

Xavier Brood

Low-token brood monitors

XAVIER oversees all operations with small, single-purpose, low-bandwidth brood clones — one job each, not fat generalists. Cap: 5 active. Pause, resume, or kill-all as needed. Runtime dispatch wires later; this is the control-plane UI.

Active 0/5

No clones yet. Pick a parent agent, describe a scoped task, then Spawn.

Xavier Audit Ledger

Append-only session log

Councils, brood spawns, kills, and dispatches land here for this browser session. Export as Markdown when you need a ship record. Persistence across reloads is a Phase 4 follow-up.

Filter events

No events yet. Call a council or spawn a brood clone to seed the ledger.