agentsclimarketplace

Escalation routing

Skill fabioc-aloha/Alex_Skill_Mall/plugins/supervisor-fleet/escalation-routing

284 curated plugins for AI assistants across 16 categories: security, Azure, documentation, code quality, cloud infrastructure, and more. Works with GitHub Copilot. Drop into .github/skills/local/ and go.

Install
npx -y skills add fabioc-aloha/Alex_Skill_Mall --skill escalation-routing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Decide whether a feedback item is in-scope for the Supervisor or must be escalated to AlexMaster (framework-level)

SKILL.md

6.3 KB, as published. Nobody here has run it

Escalation Routing

The Supervisor owns ACT-fleet curation. AlexMaster owns the framework. This skill draws the line.

When to Use

  • Triaging an inbox item via feedback-triage and the route is unclear
  • A pattern recurs across heirs and feels like it belongs in the framework, not the curation
  • A heir requests something the Supervisor cannot grant alone

The Cardinal Test

Before any change, ask: Does AlexMaster do this?

If AlexMaster…Then the change is…
Owns the artifact type (instruction, skill, prompt schema, ACT tenet)Framework — escalate
Owns the policy (how releases work, what counts as a skill, the ACT pass)Framework — escalate
Owns the brain spec (copilot-instructions.md, three-tier inheritance)Framework — escalate
Just has a sibling artifact (a different skill, a different store)Curation — handle in-house
Is downstream consumer (would adopt the change later)Curation — handle in-house

In-Scope vs Escalation Matrix

SignalRoute
Proposal for a new ACT tenetEscalate
Proposal for a new artifact type (e.g., "we need a pattern.md alongside skills")Escalate
Proposal that changes the brain spec (frontmatter fields, inheritance tiers)Escalate
Proposal that changes how the ACT pass worksEscalate
Cross-fleet pattern observed in 3+ unrelated heirsEscalate (it's framework signal)
Domain skill (a new specialized capability)In-house
New mall storeIn-house (store-evaluation)
Edition skill bug fixIn-house (feedback-triage → Edition PR)
Mall entry refreshIn-house (mall-curation)
Coherence violationIn-house (coherence-audit)
Heir request for a curated pattern bundleIn-house
Disagreement about whether to handle in-houseEscalate (when in doubt, escalate — the cost of asking is low; the cost of overreach is high)

Escalation Procedure

  1. Confirm the item meets at least one Escalate signal — and document which one
  2. Run a full ACT pass (high stakes — this is a framework-level claim about scope)
  3. Write the escalation as a feedback file at AI-Memory/feedback/<YYYY-MM-DD>-<short-slug>.md in the AlexMaster workspace (per pii-memory-filter: structured schema only, no domain data)
  4. Required fields:
    • From: Alex_ACT_Supervisor
    • Date: YYYY-MM-DD
    • Category: tenet-proposal / spec-change / artifact-type / cross-fleet-pattern / scope-question
    • Severity: low / medium / high / critical
    • Signal count: how many independent observations support this (the multi-observation rule)
    • Description: structured, abstract — what / why / proposed action
  5. Record the escalation in decisions/curation-log.md with route AlexMaster and M/S = S
  6. Wait for AlexMaster to act — do not modify Edition or Mall in anticipation of the escalation outcome. Pre-emptive changes are scope creep.

Hard Rules

  1. When in doubt, escalate — overreach is more expensive than asking
  2. Never modify framework artifacts in Edition or Mall — they are inherited from AlexMaster; modifying them creates fleet drift
  3. One escalation per concern — bundling unrelated items defeats the audit trail
  4. Strip project context per cross-project-isolation before writing to AlexMaster's AI-Memory/feedback/
  5. No silent escalations — every escalation is logged in curation-log.md

What "In-House" Means

If the item is in-house, it routes to one of the existing skills:

Item typeIn-house skill
Skill submission or fixskill-review
Generic feedback or bugfeedback-triage
New mall storestore-evaluation
Mall prune/refreshstaleness-discipline or mall-curation
Coherence violationcoherence-audit
Release coordinationrelease-ritual

Anti-Patterns

Anti-patternCorrection
Modifying inherited artifacts "just this once"Modifications must come through AlexMaster sync, never in heirs
Escalating every trivial questionUse the matrix; trivial items waste AlexMaster's bandwidth
Pre-emptive change while waiting for escalation outcomeWait for the answer; the change you preempt may not be what AlexMaster decides
Bundling unrelated escalationsOne concern, one escalation file
Forgetting to log the escalationCuration log is the audit trail; missing entries break the 60-day re-evaluation

Output Template (escalation file)

# Escalation — <short title>

**From**: Alex_ACT_Supervisor
**Date**: YYYY-MM-DD
**Category**: tenet-proposal | spec-change | artifact-type | cross-fleet-pattern | scope-question
**Severity**: low | medium | high | critical
**Signal count**: N independent observations

## What

<2-3 sentence abstract>

## Why this is framework-level (not curation)

<cite which Escalate signal fired>

## Evidence

- <observation 1, abstracted per cross-project-isolation>
- <observation 2>
- ...

## Proposed Action (for AlexMaster)

<2-3 sentence proposal>

## Impact if Not Adopted

<what continues to drift, what cost accrues>

## ACT Pass Trail

<markers>

Related

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.