foryourhealth111-pixel/Vibe-Skills
vibe
VerifiedVibe Code Orchestrator (VCO) is a governed runtime entry that freezes requirements, bounds execution, and enforces verification and phase cleanup.
scripts/router/modules/49-prompt-asset-boost.ps1
View on GitHub# Prompt Asset Boost Overlay (GPT‑5.2 assisted)
#
# Purpose:
# - When a prompt-template intent is detected, generate a prompts.chat search plan
# and a small set of "prompt overlay candidates" that the user can confirm
# and inject into subsequent tool prompts.
#
# Boundaries:
# - Advice-only (never changes selected pack/skill by default).
# - Explicit vibe only by default (requires `$vibe` or `/vibe` prefix).
# - Designed to cooperate with `prompt_overlay_advice` rather than replace it.
function Get-PromptAssetBoostPolicyDefaults {
return [pscustomobject]@{
enabled = $false
mode = "off" # off|shadow|soft|strict
activation = [pscustomobject]@{
explicit_vibe_only = $true
}
scope = [pscustomobject]@{
grade_allow = @("M", "L", "XL")
task_allow = @("planning", "coding", "review", "debug", "research")
}
trigger = [pscustomobject]@{
require_prompt_signal = $true
require_explicit_intent = $false
max_queries = 5
max_candidates = 3
}
provider = [pscustomobject]@{
type = "openai" # openai|mock
model = "your-model-id"
base_url = ""
mock_response_path = ""
timeout_ms = 9000
max_output_tokens = 1200
temperature = 0.2
top_p = 1.0
store = $false
}
output = [pscustomobject]@{
max_prompt_chars_per_candidate = 1200
}
safety = [pscustomobject]@{
fallback_on_error = $true
}
}
}
function Get-PromptAssetBoostPolicy {
param([object]$Policy)
$defaults = Get-PromptAssetBoostPolicyDefaults
if (-not $Policy) { return $defaults }
$enabled = if ($Policy.enabled -ne $null) { [bool]$Policy.enabled } else { [bool]$defaults.enabled }
$mode = if ($Policy.mode) { [string]$Policy.mode } else { [string]$defaults.mode }
$activation = if ($Policy.activation) { $Policy.activation } else { $defaults.activation }
$scope = if ($Policy.scope) { $Policy.scope } else { $defaults.scope }
$trigger = if ($Policy.trigger) { $Policy.trigger } else { $defaults.trigger }
$provider = if ($Policy.provider) { $Policy.provider } else { $defaults.provider }
$output = if ($Policy.output) { $Policy.output } else { $defaults.output }
$safety = if ($Policy.safety) { $Policy.safety } else { $defaults.safety }
return [pscustomobject]@{
enabled = $enabled
mode = $mode
activation = [pscustomobject]@{
explicit_vibe_only = if ($activation.explicit_vibe_only -ne $null) { [bool]$activation.explicit_vibe_only } else { [bool]$defaults.activation.explicit_vibe_only }
}
scope = [pscustomobject]@{
grade_allow = if ($scope.grade_allow) { @($scope.grade_allow) } else { @($defaults.scope.grade_allow) }
task_allow = if ($scope.task_allow) { @($scope.task_allow) } else { @($defaults.scope.task_allow) }
}
trigger = [pscustomobject]@{
require_prompt_signal = if ($trigger.require_prompt_signal -ne $null) { [bool]$trigger.require_prompt_signal } else { [bool]$defaults.trigger.require_prompt_signal }
require_explicit_intent = if ($trigger.require_explicit_intent -ne $null) { [bool]$trigger.require_explicit_intent } else { [bool]$defaults.trigger.require_explicit_intent }
max_queries = if ($trigger.max_queries -ne $null) { [int]$trigger.max_queries } else { [int]$defaults.trigger.max_queries }
max_candidates = if ($trigger.max_candidates -ne $null) { [int]$trigger.max_candidates } else { [int]$defaults.trigger.max_candidates }
}
provider = [pscustomobject]@{
type = if ($provider.type) { [string]$provider.type } else { [string]$defaults.provider.type }
model = if ($provider.model) { [string]$provider.model } else { [string]$defaults.provider.model }
base_url = if ($provider.base_url) { [string]$provider.base_url } else { [string]$defaults.provider.base_url }
mock_response_path = if ($provider.mock_response_path) { [string]$provider.mock_response_path } else { [string]$defaults.provider.mock_response_path }
timeout_ms = if ($provider.timeout_ms -ne $null) { [int]$provider.timeout_ms } else { [int]$defaults.provider.timeout_ms }
max_output_tokens = if ($provider.max_output_tokens -ne $null) { [int]$provider.max_output_tokens } else { [int]$defaults.provider.max_output_tokens }
temperature = if ($provider.temperature -ne $null) { [double]$provider.temperature } else { [double]$defaults.provider.temperature }
top_p = if ($provider.top_p -ne $null) { [double]$provider.top_p } else { [double]$defaults.provider.top_p }
store = if ($provider.store -ne $null) { [bool]$provider.store } else { [bool]$defaults.provider.store }
}
output = [pscustomobject]@{
max_prompt_chars_per_candidate = if ($output.max_prompt_chars_per_candidate -ne $null) { [int]$output.max_prompt_chars_per_candidate } else { [int]$defaults.output.max_prompt_chars_per_candidate }
}
safety = [pscustomobject]@{
fallback_on_error = if ($safety.fallback_on_error -ne $null) { [bool]$safety.fallback_on_error } else { [bool]$defaults.safety.fallback_on_error }
}
}
}
function Test-PromptAssetBoostScope {
param(
[object]$Policy,
[object]$PromptNormalization,
[string]$Grade,
[string]$TaskType
)
$resolved = Get-PromptAssetBoostPolicy -Policy $Policy
if (-not $resolved.enabled) {
return [pscustomobject]@{ enabled = $false; mode = "off"; scope_applicable = $false; reasons = @("policy_disabled") }
}
$mode = if ($resolved.mode) { [string]$resolved.mode } else { "off" }
if ($mode -eq "off") {
return [pscustomobject]@{ enabled = $false; mode = "off"; scope_applicable = $false; reasons = @("mode_off") }
}
$reasons = @()
if ($resolved.activation.explicit_vibe_only) {
$prefixDetected = [bool]($PromptNormalization -and $PromptNormalization.prefix_detected)
if (-not $prefixDetected) {
$reasons += "explicit_vibe_only"
}
}
if ($resolved.scope.grade_allow.Count -gt 0 -and -not ($resolved.scope.grade_allow -contains $Grade)) { $reasons += "grade_not_allowed" }
if ($resolved.scope.task_allow.Count -gt 0 -and -not ($resolved.scope.task_allow -contains $TaskType)) { $reasons += "task_not_allowed" }
$scopeApplicable = ($reasons.Count -eq 0)
return [pscustomobject]@{
enabled = $true
mode = $mode
scope_applicable = $scopeApplicable
reasons = @($reasons)
}
}
function Get-PromptAssetBoostJsonSchema {
param(
[int]$MaxQueries = 5,
[int]$MaxCandidates = 3,
[int]$MaxPromptCharsPerCandidate = 1200
)
$qMax = [Math]::Max(1, [Math]::Min(8, [int]$MaxQueries))
$cMax = [Math]::Max(1, [Math]::Min(3, [int]$MaxCandidates))
$pMax = [Math]::Max(200, [Math]::Min(2000, [int]$MaxPromptCharsPerCandidate))
return [ordered]@{
type = "object"
additionalProperties = $false
required = @("search_plan", "overlay_candidates")
properties = [ordered]@{
search_plan = [ordered]@{
type = "object"
additionalProperties = $false
required = @("queries")
properties = [ordered]@{
queries = [ordered]@{
type = "array"
maxItems = $qMax
items = [ordered]@{ type = "string"; maxLength = 120 }
}
category_hints = [ordered]@{
type = "array"
maxItems = 4
items = [ordered]@{ type = "string"; maxLength = 40 }
}
tag_hints = [ordered]@{
type = "array"
maxItems = 8
items = [ordered]@{ type = "string"; maxLength = 40 }
}
type_hints = [ordered]@{
type = "array"
maxItems = 4
items = [ordered]@{ type = "string"; enum = @("TEXT", "STRUCTURED", "IMAGE", "VIDEO", "AUDIO") }
}
limit = [ordered]@{
type = "integer"
minimum = 1
maximum = 50
}
}
}
overlay_candidates = [ordered]@{
type = "array"
maxItems = $cMax
items = [ordered]@{
type = "object"
additionalProperties = $false
required = @("id", "title", "role", "prompt", "when_to_use")
properties = [ordered]@{
id = [ordered]@{ type = "string"; maxLength = 24 }
title = [ordered]@{ type = "string"; maxLength = 80 }
role = [ordered]@{ type = "string"; enum = @("system", "user") }
prompt = [ordered]@{ type = "string"; maxLength = $pMax }
when_to_use = [ordered]@{ type = "string"; maxLength = 220 }
variables = [ordered]@{
type = "array"
maxItems = 12
items = [ordered]@{ type = "string"; maxLength = 40 }
}
confidence = [ordered]@{
type = "number"
minimum = 0
maximum = 1
}
}
}
}
confirm_questions = [ordered]@{
type = "array"
maxItems = 6
items = [ordered]@{ type = "string"; maxLength = 140 }
}
rationale = [ordered]@{
type = "string"
maxLength = 480
}
}
}
}
function New-PromptAssetBoostInputText {
param(
[string]$PromptText,
[object]$PromptNormalization,
[string]$Grade,
[string]$TaskType,
[string]$RouteMode,
[string]$SelectedPackId,
[string]$SelectedSkill,
[object]$PromptOverlayAdvice,
[int]$MaxQueries,
[int]$MaxCandidates,
[int]$MaxPromptCharsPerCandidate
)
$normalized = if ($PromptNormalization -and $PromptNormalization.normalized) { [string]$PromptNormalization.normalized } else { $PromptText }
$facetMatches = $null
$matchedFacets = @()
$promptSignalHit = $false
$docSurfaceHit = $false
try {
if ($PromptOverlayAdvice) {
$promptSignalHit = [bool]$PromptOverlayAdvice.prompt_signal_hit
$docSurfaceHit = [bool]$PromptOverlayAdvice.doc_surface_hit
$matchedFacets = @($PromptOverlayAdvice.matched_intent_facets)
$facetMatches = $PromptOverlayAdvice.facet_matches
}
} catch { }
$ctx = [ordered]@{
vco = [ordered]@{
grade = $Grade
task_type = $TaskType
route_mode = $RouteMode
selected_pack = if ($SelectedPackId) { $SelectedPackId } else { "none" }
selected_skill = if ($SelectedSkill) { $SelectedSkill } else { "none" }
}
prompt = [ordered]@{
original = $PromptText
normalized = $normalized
}
prompt_overlay_signals = [ordered]@{
prompt_signal_hit = [bool]$promptSignalHit
doc_surface_hit = [bool]$docSurfaceHit
matched_intent_facets = @($matchedFacets)
facet_matches = if ($facetMatches) { $facetMatches } else { [pscustomobject]@{} }
}
constraints = [ordered]@{
max_queries = [int]$MaxQueries
max_candidates = [int]$MaxCandidates
max_prompt_chars_per_candidate = [int]$MaxPromptCharsPerCandidate
}
}
$json = ($ctx | ConvertTo-Json -Depth 12 -Compress)
$text = @()
$text += "You generate prompts.chat search plans and small prompt overlay candidates."
$text += "Return ONLY JSON that matches the JSON Schema. No markdown. No extra keys."
$text += ""
$text += "Guidelines:"
$text += "- Queries should be short and directly searchable (keywords, not long sentences)."
$text += "- Candidates must be safe to inject as a system/user prompt (no tool calls)."
$text += "- Keep each candidate prompt concise; prefer variables like {project}, {goal}."
$text += ""
$text += "Context JSON:"
$text += $json
return ($text -join "`n")
}
function Invoke-PromptAssetBoostProvider {
param(
[object]$PolicyResolved,
[string]$RepoRoot,
[string]$InputText,
[object]$Schema
)
$providerType = if ($PolicyResolved -and $PolicyResolved.provider -and $PolicyResolved.provider.type) { [string]$PolicyResolved.provider.type } else { "openai" }
if ($providerType -eq "mock") {
$mockRel = if ($PolicyResolved.provider.mock_response_path) { [string]$PolicyResolved.provider.mock_response_path } else { "" }
if (-not $mockRel) {
return [pscustomobject]@{
ok = $false
abstained = $true
reason = "mock_missing_path"
api = "mock"
latency_ms = 0
output_text = $null
error = $null
}
}
$base = if ($RepoRoot) { [string]$RepoRoot } else { "" }
$path = if ([System.IO.Path]::IsPathRooted($mockRel)) { $mockRel } else { if ($base) { Join-Path $base $mockRel } else { $mockRel } }
if (-not (Test-Path -LiteralPath $path)) {
return [pscustomobject]@{
ok = $false
abstained = $true
reason = "mock_file_not_found"
api = "mock"
latency_ms = 0
output_text = $null
error = $path
}
}
$raw = Get-Content -LiteralPath $path -Raw -Encoding UTF8
return [pscustomobject]@{
ok = $true
abstained = $false
reason = "mock_ok"
api = "mock"
latency_ms = 0
output_text = $raw
error = $null
}
}
if ($providerType -ne "openai") {
return [pscustomobject]@{
ok = $false
abstained = $true
reason = "unsupported_provider"
api = "none"
latency_ms = 0
output_text = $null
error = $null
}
}
if (-not (Get-OpenAiApiKey)) {
return [pscustomobject]@{
ok = $false
abstained = $true
reason = "missing_openai_api_key"
api = "none"
latency_ms = 0
output_text = $null
error = $null
}
}
$model = if ($PolicyResolved.provider.model) { [string]$PolicyResolved.provider.model } else { "your-model-id" }
$baseUrl = if ($PolicyResolved.provider.base_url) { [string]$PolicyResolved.provider.base_url } else { "" }
$timeoutMsSafe = [Math]::Max(500, [int]$PolicyResolved.provider.timeout_ms)
$maxTokens = [int]$PolicyResolved.provider.max_output_tokens
$temperature = [double]$PolicyResolved.provider.temperature
$topP = [double]$PolicyResolved.provider.top_p
$store = [bool]$PolicyResolved.provider.store
$textFormat = [ordered]@{
type = "json_schema"
name = "vco_prompt_asset_boost"
strict = $true
schema = $Schema
}
$inputItems = @(
[ordered]@{
role = "user"
content = @(
[ordered]@{
type = "input_text"
text = $InputText
}
)
}
)
$instructions = "Return ONLY JSON that matches the JSON Schema. No markdown. No extra keys."
$chatResponseFormat = [ordered]@{
type = "json_schema"
json_schema = [ordered]@{
name = "vco_prompt_asset_boost"
strict = $true
schema = $Schema
}
}
$chatMessages = @(
[ordered]@{ role = "system"; content = $instructions },
[ordered]@{ role = "user"; content = $InputText }
)
$preferChat = $false
if ($baseUrl -and ($baseUrl -notmatch "openai\\.com")) { $preferChat = $true }
$invokeResponses = {
$r = Invoke-OpenAiResponsesCreate `
-Model $model `
-BaseUrl $baseUrl `
-InputItems $inputItems `
-TextFormat $textFormat `
-Instructions $instructions `
-MaxOutputTokens $maxTokens `
-Temperature $temperature `
-TopP $topP `
-TimeoutMs $timeoutMsSafe `
-Store:([bool]$store)
$r | Add-Member -NotePropertyName api -NotePropertyValue "responses" -Force
return $r
}
$invokeChat = {
$r = Invoke-OpenAiChatCompletionsCreate `
-Model $model `
-BaseUrl $baseUrl `
-Messages $chatMessages `
-ResponseFormat $chatResponseFormat `
-MaxTokens $maxTokens `
-Temperature $temperature `
-TopP $topP `
-TimeoutMs $timeoutMsSafe
$r | Add-Member -NotePropertyName api -NotePropertyValue "chat_completions" -Force
return $r
}
$providerResult = $null
if ($preferChat) {
$providerResult = & $invokeChat
if (-not ([bool]$providerResult.ok -and (-not [bool]$providerResult.abstained) -and $providerResult.output_text)) {
$providerResult = & $invokeResponses
}
} else {
$providerResult = & $invokeResponses
if (-not ([bool]$providerResult.ok -and (-not [bool]$providerResult.abstained) -and $providerResult.output_text)) {
if ([string]$providerResult.reason -ne "missing_openai_api_key") {
$providerResult = & $invokeChat
}
}
}
if (-not $providerResult -or -not [bool]$providerResult.ok -or [bool]$providerResult.abstained -or (-not $providerResult.output_text)) {
$reason = if ($providerResult -and $providerResult.reason) { [string]$providerResult.reason } else { "provider_abstained" }
return [pscustomobject]@{
ok = $false
abstained = $true
reason = $reason
api = if ($providerResult -and $providerResult.api) { [string]$providerResult.api } else { "unknown" }
latency_ms = if ($providerResult -and $providerResult.latency_ms -ne $null) { [int]$providerResult.latency_ms } else { 0 }
output_text = $null
error = if ($providerResult -and $providerResult.error) { [string]$providerResult.error } else { $null }
}
}
return [pscustomobject]@{
ok = $true
abstained = $false
reason = "ok"
api = if ($providerResult.api) { [string]$providerResult.api } else { "unknown" }
latency_ms = if ($providerResult.latency_ms -ne $null) { [int]$providerResult.latency_ms } else { 0 }
output_text = [string]$providerResult.output_text
error = $null
}
}
function Get-PromptAssetBoostAdvice {
param(
[string]$PromptText,
[object]$PromptNormalization,
[string]$PromptLower,
[string]$Grade,
[string]$TaskType,
[string]$RouteMode,
[string]$SelectedPackId,
[string]$SelectedSkill,
[object]$PromptOverlayAdvice,
[object]$PromptAssetBoostPolicy,
[string]$RepoRoot
)
if (-not $PromptAssetBoostPolicy) {
return [pscustomobject]@{
enabled = $false
mode = "off"
scope_applicable = $false
enforcement = "none"
reason = "policy_missing"
preserve_routing_assignment = $true
should_apply_hook = $false
prompt_signal_hit = [bool]($PromptOverlayAdvice -and $PromptOverlayAdvice.prompt_signal_hit)
explicit_intent = [bool]($PromptOverlayAdvice -and @($PromptOverlayAdvice.matched_intent_facets).Count -gt 0)
recommended_skill = $null
confirm_required = $false
search_plan = $null
overlay_candidates = @()
provider = [pscustomobject]@{
ok = $false
abstained = $true
reason = "policy_missing"
api = "none"
latency_ms = 0
error = $null
}
}
}
$resolved = Get-PromptAssetBoostPolicy -Policy $PromptAssetBoostPolicy
$scope = Test-PromptAssetBoostScope -Policy $resolved -PromptNormalization $PromptNormalization -Grade $Grade -TaskType $TaskType
if (-not [bool]$scope.enabled -or -not [bool]$scope.scope_applicable) {
return [pscustomobject]@{
enabled = $false
mode = [string]$scope.mode
scope_applicable = $false
enforcement = "none"
reason = if ($scope.reasons -and $scope.reasons.Count -gt 0) { ($scope.reasons -join ",") } else { "outside_scope" }
preserve_routing_assignment = $true
should_apply_hook = $false
prompt_signal_hit = [bool]($PromptOverlayAdvice -and $PromptOverlayAdvice.prompt_signal_hit)
explicit_intent = [bool]($PromptOverlayAdvice -and @($PromptOverlayAdvice.matched_intent_facets).Count -gt 0)
recommended_skill = $null
confirm_required = $false
search_plan = $null
overlay_candidates = @()
provider = [pscustomobject]@{
ok = $false
abstained = $true
reason = "outside_scope"
api = "none"
latency_ms = 0
error = $null
}
}
}
$mode = [string]$scope.mode
$promptSignalHit = [bool]($PromptOverlayAdvice -and $PromptOverlayAdvice.prompt_signal_hit)
$explicitIntent = [bool]($PromptOverlayAdvice -and @($PromptOverlayAdvice.matched_intent_facets).Count -gt 0)
$triggerReasons = @()
if ([bool]$resolved.trigger.require_prompt_signal -and (-not ($promptSignalHit -or $explicitIntent))) {
$triggerReasons += "missing_prompt_signal"
}
if ([bool]$resolved.trigger.require_explicit_intent -and (-not $explicitIntent)) {
$triggerReasons += "missing_explicit_intent"
}
if ($triggerReasons.Count -gt 0) {
return [pscustomobject]@{
enabled = $true
mode = $mode
scope_applicable = $true
enforcement = "advisory"
reason = ($triggerReasons -join ",")
preserve_routing_assignment = $true
should_apply_hook = $false
prompt_signal_hit = $promptSignalHit
explicit_intent = $explicitIntent
recommended_skill = $null
confirm_required = $false
search_plan = $null
overlay_candidates = @()
provider = [pscustomobject]@{
ok = $false
abstained = $true
reason = ($triggerReasons -join ",")
api = "none"
latency_ms = 0
error = $null
}
}
}
$enforcement = "advisory"
$reason = "soft_advisory"
switch ($mode) {
"shadow" {
$enforcement = "advisory"
$reason = "shadow_advisory"
}
"soft" {
$enforcement = "advisory"
$reason = "soft_advisory"
}
"strict" {
$enforcement = "required"
$reason = "strict_required"
}
default {
$enforcement = "advisory"
$reason = "unknown_mode_advisory"
}
}
$recommendedSkill = "prompt-lookup"
$shouldApplyHook = $true
if ($mode -eq "shadow") {
return [pscustomobject]@{
enabled = $true
mode = $mode
scope_applicable = $true
enforcement = $enforcement
reason = $reason
preserve_routing_assignment = $true
should_apply_hook = $shouldApplyHook
prompt_signal_hit = $promptSignalHit
explicit_intent = $explicitIntent
recommended_skill = $recommendedSkill
confirm_required = $false
search_plan = $null
overlay_candidates = @()
provider = [pscustomobject]@{
ok = $false
abstained = $true
reason = "shadow_mode_no_provider"
api = "none"
latency_ms = 0
error = $null
}
}
}
$maxQueries = [Math]::Max(1, [Math]::Min(8, [int]$resolved.trigger.max_queries))
$maxCandidates = [Math]::Max(1, [Math]::Min(3, [int]$resolved.trigger.max_candidates))
$maxPromptChars = [Math]::Max(200, [Math]::Min(2000, [int]$resolved.output.max_prompt_chars_per_candidate))
$schema = Get-PromptAssetBoostJsonSchema -MaxQueries $maxQueries -MaxCandidates $maxCandidates -MaxPromptCharsPerCandidate $maxPromptChars
$inputText = New-PromptAssetBoostInputText `
-PromptText $PromptText `
-PromptNormalization $PromptNormalization `
-Grade $Grade `
-TaskType $TaskType `
-RouteMode $RouteMode `
-SelectedPackId $SelectedPackId `
-SelectedSkill $SelectedSkill `
-PromptOverlayAdvice $PromptOverlayAdvice `
-MaxQueries $maxQueries `
-MaxCandidates $maxCandidates `
-MaxPromptCharsPerCandidate $maxPromptChars
$provider = Invoke-PromptAssetBoostProvider -PolicyResolved $resolved -RepoRoot $RepoRoot -InputText $inputText -Schema $schema
$searchPlan = $null
$candidates = @()
$confirmQuestions = @()
$rationale = $null
if ([bool]$provider.ok -and $provider.output_text) {
try {
$obj = ($provider.output_text.Trim() | ConvertFrom-Json)
if ($obj -and $obj.search_plan) { $searchPlan = $obj.search_plan }
if ($obj -and $obj.overlay_candidates) { $candidates = @($obj.overlay_candidates) }
if ($obj -and $obj.confirm_questions) { $confirmQuestions = @($obj.confirm_questions) }
if ($obj -and $obj.rationale) { $rationale = [string]$obj.rationale }
} catch {
}
}
$cleanCandidates = @()
foreach ($c in @($candidates | Select-Object -First $maxCandidates)) {
if (-not $c) { continue }
$prompt = if ($c.prompt) { [string]$c.prompt } else { "" }
if ($prompt.Length -gt $maxPromptChars) { $prompt = $prompt.Substring(0, $maxPromptChars) }
if (-not $prompt.Trim()) { continue }
$cleanCandidates += [pscustomobject]@{
id = if ($c.id) { [string]$c.id } else { "candidate_{0}" -f ($cleanCandidates.Count + 1) }
title = if ($c.title) { [string]$c.title } else { "Prompt Candidate {0}" -f ($cleanCandidates.Count + 1) }
role = if ($c.role) { [string]$c.role } else { "system" }
prompt = $prompt.Trim()
when_to_use = if ($c.when_to_use) { [string]$c.when_to_use } else { "" }
variables = if ($c.variables) { Get-ArraySafe -Value $c.variables } else { Get-ArraySafe -Value $null }
confidence = if ($c.confidence -ne $null) { [double]$c.confidence } else { 0.55 }
}
}
$confirmRequired = ($cleanCandidates.Count -gt 0)
return [pscustomobject]@{
enabled = $true
mode = $mode
scope_applicable = $true
enforcement = $enforcement
reason = if ([bool]$provider.ok) { $reason } else { "provider_abstained" }
preserve_routing_assignment = $true
should_apply_hook = $shouldApplyHook
prompt_signal_hit = $promptSignalHit
explicit_intent = $explicitIntent
recommended_skill = $recommendedSkill
confirm_required = $confirmRequired
search_plan = $searchPlan
overlay_candidates = @($cleanCandidates)
confirm_questions = @($confirmQuestions | Where-Object { $_ } | Select-Object -First 6)
rationale = if ($rationale) { $rationale.Trim() } else { $null }
provider = [pscustomobject]@{
ok = [bool]$provider.ok
abstained = [bool]$provider.abstained
reason = [string]$provider.reason
api = [string]$provider.api
latency_ms = [int]$provider.latency_ms
error = if ($provider.error) { [string]$provider.error } else { $null }
}
}
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |