@pennyfarthing/core 9.2.0 → 9.3.0

package/pennyfarthing-dist/skills/yq/SKILL.md

@@ -1,272 +0,0 @@
----
-name: yq
-description: yq is a YAML processor. Use this skill when reading, modifying, or querying YAML files from the command line.
----
-
-# yq - YAML Processor Skill
-
-<run>
-yq [expression] [file]
-</run>
-
-<output>
-YAML/JSON output depending on specified format
-</output>
-
-## Overview
-
-`yq` is a lightweight and portable command-line YAML processor. It's like `jq` but for YAML. This skill covers the mikefarah/yq version (v4+).
-
-## Installation
-
-```bash
-# macOS
-brew install yq
-
-# Check version (should be v4+)
-yq --version
-```
-
-## Basic Operations
-
-### Read a Value
-
-```bash
-# Read a top-level key
-yq '.theme' config.yaml
-# Output: discworld
-
-# Read nested value
-yq '.agents.sm.character' personas.yaml
-# Output: Captain Carrot Ironfoundersson
-
-# Read with default if missing
-yq '.missing // "default"' config.yaml
-# Output: default
-```
-
-### Read Entire Section
-
-```bash
-# Get all properties of a key
-yq '.agents.sm' personas.yaml
-# Output: the entire sm object as YAML
-
-# Output as JSON
-yq -o json '.agents.sm' personas.yaml
-```
-
-### Check if Key Exists
-
-```bash
-# Returns "true" or "false"
-yq '.agents.sm | has("helper")' personas.yaml
-
-# Returns null if doesn't exist
-yq '.agents.missing' personas.yaml
-# Output: null
-```
-
-## Modifying YAML
-
-### In-Place Edit
-
-```bash
-# Edit file in place
-yq -i '.theme = "startrek"' config.yaml
-
-# Add a new key
-yq -i '.newkey = "value"' config.yaml
-
-# Delete a key
-yq -i 'del(.unwanted)' config.yaml
-```
-
-### Output to Stdout (Non-Destructive)
-
-```bash
-# Modify and output to stdout (file unchanged)
-yq '.count = 42' config.yaml
-
-# Pipe to new file
-yq '.theme = "newtheme"' config.yaml > new-config.yaml
-```
-
-## Common Patterns
-
-### Iterate Over Arrays
-
-```bash
-# List all items
-yq '.stories[]' sprint.yaml
-
-# Get specific field from each item
-yq '.stories[].title' sprint.yaml
-
-# Filter items
-yq '.stories[] | select(.status == "in-progress")' sprint.yaml
-```
-
-### Iterate Over Object Keys
-
-```bash
-# Get all agent names
-yq '.agents | keys' personas.yaml
-# Output:
-# - sm
-# - tea
-# - dev
-# - reviewer
-
-# Iterate with key-value
-yq '.agents | to_entries | .[] | .key + ": " + .value.character' personas.yaml
-```
-
-### Merge Files
-
-```bash
-# Merge two YAML files (second overwrites first)
-yq '. *= load("overrides.yaml")' base.yaml
-
-# Merge specific key
-yq '.agents.sm *= load("sm-overrides.yaml")' personas.yaml
-```
-
-### Extract to Environment Variables
-
-```bash
-# Export as shell variables
-eval $(yq -o shell '.database' config.yaml)
-# Creates: $host, $port, $name
-
-# Specific format
-yq '.theme' config.yaml | xargs -I{} echo "THEME={}"
-```
-
-## Working with Multi-Document YAML
-
-```bash
-# Select specific document (0-indexed)
-yq 'select(document_index == 0)' multi.yaml
-
-# Process all documents
-yq '... | select(has("kind"))' k8s-manifests.yaml
-```
-
-## String Operations
-
-```bash
-# Uppercase
-yq '.name | upcase' config.yaml
-
-# String contains
-yq '.agents | to_entries | .[] | select(.value.style | contains("honest"))' personas.yaml
-
-# String interpolation
-yq '"Theme is: " + .theme' config.yaml
-```
-
-## Conditional Logic
-
-```bash
-# If-then-else
-yq 'if .count > 10 then "many" else "few" end' config.yaml
-
-# Select with condition
-yq '.stories[] | select(.points >= 5)' sprint.yaml
-
-# Null coalescing
-yq '.optional // "default"' config.yaml
-```
-
-## Output Formats
-
-```bash
-# YAML (default)
-yq '.data' file.yaml
-
-# JSON
-yq -o json '.data' file.yaml
-
-# Properties format
-yq -o props '.data' file.yaml
-
-# CSV (for arrays)
-yq -o csv '.items' file.yaml
-```
-
-## Common Use Cases in Pennyfarthing
-
-### Read Persona Config
-
-```bash
-# Get current theme
-yq '.theme' .claude/persona-config.yaml
-
-# Get agent persona
-THEME=$(yq '.theme' .claude/persona-config.yaml)
-yq ".agents.sm" personas/themes/${THEME}.yaml
-```
-
-### Query Sprint Data
-
-```bash
-# Get current sprint stories
-yq '.stories[] | select(.status == "todo")' sprint/current-sprint.yaml
-
-# Count stories by status
-yq '[.stories[].status] | group_by(.) | map({(.[0]): length}) | add' sprint/current-sprint.yaml
-```
-
-### Update Session Files
-
-```bash
-# Update status in YAML frontmatter (if using YAML)
-yq -i '.status = "in-progress"' .session/current_work.yaml
-```
-
-## Quick Reference
-
-| Action | Command |
-|--------|---------|
-| Read value | `yq '.key' file.yaml` |
-| Read nested | `yq '.parent.child' file.yaml` |
-| Set value | `yq -i '.key = "value"' file.yaml` |
-| Delete key | `yq -i 'del(.key)' file.yaml` |
-| Array length | `yq '.array | length' file.yaml` |
-| Filter array | `yq '.array[] \| select(.field == "value")' file.yaml` |
-| Output JSON | `yq -o json '.' file.yaml` |
-| Merge files | `yq '. *= load("other.yaml")' base.yaml` |
-| Default value | `yq '.key // "default"' file.yaml` |
-
-## Troubleshooting
-
-### "null" Output
-
-The key doesn't exist. Use `// "default"` for fallback:
-```bash
-yq '.missing // "not found"' file.yaml
-```
-
-### Quotes in Output
-
-yq preserves YAML formatting. Use `-r` (raw) for unquoted strings:
-```bash
-yq -r '.theme' config.yaml
-```
-
-### Boolean as String
-
-YAML booleans (`true`/`false`) are parsed as booleans. Compare accordingly:
-```bash
-# CORRECT
-yq '.enabled == true' config.yaml
-
-# WRONG (comparing to string)
-yq '.enabled == "true"' config.yaml
-```
-
-## Reference
-
-- **Official Docs:** https://mikefarah.gitbook.io/yq/
-- **GitHub:** https://github.com/mikefarah/yq

package/pennyfarthing-dist/templates/LEADERBOARD.schema.yaml

@@ -1,187 +0,0 @@
-# Leaderboard Schema Specification
-# Version: 1.0.0
-# Generated leaderboards must conform to this structure
-
-schema_version: 1
-
-# Header section - metadata about the benchmark
-header:
-  scenario_name:
-    type: string
-    example: "race-condition-cache"
-    description: "Scenario identifier from scenarios/ directory"
-
-  scenario_title:
-    type: string
-    example: "Race Condition Cache"
-    description: "Human-readable scenario name (title case)"
-
-  difficulty:
-    type: enum
-    values: [easy, medium, hard, extreme]
-    description: "Scenario difficulty level"
-
-  role:
-    type: string
-    example: "dev"
-    description: "Agent role (dev, reviewer, tea, sm, architect)"
-
-  role_title:
-    type: string
-    example: "Developer"
-    description: "Human-readable role name"
-
-  baseline:
-    mean:
-      type: number
-      precision: 2
-      description: "Control agent mean score"
-    std_dev:
-      type: number
-      precision: 2
-      description: "Control agent standard deviation"
-    n:
-      type: integer
-      description: "Number of control runs"
-
-  version:
-    type: string
-    example: "5.0.1"
-    description: "Pennyfarthing version used"
-
-  generated_timestamp:
-    type: string
-    format: "ISO8601"
-    example: "2026-01-03T03:45:00Z"
-    description: "When leaderboard was generated"
-
-# Rankings table columns
-rankings:
-  columns:
-    - name: Rank
-      type: integer
-      description: "Position by mean score (ties share rank)"
-
-    - name: Theme
-      type: string
-      description: "Theme identifier"
-
-    - name: Character
-      type: string
-      description: "Persona character name for this role. Cross-role entries show '(as {role})' suffix."
-
-    - name: Mean
-      type: number
-      precision: 2
-      description: "Mean score across all runs"
-
-    - name: Std Dev
-      type: number
-      precision: 2
-      description: "Standard deviation of scores"
-
-    - name: n
-      type: integer
-      description: "Number of runs"
-
-    - name: Delta
-      type: number
-      precision: 2
-      signed: true
-      description: "Difference from baseline mean (+ or -)"
-
-    - name: "Cohen's d"
-      type: number
-      precision: 2
-      signed: true
-      description: "Effect size using pooled standard deviation"
-
-  # Cross-role metadata (optional, present when cross_role=true in summary.yaml)
-  cross_role_fields:
-    - name: Cross-Role
-      type: boolean
-      optional: true
-      description: "True if character is performing outside their native role"
-
-    - name: Source Role
-      type: string
-      optional: true
-      description: "Character's native role (e.g., 'sm' for Prospero)"
-
-    - name: Effective Role
-      type: string
-      optional: true
-      description: "Role being performed (matches the leaderboard's role)"
-
-  special_rows:
-    control_row:
-      position: "After last positive delta, before first negative"
-      format: "| --- | **CONTROL** | **Baseline** | **{mean}** | **{std}** | **{n}** | **0.00** | **0.00** |"
-      description: "Bold separator row showing baseline"
-
-  sorting:
-    primary: mean
-    order: descending
-    tie_handling: "Same rank, listed alphabetically by theme"
-
-# Cohen's d calculation
-cohens_d:
-  formula: "d = (M1 - M2) / sqrt(((n1-1)*s1² + (n2-1)*s2²) / (n1+n2-2))"
-  interpretation:
-    negligible: "< 0.20"
-    small: "0.20"
-    medium: "0.50"
-    large: "0.80"
-    very_large: "> 1.20"
-  notes:
-    - "Positive d = outperforms baseline"
-    - "Negative d = underperforms baseline"
-    - "Uses pooled standard deviation for unequal sample sizes"
-
-# Summary statistics section
-summary:
-  total_themes:
-    type: integer
-    description: "Count of all themed agents benchmarked"
-
-  themes_above:
-    type: integer
-    description: "Count with mean > baseline mean"
-
-  themes_near:
-    type: integer
-    description: "Count within 1 point of baseline"
-
-  total_runs:
-    type: integer
-    description: "Sum of all n values"
-
-# Top/Bottom performer tables
-highlights:
-  top_performers:
-    count: 5
-    criteria: "Highest Cohen's d (most positive)"
-    columns: [Theme, Character, Mean, Delta, "Cohen's d"]
-
-  bottom_performers:
-    count: 5
-    criteria: "Lowest Cohen's d (most negative)"
-    columns: [Theme, Character, Mean, Delta, "Cohen's d"]
-
-# Data sources
-data_sources:
-  baseline: "internal/results/baselines/{scenario}/{role}/summary.yaml"
-  themes: "internal/results/benchmarks/{scenario}/{theme}-{role}/summary.yaml"
-  cross_role: "internal/results/benchmarks/{scenario}/{theme}-{character}-as-{role}/summary.yaml"
-  ocean_analysis: "./OCEAN-ANALYSIS.md"
-
-# Generation requirements
-requirements:
-  minimum_n: 4
-  recommended_n: 5
-  baseline_n: 10
-  character_required: true
-  notes:
-    - "All themes should have consistent n for valid comparison"
-    - "Character names must be populated from theme YAML"
-    - "Model and token data tracked per-run, not in leaderboard header"

package/pennyfarthing-dist/templates/LEADERBOARD.template.md

@@ -1,59 +0,0 @@
-# {SCENARIO_TITLE} - {ROLE_TITLE} Leaderboard
-
-**Scenario:** `{scenario_name}` ({difficulty} difficulty)
-**Role:** {role_title} (`{role}`)
-**Baseline:** Control agent, mean {baseline_mean}, std_dev {baseline_std}, n={baseline_n}
-**Pennyfarthing Version:** {version}
-**Generated:** {generated_timestamp}
-
-**Analysis:** [OCEAN Personality Correlation Analysis](./OCEAN-ANALYSIS.md)
-
----
-
-## Cohen's d Interpretation
-
-| Effect Size | Cohen's d |
-|-------------|-----------|
-| Negligible | |d| < 0.20 |
-| Small | |d| = 0.20 |
-| Medium | |d| = 0.50 |
-| Large | |d| = 0.80 |
-| Very Large | |d| > 1.20 |
-
----
-
-## Rankings
-
-| Rank | Theme | Character | Mean | Std Dev | n | Delta | Cohen's d |
-|------|-------|-----------|------|---------|---|-------|-----------|
-{RANKINGS_ROWS}
-
----
-
-## Summary Statistics
-
-- **Total Themes:** {total_themes}
-- **Themes Beating Baseline:** {themes_above}
-- **Themes At/Near Baseline (within 1 point):** {themes_near}
-- **Total Benchmark Runs:** {total_runs}
-
-## Top Performers (Above Baseline)
-
-| Theme | Character | Mean | Delta | Cohen's d |
-|-------|-----------|------|-------|-----------|
-{TOP_PERFORMERS_ROWS}
-
-## Bottom Performers
-
-| Theme | Character | Mean | Delta | Cohen's d |
-|-------|-----------|------|-------|-----------|
-{BOTTOM_PERFORMERS_ROWS}
-
----
-
-## Notes
-
-- Cohen's d uses pooled standard deviation: `d = (M1 - M2) / sqrt(((n1-1)*s1² + (n2-1)*s2²) / (n1+n2-2))`
-- Positive d indicates theme outperforms baseline; negative indicates underperformance
-- Model and token usage tracked per-run in individual result files
-- {scenario_description}