agentleFS
mikeparcewskimikeparcewski/wicked-garden/skills/persona/SKILL.md

wicked-garden-persona

Skill9 starsChanged 7 months agoMITView on GitHub

On-demand persona invocation system for applying named perspectives to any task. Sub-actions: `as <name> <task>` (invoke a persona), `list [--role R]` (discover personas), `define <name> --focus "..." [--save]` (create or update a custom persona). Use when: invoking a named persona ("as platform, review this auth flow"), applying a specific perspective or role lens to a task, listing available personas, defining a house persona that encodes a failure-mode defense, or reviewing work through a specific role's perspective.

  • Installs packages

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 3 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If none of these is available, or Python 3 is missing, skip the script-backed step, say so, and follow the manual alternative where one is …

    line 26
  2. Very specific:

    If the first token is none of these, treat the args as as <args> when they start with a known persona name followed by a task.

    line 40
  3. Distinctive:

    methodology exemplars: keep rich profiles (they show the GOOD pattern).

    line 69
  4. Very specific:

    Create your own with the define action — the mechanism that lets an enterprise inject ITS house personas.

    line 107
  5. Very specific:

    Use --save to promote to plugin-level cache for cross-project reuse:

    line 117
  6. Very specific:

    If specialist.json is unavailable, three built-in personas are always available:

    line 151
  7. Distinctive:

    When names collide across sources, higher priority wins:

    line 158

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Persona System

On-demand persona invocation for applying named perspectives to any task. Personas are behavioral modifiers — they change the lens through which a task is executed, not the tools available.

Runtime

Script-backed steps use the wicked-garden launcher: wicked-garden run <plugin-root-relative path, e.g. scripts/…> [args]. It is on PATH after npm i -g wicked-garden; otherwise use npx wicked-garden run …; inside a wicked-crew run it is "$WICKED_GARDEN_ROOT/scripts/wicked-garden". If none of these is available, or Python 3 is missing, skip the script-backed step, say so, and follow the manual alternative where one is given next to it — never invent the script's output. Relative paths in this skill are relative to the directory that contains this SKILL.md.

Sub-Action Routing

Route on the first token of the args, then read the matching ref for the full flow:

Sub-actionArgsRefPurpose
as<persona-name> <task description>refs/as.mdInvoke a persona for any task — registry lookup, profile assembly, then embody the persona inline (the former persona-agent worker lives in this ref)
list[--role <role>]refs/list.mdList all available personas, tiered Methodology vs Generic
define<name> --focus "..." [--traits "..."] [--constraints "FAILURE MODE — ..."] [--not-focus "..."] [--role <role>] [--save]refs/define.mdCreate or update a custom persona — the enterprise house-persona injection mechanism; --save promotes to the plugin cache

If the first token is none of these, treat the args as as <args> when they start with a known persona name followed by a task; otherwise show the routing table above and STOP.

Quick Start

/wicked-garden-persona as engineering "review this auth flow"
/wicked-garden-persona list
/wicked-garden-persona define pragmatic-tech-lead --focus "delivery over perfection"

How It Works

  1. Registry merges built-in specialists + custom personas + plugin cache
  2. The as action looks up a persona, assembles its profile and embodies it inline (refs/as.md carries the behavioral guidelines that were the persona-agent worker — retired in B9)
  3. The task runs under the persona's behavioral profile
  4. Behavior emerges from personality, constraints, memories, and preferences

Built-in Personas

Names come from .claude-plugin/specialist.json; their rich profiles live in scripts/persona/registry.py::_BUILTIN_RICH. The built-ins are illustrative exemplars, not the product. A blinded, independently-graded lift eval (tests/persona/EVAL_RESULTS.md, run 2026-06-12) found the built-in personas produce lift=0 against a strong base model — it already flags the targeted failure modes unprompted. So the curated surface was reduced: only the three methodology exemplars keep rich profiles (they show the GOOD pattern); the generic role personas were demoted to thin role records. For real leverage, define your OWN house persona (see "Custom Personas") — that is the actual product.

Methodology personas (carry a failure-mode defense — the GOOD pattern)

Their registry records encode FAILURE MODE — … constraints + a not_focus scope guard the base model does not reliably self-apply. Kept as illustrative exemplars of what a worth-keeping persona looks like.

NameRoleDefends against
platformdevsecopsSilent secret exposure, irreversible deploy, privilege creep, unbounded blast radius
qequality-engineeringSelf-graded "done", happy-path-only tests, untested recovery, coverage theater
agenticagentic-architectureRunaway loops, ungated irreversible actions, over-broad tool access, non-idempotent retries

(skeptic, a fallback persona, is also methodology — it forces an edge case before any approval.)

Generic role names (thin lens — no curated profile)

These specialist names still resolve via the as action, but carry no curated constraints — the eval showed the base model already plays these roles, so a rich profile added surface without lift. They apply the role as a plain lens. Any role name not in this list (e.g. an ad-hoc one) falls back gracefully — the as action lists what's available rather than crashing. Want durable value from one of these lenses? Use the define action to create your house version with a failure-mode constraint.

NameRolePlain lens for
engineeringengineeringCode quality, architecture, implementation
productproductRequirements, UX, design review, business strategy
datadata-engineeringPipeline design, ML guidance, analytics
jambrainstormingIdeation, exploration, multi-perspective analysis

Custom Personas

Create your own with the define action — the mechanism that lets an enterprise inject ITS house personas. Stored in project-scoped DomainStore.

/wicked-garden-persona define pragmatic-tech-lead \
  --focus "delivery over perfection — ship iteratively, measure, adjust" \
  --traits "direct,pragmatic,cost-aware" \
  --role engineering

Use --save to promote to plugin-level cache for cross-project reuse:

/wicked-garden-persona define my-persona --focus "..." --save

Persona Schema

Each persona has rich characteristics that create a genuine perspective:

name        — kebab-case identifier
description — one-line summary
focus       — the lens this persona applies
traits      — behavioral adjectives
role        — category for --role filtering
personality — style, temperament, humor
constraints — non-negotiable rules (MUST follow)
memories    — formative experiences that inform judgment
preferences — communication, code_style, review_focus, decision_making

Cross-Domain Integration

Route any code review through a persona's lens via the engineering domain's review action with --persona:

engineering review --persona platform
engineering review --persona qe

Falls back to the built-in architect persona if the named persona is not found.

Fallback Personas

If specialist.json is unavailable, three built-in personas are always available:

  • architect — system design, trade-offs, structural decisions
  • skeptic — challenges assumptions, finds edge cases
  • advocate — champions end-user perspective and simplicity

Source Priority

When names collide across sources, higher priority wins:

custom (DomainStore) > cache (the harness's plugin cache) > builtin > fallback

This lets you override a built-in persona with a project-specific version.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.