fnd-skill-finder

name	fnd - Skill Finder
description	Find the right GOSM skill for what you want to do. Describe your need in plain language and get matched to the best skill(s).
output	{"format":"prose"}

Skill Finder

Input: $ARGUMENTS

---

Purpose

You have 318 skills. You don't need to remember them. Describe what you want to do and this skill matches you to the right one(s).

---

Instructions

1. Read the user's input (what they want to do)
2. Scan the registry below for the best semantic matches
3. Return top 3-5 matches, ranked by relevance
4. For each match, explain in one sentence why it fits and how it differs from the other matches
5. If no strong match exists, say so and show the closest options
6. If the input is ambiguous, briefly explain the tradeoff between the top candidates so the user can choose

Output Format

SKILL FINDER RESULTS
====================

Your request: "[user's input]"

TOP MATCHES:

1. /[skill_name] - [description]
   Why: [1 sentence on why this fits]

2. /[skill_name] - [description]
   Why: [1 sentence on why this fits]

3. /[skill_name] - [description]
   Why: [1 sentence on why this fits]

[If matches are similar:]
KEY DIFFERENCES:
- /skill_a focuses on [X], while /skill_b focuses on [Y]
- Use /skill_a when [condition], use /skill_b when [condition]

Matching Priority

Match on intent, not keywords. The user might say:

• "figure out what went wrong" → root_cause_analysis, debugging, failure_attribution
• "help me decide" → multi_criteria_decision, decision_trees, selection
• "think through this deeply" → araw, uaua, meta_reasoning_core
• "make sure I haven't missed anything" → mece_validation, comprehensive_aspects, checklist_search

When Multiple Skills Form a Workflow

If the user's need spans multiple skills, say so:

SUGGESTED WORKFLOW:
1. First: /skill_a (to do X)
2. Then: /skill_b (to do Y)
3. Finally: /skill_c (to do Z)

---

Skill Registry (318 skills)

Categories

Skills fall into these categories (a skill can belong to multiple):

• communication - Writing, speaking, presenting, narrative
• decision_making - Choosing between options, evaluating tradeoffs
• goal_analysis - Understanding, decomposing, refining goals
• guessing - Generating hypotheses, exploring unknowns
• meta - GOSM system internals, framework maintenance
• planning - Ordering, scheduling, strategy, roadmaps
• problem_solving - Diagnosing issues, finding root causes
• reasoning - Logic, analysis, mental models, evidence
• search - Finding information, options, solutions
• validation - Checking, verifying, testing, auditing

Full Registry

Skill	Description	Categories
academic_mastery	Master academic subjects through structured learning, concept mapping, and competency verification	-
active_listening	Systematic procedure for listening deeply to understand others, build trust, and improve communication outcomes	goal_analysis
active_recall	Use retrieval practice and self-testing to strengthen learning and identify gaps	problem_solving, validation
adaptive_extraction_pipeline	Breadth-first, learned extraction pipeline that clarifies goals first, samples broadly, learns user preferences, and extracts selectively from highest-value items	goal_analysis
adversarial_review	Nothing is a fact until it survives an assassination attempt. Builder constructs claims, Breaker tries to destroy them	validation, guessing
advocacy_infrastructure_setup	Set up foundational infrastructure for autonomous advocacy operations	-
after_action_review	Structured debrief to learn from completed actions - what happened, why, what to do differently	-
ai_biomedical_agent	AI agent for biomedical research tasks	meta
algorithmic_optimization	Orderings based on classic algorithm design paradigms: greedy, dynamic programming, divide-and-conquer	planning
alphacode_pass1_explicit	First pass explicit analysis for competitive programming problems	meta
alphacode_pass2_implicit	Second pass implicit analysis for competitive programming problems	meta
alphacode_pass3_meta	Third pass meta-analysis for competitive programming problems	meta
analogy_search	Many problems have been solved before - just in different domains. Find and transfer solutions	problem_solving, search
anticipated_failures_analysis	Pre-mortem: what could go wrong and how to prevent it	reasoning
api_design	Procedure for designing, implementing, and documenting effective APIs	-
api_middleman_strategies	Strategies for working around API limitations including rate limits and access restrictions