fractal

Delegate analysis- and context-heavy work to Fractal, an agentic CLI powered by a self-harnessed Recursive Language Model (predict-rlm), by running it non-interactively (fractal -p). Reach for it when a task needs reasoning over a large or deep codebase, synthesizing an answer across many files, auditing, or open-ended investigation — work that would otherwise flood your own context. The RLM reasons over context programmatically (no context rot) and returns a distilled answer. Use when asked to run Fractal headless, script it, call it from CI or another agent, or offload a heavy analysis/large-context task. Less suited to trivial single-file edits you can do directly.

Quality

96%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Quality

Content

92%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A tight, highly actionable body with a clear sequenced workflow and validation checkpoints; its only real weakness is progressive disclosure, where the repeatedly-referenced RECIPES.md bundle file is missing.

Suggestions

Add the missing RECIPES.md bundle file (or remove the RECIPES.md links and inline the promised CI-gate, multi-turn-driver, diff-review, and structured-output patterns), so the signaled one-level-deep references actually resolve.

Remove the duplicated stream-capture tip (it appears in both Quick start and Output contract) to save tokens without losing clarity.

Tighten the 'When to reach for it' section, whose example prompts and trigger framing largely restate the frontmatter description.

Dimension	Reasoning	Score
Conciseness	The body is lean and operational throughout — install check, preflight, quick-start command, an output-contract table, and a flags list — assuming Claude's competence and avoiding explanations of concepts Claude already knows. Minor repetition of the stream-capture tip and some value-prop prose keeps it just at level-3 rather than the padded level-1 or loosely-tightened level-2.	3 / 3
Actionability	Provides fully executable, copy-paste-ready commands — the install one-liner, `fractal -p "..." --workspace /path`, `docker info`, `fractal config status`, `git diff \| fractal -p "review this diff"`, and concrete flag syntax — matching the level-3 'fully executable, copy-paste ready' anchor.	3 / 3
Workflow Clarity	Sections form a clear sequence (Install → Preflight → Quick start → Output contract → Flags → Tips) with explicit validation checkpoints in Preflight ('docker info >/dev/null \|\| echo ...', 'fractal config status') and exit-code semantics as a feedback loop; operations are not destructive/batch so the level-2 cap does not apply.	3 / 3
Progressive Disclosure	The body is well-organized and defers detail to a one-level-deep, clearly signaled `[RECIPES.md](RECIPES.md)` reference, but RECIPES.md is not present in the bundle, so the signaled navigation is broken — landing it at level-2 ('references present but ... could be better organized') rather than the clean level-3 'easy navigation' anchor.	2 / 3
	Total	11 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, third-person description that clearly states concrete capabilities, gives multiple explicit 'use when' triggers, and carves out a distinct niche with negative guidance. No fluff or over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'running it non-interactively (fractal -p)', 'reasoning over a large or deep codebase, synthesizing an answer across many files, auditing, or open-ended investigation', and 'script it, call it from CI or another agent' — matching the level-3 anchor of several specific concrete actions, not the partial level-2.	3 / 3
Completeness	Explicitly answers what (delegate analysis/context-heavy work to Fractal run via `fractal -p`) and when via two explicit trigger clauses — 'Reach for it when...' and 'Use when...' — plus negative guidance ('Less suited to trivial single-file edits'), satisfying the level-3 'what AND when with explicit triggers' anchor.	3 / 3
Trigger Term Quality	Covers natural phrasings a user would say — 'audit', 'synthesizing an answer across many files', 'open-ended investigation', 'run Fractal headless', 'call it from CI or another agent', 'offload a heavy analysis/large-context task' — giving good coverage of natural terms rather than jargon-only.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (offloading heavy/large-context analysis to one specific CLI) with distinct triggers and explicit 'less suited to' negative guidance, making it unlikely to fire for the wrong skill.	3 / 3
	Total	12 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
relative_links	Relative link issues: 2 missing	Warning

	Total	15 / 16 Passed

Repository: jasonkneen/tiny-world-builder
Path: .agents/skills/fractal/SKILL.md
Commit: 039a20a

Reviewed: 3 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.