paper-narrative

Judge and reshape the STORY a paper's figures tell. Input is the work itself — manuscript (or abstract) + figure deck — no hand-written brief. `paper_brief_prompt(abstract, captions)` hands you the prompt to write the brief yourself (pitch/vision/per-figure-claims); then you play a handling editor over the full deck and return hook_verdict (would Fig 1 make me send this for review?), arc (hook→mechanism→evidence→application), figure_moves (panels in the wrong figure), missing_panels (concrete analyses to RUN), kill_list, and boldest_defensible_fig1. Hands per-figure claims to `figure-composer`. Load when writing or revising a paper.

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

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, actionable, well-sequenced skill body with a clear convergence loop; the only real gap is that the referenced kernel.py is not bundled, leaving the one external reference pointing at a missing file.

Suggestions

Bundle kernel.py under scripts/ (or references/) so the exec() reference resolves to a real file in the skill bundle.

Confirm the path used in exec(open("paper-narrative/kernel.py")) matches the bundled location, or use a path relative to the skill directory.

Dimension	Reasoning	Score
Conciseness	Lean and purposeful (~55 lines): it assumes Claude's competence, names builders and outputs directly, and avoids explaining concepts Claude already knows; the Setup prose about the pure-skill architecture earns its place by explaining how to load the helpers.	3 / 3
Actionability	Gives concrete executable guidance — the exec(open("paper-narrative/kernel.py")) line and named builders (paper_brief_prompt, narrative_review_task, narrative_review_schema) with specific JSON output fields and per-step actions; absence of full code is acceptable for this instruction-oriented skill.	3 / 3
Workflow Clarity	Five clearly sequenced steps culminating in an explicit feedback/convergence loop ("Re-run step 2... Converge when would_send_for_review=='yes' and figure_moves / missing_panels are empty"), providing a clear validation checkpoint.	3 / 3
Progressive Disclosure	Sections are well-organized and the single reference to paper-narrative/kernel.py is clearly signaled and one level deep, but that referenced file is not present in the bundle (no scripts/ directory), so the disclosure structure cannot be verified and is incomplete.	2 / 3
	Total	11 / 12 Passed

Description

75%

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 highly specific, complete, and distinctive description with an explicit load trigger, weakened mainly by second-person voice and an over-reliance on internal jargon that crowds out natural user trigger terms.

Suggestions

Rewrite in third person (remove "you") to recover the specificity point lost to voice, e.g. "Judges and reshapes... hands the agent the prompt...".

Lead with natural user trigger terms ("review or reorganize a paper's figures", "is Figure 1 a hook") and push internal jargon (hook_verdict, kill_list, boldest_defensible_fig1) into the body.

Dimension	Reasoning	Score
Specificity	Names many concrete actions and outputs ("write the brief", "play a handling editor", "return hook_verdict...arc...figure_moves...kill_list...boldest_defensible_fig1"), which would be a 3, but the description uses second person ("hands you the prompt", "you play a handling editor"), so the specificity score is reduced by 1 per the voice guideline.	2 / 3
Completeness	Clearly answers both what (judge/reshape the figure story, write a brief, play handling editor, return named JSON fields) and when ("Load when writing or revising a paper" is an explicit trigger clause, so completeness is not capped at 2).	3 / 3
Trigger Term Quality	Includes the natural trigger "Load when writing or revising a paper" and domain terms (paper, figures, manuscript), but the description is dominated by internal jargon (hook_verdict, kill_list, boldest_defensible_fig1, paper_brief_prompt) that users would not naturally say, and common variations like "review/reorganize my figures" are missing.	2 / 3
Distinctiveness Conflict Risk	A clear niche — reshaping a paper's figure narrative and handing per-figure claims to figure-composer — with distinct triggers unlikely to fire for unrelated skills.	3 / 3
	Total	10 / 12 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: UnicomAI/wanwu
Path: configs/microservice/bff-service/configs/agent-skills/claude-science/paper-narrative/SKILL.md
Commit: f618458

Reviewed: about 13 hours 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.