content-research-writer

A token-efficient writing partner that uses bounded research, numbered citations, evidence packets, and parallel sub-agents to outline, draft, and refine high-quality content with minimal context growth.

Install with Tessl CLI

npx tessl i github:0xrabbidfly/eric-cartman --skill content-research-writer

What are skills?

1.34x

Quality

40%

Does it follow best practices?

Impact

94%

1.34x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./.github/skills/content-research-writer/SKILL.md

SKILL.md

Review

Evals

Evaluation results

92%

41%

Setting Up a Research Project for a Technical Article

Evidence packet structure and folder organization

Criteria

Without context

With context

Folder structure prefix naming

20%

Evidence packet file naming

70%

100%

Packet ID field

100%

Target claim field

100%

Evidence bullets cap

100%

Quote cap

100%

Does not prove field

50%

100%

Confidence field

100%

Intended placement field

100%

Citation-ready reference entry

40%

100%

One packet per claim (atomic)

100%

Without context: $0.7069 · 2m 37s · 19 turns · 22 in / 5,797 out tokens

With context: $1.3076 · 4m 12s · 30 turns · 2,502 in / 8,943 out tokens

100%

17%

Research Planning for a Cloud Cost Optimization Article

Research brief and source planning

Criteria

Without context

With context

Claims-to-prove count

100%

Claim type labels

100%

Audience definition

100%

Goal and success criteria

100%

Scope boundaries

100%

Voice constraints

100%

Source log file exists

100%

Selected/rejected decisions

100%

First-party source preference

100%

Publication date noted

80%

100%

Without context: $1.2469 · 4m 42s · 17 turns · 19 in / 7,987 out tokens

With context: $1.2579 · 3m 43s · 28 turns · 284 in / 7,198 out tokens

91%

15%

Write a Cited Technical Blog Post on Observability

Citation-first drafting with numbered references

Criteria

Without context

With context

Numbered in-text citations

100%

No uncited factual claim

80%

100%

References section present

100%

Reference entry format: author/org

100%

Reference entry format: date and URL

50%

75%

Reference entry format: accessed date

75%

Evidence packets exist

100%

No large verbatim evidence blocks

100%

Source budget respected

100%

50%

Draft file naming

100%

Without context: $1.3874 · 4m 33s · 31 turns · 225 in / 9,837 out tokens

With context: $1.5728 · 5m 40s · 34 turns · 1,578 in / 12,416 out tokens

Evaluated: 1 day ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Setting Up a Research Project for a Technical Article Research Planning for a Cloud Cost Optimization Article Write a Cited Technical Blog Post on Observability

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.