Skip to content
SkillModel-led6,965 tokens in use

Reimagine Work

A weekly review of an agent fleet that finds where work waits on a human, ranks fixes and sets up a shadow owner judge, while every hard gate stays human.

3,862 stars on the repository MIT Licenseericosiu/ai-marketing-skills · reimagine-work/skills/reimagine-work
npx skills add ericosiu/ai-marketing-skills -s reimagine-work
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing Reimagine Work costs

28tokens idle
6,965tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

Nothing detected: no shell, network, file writes, git or credentials in its files.

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Excellent
  • Frontmatter8/15
  • Structure25/25
  • Examples15/15
  • Output15/15
  • Size13/15
  • Resources10/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity88
  • Maintenance44
  • Community56
  • Maturity28
  • Momentum40

Reimagine Work is a template for a weekly pass over a fleet of permanent agents. It runs five steps in order: question where agents wait on a human, delete what protects nothing, simplify to one owner per job, speed up the waits and automate through an escalation ladder.

Findings come from files the agent opens in that run, such as the approval ledger, outcome records, judge logs and queues. The output is a short ranked list, a leverage score and exact yes lines for the owner.

What You Get

  • A top-ranked list of moves with the evidence path, count and estimated minutes saved per week with the math shown
  • Copy-paste yes lines for each gated move, plus one delete line and one ask for a tenfold improvement
  • A leverage score out of 100 with what would raise it next week
  • A stream check marking each agent lane as drop or stream, with signal-to-shipped time and cost per changed outcome
  • The judge’s status by ask type against the go-live bar of 30 shadow cases with zero misses

Common Use Cases

  • Finding approvals the owner keeps giving for the same kind of ask
  • Merging agents that do the same job under one owner
  • Moving a lane from batch work to continuous work without removing its hard gates

What Makes It Unique

The loop never asks the owner something already answered: it first checks the owner’s own words, a lane owner’s approval, a standing approval with a cap and whether the action is reversible. New checks start in shadow mode, earn autonomy per ask type and drop back to shadow after one miss. Sends, spend, publishing, deletes and merges always stay human. A weekly cold-start test and an outside reviewer model challenge the findings.

Under The Hood

The skill is a single long instructions file with no scripts. It tells the agent to read ledgers, logs, queues, chats, email and calendar data in read-only mode, to run the existing tests before and after any build, and to write only its own report, its own state file and reversible internal checks with a backup beside each changed file. It sends one call to a second model for an outside review each week. It states that it sends, posts, spends, deletes, merges and edits skills never, and only proposes them for the owner to decide.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers