unbias
Skill1 starsChanged 4 months agoMITView on GitHub
Use when the user wants to de-slop, humanize, unbias, or scrub their own draft, or to clean the writing they are working on of AI tells. Applies the BIAS.md tells catalog in the current session.
In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- biascheckSkill
- crew-consultSkill
- crew-reviewSkill
- experimentSkill
- multiplechoiceSkill
- priorartSkill
- secure-researchSkill
The file
Files
unbias
Overview
Apply the tells catalog at ../../shared/BIAS.md to the draft the operator is working on, in this
session, with the model already loaded. No script, no external model calls. This is the fast, everyday
de-slop pass. For an independent, outside-family authenticity score, that is the separate biascheck
skill.
Modes
- review (default): for each tell, quote the phrase, name the tell, propose the fix. Do not edit. Close with the cluster read: are the tells dense in one passage (AI-shaped) or scattered (probably just voice)?
- apply: rewrite in place, changing only what a flagged tell requires and preserving the author's voice, then give a one-line-per-fix change summary.
The passes (run in order, from the catalog)
- Scrub 1, machine tells: vocabulary by tier, punctuation led by the em-dash (do a literal strip
pass), structure and rhythm. See
../../shared/BIAS.md, "Scrub 1". - Scrub 2, reasoning-posture: the confession / defend / over-explain / announce-honesty family, at clause, paragraph, and section altitude. See "Scrub 2".
- Cluster and false-positive guard: a single tell is noise; flag with confidence only on clusters. Preserve genuine human signal (specifics, dated references, real asides, plain or non-native voice). See "How to use this".
Note
The catalog is the source of truth; do not restate its tell list here. Read ../../shared/BIAS.md.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.