heretic SKILL.md skillA

heretic from akillness/jeo-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add akillness/jeo-skills/heretic@v1.0

Pins v1.0 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
akillness/jeo-skills
Path
.agent-skills/heretic/SKILL.md
Versions recorded
2
Latest version
v1.0, committed Jul 27, 2026
First version recorded
Jul 1, 2026
Size
11,979 bytes
Repository stars
52
License
none declared

Runs directional ablation and refusal-direction analysis for open-weight models the user may modify; use to reduce benign over-refusal or measure refusal/KL trade-offs, not for training.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# heretic — Automatic Abliteration & Refusal-Direction Interpretability

> **Keyword**: `heretic` · `abliterate` · `decensor a model` · `remove refusals` · `refusal direction` · `plot residuals`
>
> **Responsible use.** Heretic is AGPL-3.0 software for research and for customizing models **you have
> the right to modify**. Respect each base model's license and every platform's Terms of Service when
> downloading, modifying, uploading, or serving weights. Abliteration lowers a model's refusal rate on
> *all* topics, including genuinely harmful ones — do not use it to produce a model for illegal content
> (e.g. CSAM, weapons-of-mass-destruction uplift, targeted harassment) or to evade safety controls you
> are contractually bound by. Legitimate uses include interpretability/alignment research, red-teaming,
> reducing benign over-refusal, and running uncensored models on your own hardware for lawful work.

heretic packages the [p-e-w/heretic](https://github.com/p-e-w/heretic) tool (AGPL-3.0) as an installable
jeo-skills plugin. Heretic implements a **parametrized variant of directional ablation** (abliteration,
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add akillness/jeo-skills/heretic@sha256:926568b18a360c29

The line at the top of this page pins the label v1.0, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_babz7txjsfpln5on.svg)](https://markdownregistry.com/a/art_babz7txjsfpln5on)

4 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 2 versions of this file, between Jul 1, 2026 and Jul 27, 2026, about one every 26 days.

versioncommittedcommitsizeaudit
v1.0 latest2026-07-27 056e3f8 11,979 BA view · diff
v1.02026-07-01 3fa4ca3 12,851 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (11979 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

akillness/jeo-skills · 52 stars · license none declared · pushed 2026-11-05 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_babz7txjsfpln5on
GET https://markdownregistry.com/api/v1/resolve?ref=akillness/jeo-skills/heretic
GET https://markdownregistry.com/api/v1/blob/926568b18a360c2943e5cbe9b02f1484d3920b59027bdff0d38052579f88ebab

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from akillness/jeo-skills

a2a-protocol SKILL.md skill
akillness/jeo-skills · .agent-skills/a2a-protocol/SKILL.md · Agent-to-Agent horizontal messaging — Linux Foundation AAIF 2026 standard. Use when multiple agents across…
git:20260920.2a2b733 · audit A · 52 stars
ab-test-analysis SKILL.md skill
akillness/jeo-skills · .agent-skills/ab-test-analysis/SKILL.md · Analyze A/B test results with statistical significance, sample size validation, confidence intervals, and…
git:20260919.9f2a908 · audit A · 52 stars
academic-research SKILL.md skill
akillness/jeo-skills · .agent-skills/academic-research/SKILL.md · Full academic research pipeline from discovery to publication — 4 reference pipelines, 27 modes, 39-agent ensemble…
v1.0.0 · audit A · 52 stars
agent-beacon SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-beacon/SKILL.md · Operate Asymptote-Labs/agent-beacon, the cross-harness coding-agent session history and self-improving memory layer…
v1.0.0 · audit A · 52 stars
agent-benchmarking SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-benchmarking/SKILL.md · Map agent capabilities to standard benchmarks (SWE-bench, WebArena, OSWorld, GAIA, TauBench) plus custom regression…
git:20260919.9f2a908 · audit A · 52 stars
agent-guardrails SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-guardrails/SKILL.md · Layered agent safety — prompt-injection defense, data exfiltration prevention, tool-misuse blocking via NeMo Guardrails…
git:20260919.9f2a908 · audit B · 52 stars
agent-manager SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-manager/SKILL.md · tmux + Python agent lifecycle manager — start, stop, monitor, and assign tasks to AI agents running in tmux sessions…
v1.0 · audit A · 52 stars
agent-memory-architecture SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-memory-architecture/SKILL.md · Design tiered agent memory — core / recall / archival — with backend selection (mem0, Letta, Zep/Graphiti, LangMem) and…
git:20260919.9f2a908 · audit A · 52 stars
agent-observability SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-observability/SKILL.md · Instrument LLM agents with traces, metrics, and replay. Use when an agent in production is silently failing…
git:20260919.9f2a908 · audit A · 52 stars
agent-principles SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-principles/SKILL.md · Core principles for collaborative development with AI agents. Defines divide-and-conquer, context management…
v2.0.0 · audit A · 52 stars
agent-pulse SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-pulse/SKILL.md · Operate and extend barretlee/agent-pulse, the evidence-backed AI industry intelligence system: inspect source catalog…
v1.0 · audit A · 52 stars
agent-reach SKILL.md skill
akillness/jeo-skills · .agent-skills/agent-reach/SKILL.md · Operate Panniantong/Agent-Reach, the MIT-licensed installer, doctor, and routing layer for platform-specific internet…
v1.0.0 · audit A · 52 stars

Every file in akillness/jeo-skills

Browse by kind, by grade A, or by owner.