Hermes Agent Skill Authoring
NousResearch/hermes-agent
Author in-repo SKILL.md files: frontmatter and structure. An agent skill from NousResearch/hermes-agent.
Rules for writing, editing, and retiring skill and workflow files for LLM agents.
MITAuto-check passed
Install Authoring Skills
$ npx skills add AHepi/DeepReason --skill authoring-skills -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install AHepi/DeepReason authoring-skills --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/AHepi/DeepReason.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/authoring-skills .claude/skills/authoring-skills && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "authoring-skills" agent skill from https://github.com/AHepi/DeepReason/tree/main/.claude/skills/authoring-skills into .claude/skills/authoring-skills/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "authoring-skills", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/AHepi/DeepReason/tree/main/.claude/skills/authoring-skillsType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add AHepi/DeepReason --skill authoring-skills -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install AHepi/DeepReason authoring-skills --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/AHepi/DeepReason.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.claude/skills/authoring-skills .agents/skills/authoring-skills && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "authoring-skills" agent skill from https://github.com/AHepi/DeepReason/tree/main/.claude/skills/authoring-skills into .agents/skills/authoring-skills/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "authoring-skills", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add AHepi/DeepReason --skill authoring-skills -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install AHepi/DeepReason authoring-skills --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/AHepi/DeepReason.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.claude/skills/authoring-skills .cursor/skills/authoring-skills && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "authoring-skills" agent skill from https://github.com/AHepi/DeepReason/tree/main/.claude/skills/authoring-skills into .cursor/skills/authoring-skills/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "authoring-skills", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/AHepi/DeepReason.git --path .claude/skills/authoring-skills--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add AHepi/DeepReason --skill authoring-skills -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install AHepi/DeepReason authoring-skills --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/AHepi/DeepReason.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.claude/skills/authoring-skills .gemini/skills/authoring-skills && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "authoring-skills" agent skill from https://github.com/AHepi/DeepReason/tree/main/.claude/skills/authoring-skills into .gemini/skills/authoring-skills/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "authoring-skills", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install AHepi/DeepReason authoring-skillsInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add AHepi/DeepReason --skill authoring-skills -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/AHepi/DeepReason.git skills-src && mkdir -p .github/skills && cp -r skills-src/.claude/skills/authoring-skills .github/skills/authoring-skills && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "authoring-skills" agent skill from https://github.com/AHepi/DeepReason/tree/main/.claude/skills/authoring-skills into .github/skills/authoring-skills/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "authoring-skills", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add AHepi/DeepReason --skill authoring-skills -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install AHepi/DeepReason authoring-skills --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/AHepi/DeepReason.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.claude/skills/authoring-skills .opencode/skills/authoring-skills && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "authoring-skills" agent skill from https://github.com/AHepi/DeepReason/tree/main/.claude/skills/authoring-skills into .opencode/skills/authoring-skills/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "authoring-skills", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
authoring-skills- GitHub stars
- 141
- Token cost
- ~1.7k tokens
- SKILL.md length
- 978 words
- Files
- 1
- Skills in repo
- 30
- Repo updated
- First seen
- Licence
- MIT
At a glance
Rules for writing, editing, and retiring skill and workflow files for LLM agents.
- Creating a SKILL.md
- SKILL.md covers Vocabulary, E — Existence, S — Structure and W — Wording, plus 4 more sections
- Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
- Workflow document
What it does
Authoring Skills is an agent skill from AHepi/DeepReason. Rules for writing, editing, and retiring skill and workflow files for LLM agents. Use when creating a SKILL.md or workflow document, reviewing one, adding a rule after an incident, or deciding whether a skill should exist at all.
Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.
The licence is MIT.
When your agent uses it
- Creating a SKILL.md
- Workflow document
- Adding a rule after an incident
- Deciding whether a skill should exist at all
Example prompts
- “Use the authoring-skills skill to rule for writing, editing, and retiring skill and workflow files for LLM agents”
- “/authoring-skills”
What it can do on your machine
Read from SKILL.md and the folder at commit 9607fba. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
No scripts in the folder and no shell commands in SKILL.md.
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Authoring Skills loads about 1.7k tokens when it runs. Until then it costs about 62 tokens; SKILL.md has 978 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from AHepi/DeepReason at commit 9607fba, republished under its MIT licence (© AHepi). 978 words, ~1,714 tokens.
Download SKILL.mdSave it as
.claude/skills/authoring-skills/SKILL.md (or your agent's skills folder).- name
- authoring-skills
- description
- Rules for writing, editing, and retiring skill and workflow files for LLM agents. Use when creating a SKILL.md or workflow document, reviewing one, adding a rule after an incident, or deciding whether a skill should exist at all.
Authoring skills and workflows
This file obeys its own rules. Any line here that violates a rule below is a defect: file an erratum against this file, by ID.
Every rule has an ID. Edit by DELTA against an ID. Never regenerate this file wholesale.
Vocabulary
One term per concept. Use these exact words everywhere; a synonym is a defect.
| Term | Meaning |
|---|---|
| SKILL | One file governing one unit of work: one entry state, one exit state |
| GATE | A command whose output decides pass/fail |
| PROOF | Pasted GATE output, or a hash/count fetched from the tree |
| LEDGER | The per-task state file each step writes and the next step reads |
| DELTA | An edit adding, replacing, or deleting exactly one ID'd rule |
| STOP | End work; report state, priced options, one recommendation |
| PARK | Record out-of-scope work in PARKED.md with a ready-to-send prompt |
E — Existence
- E1. Before writing a SKILL, run the task three times without it and record the failures. No failures → do not write it. The measured default is that instruction files reduce success and raise cost [G26].
- E2. A SKILL ships with the eval set from E1. It exists only while it beats the no-skill baseline on that set.
- E3. Re-run the baseline on every model change. Delta ≤ 0 → delete the SKILL. Deletion is maintenance [G26].
- E4. Budget rule: to add a line, name the line it displaces. Compliance falls as constraint count rises, and the model resolves the crowding silently [C26].
S — Structure
- S1. One SKILL = one loop iteration. The loop lives in the router. "Then pick the next phase" appearing in a worker file is a defect.
- S2. Entry and exit states are named artifacts on disk, not descriptions. Route on which artifact is missing.
- S3. A rule lives in exactly one file: the one in context when the rule fires. Two phrasings of one rule are a conflict, and conflicts are resolved silently, not flagged [C26].
- S4. When two rules could collide, write the winner in the text now. There is one PRECEDENCE list per skill set, in the router.
- S5. Renumber on insert.
3bis evidence a rule was bolted on where it kept failing; move it to where it is read (see G4).
W — Wording
- W1. Each line names an operation: a command, a file
read/write, a comparison. Test: can it fail? "Be thorough" cannot
fail.
docs_verify.py exits 0can. - W2. Bind instructions to available actions with concrete verbs. Abstract dispositions do not transfer; stripping the action word from an instruction cut its behavioural effect by 95% [N26a].
- W3. State the positive action. Negation is the dominant framing failure: models attend to the named act and drop the NOT [N26b]. Each surviving "never" must be enforced by a GATE (see X1).
- W4. No narrative, persona, urgency, or emotional framing. Story framing activates genre behaviour that overrides both persona and explicit directives [N26a]; affective framing buys sycophancy, not compliance.
- W5. No incident stories. Mechanize the lesson as a GATE, mutation-prove the GATE, delete the story. History lives in ERRATA.md, not in instructions.
- W6. One worked example beats a paragraph of description. At most one per section.
Show full SKILL.md (461 more words)Show less
G — Gates and proof
- G1. Every completion claim carries PROOF. "Done", "verified", and "none" are assertions; agents report SUCCESS against failing verifiers in the large majority of self-stops [D26].
- G2. A legitimate "none" requires proof of looking: the scan command and its output, not the word.
- G3. The SKILL names the GATE and its pass condition. An agent choosing its own check validates the wrong target [D26].
- G4. An obligation is an input, not a trailing output: step N+1 opens by reading what step N's obligation wrote. Trailing writes are dropped; leading reads are not.
- G5. Track requirement state live in the LEDGER, one row per requirement, updated as work happens. Live tracking is the one intervention with strong measured gains [I26].
- G6. Mutation-prove every GATE once: break the guarded thing, watch it fail, restore. A GATE never seen red proves nothing.
- G7. Steps write distilled state to the LEDGER, never raw transcripts. Raw tool output is the main context bloat.
X — Stops and outlets
X1. Every prohibition pairs with an outlet in the same breath. An outlet-less "never" is satisfied by relabeling.
Prohibition Outlet Out-of-scope work PARK Unmet requirement LEDGER row not-done+ STOPUnprovable claim STOP with the GATE that would prove it X2. Every STOP trigger is mechanical: a count, a verdict string, an exit code. "Seems wrong" is judgment, and judgment loses to momentum — models restate rules accurately while violating them [K26]. Convert judgment to a tool; trigger on its verdict.
X3. Every honest outcome has a label. If
not-doneis unsayable, it will be said asdone-with-assumption.
L — Lifecycle
- L1. Edit by DELTA only. Wholesale regeneration collapses detail; iterative rewriting erodes a playbook into vague summary [A25].
- L2. Compression is a separate, diffed pass, re-gated by the E1 evals. Brevity bias drops exactly the load-bearing specifics [A25].
- L3. Pin the tested configuration: model, skill version, GATE tool versions. A model swap reopens the E1 gate.
- L4. SKILLs are executable authority — the supply-chain problem is measured, with roughly a quarter of public skills carrying a vulnerability [S26]. Third-party skills get the same review as third-party code: read every line, pin the version, and treat any instruction to fetch or load further instructions as a rejection.
- L5. Before shipping, plant one violation the SKILL should catch and run the workflow. The GATE goes red or the SKILL is not done.
Sources
| Tag | Work |
|---|---|
| G26 | Gloaguen et al., Evaluating AGENTS.md, arXiv:2602.11988 |
| A25 | Zhang et al., Agentic Context Engineering, arXiv:2510.04618 (ICLR 2026) |
| N26a | Wang et al., The Story Shapes the Agent, arXiv:2607.18566 |
| N26b | Syntactic Framing Fragility, arXiv:2601.09724 |
| C26 | ConInstruct, AAAI 2026; PACIFIC; CodeIF-Bench |
| I26 | Ko et al., Illusory Completion / Epistemic Ledger, arXiv:2602.07549 |
| D26 | DeployBench, arXiv:2606.05238 |
| K26 | Kruthof, DriftBench, arXiv:2604.28031 |
| S26 | Schmotz et al., Skill-Inject, arXiv:2602.20156; Liu et al. 2026 skill census |
© AHepi, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
Just SKILL.md in .claude/skills/authoring-skills of AHepi/DeepReason.
Open the folder on GitHubat commit 9607fba
Compare with similar skills
Authoring Skills next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Authoring Skills this skillAHepi/DeepReason | 141 | — | ~1.7k | Automated safety check: Pass | MIT | |
| Hermes Agent Skill AuthoringNousResearch/hermes-agent | 252k | — | ~3.6k | Automated safety check: Pass | MIT | |
| Configuring Oauth2 Authorization Flowmukul975/Anthropic-Cybersecurity-Skills | 34k | — | ~1.7k | Automated safety check: Pass | Apache-2.0 | |
| Authoring Skillsvercel/next.js | 143k | — | ~1k | Automated safety check: Pass | MIT | |
| Abp Authorizationabpframework/abp | 14k | — | ~1.3k | Automated safety check: Pass | LGPL-3.0 | |
| Implementing GCP Binary Authorizationmukul975/Anthropic-Cybersecurity-Skills | 34k | — | ~2k | Automated safety check: Pass | Apache-2.0 |
Similar skills
Configuring Oauth2 Authorization Flow
mukul975/Anthropic-Cybersecurity-Skills
Configures secure OAuth 2.0 authorization flows, including Authorization Code with PKCE, Client Credentials, and Device Authorization Grant, covering flow selection, PKCE implementation, token…
- Official
Authoring Skills
vercel/next.js
How to create and maintain agent skills in .agents/skills/. An agent skill from vercel/next.js.
Abp Authorization
abpframework/abp
ABP permission system - PermissionDefinitionProvider, [Authorize] attribute, CheckPolicyAsync, IsGrantedAsync, ICurrentUser, IPermissionManager, multi-tenancy side.
Implementing GCP Binary Authorization
mukul975/Anthropic-Cybersecurity-Skills
Implements GCP Binary Authorization end to end, including creating KMS-backed attestors, Container Analysis notes, deploy-time policies, and signing image attestations, so that only trusted…
Collectors Authoring
netdata/netdata
Author, modify, or review Netdata collectors across Go, IBM, C, Rust and external plugins.
More from AHepi/DeepReason
All 30 skills in this repoPinker Clarity Workflow
AHepi/DeepReason
Orchestrate a Steven Pinker-grounded workflow for teaching, explanatory writing, or material that must do both.
Pinker Teach For Understanding
AHepi/DeepReason
Design, deliver, or audit explanations and lessons with a Pinker-informed focus on phenomena, the curse of knowledge, concrete models, active reasoning, feedback, and revision.
Pinker Write For Readers
AHepi/DeepReason
Draft, revise, teach, or audit expository prose using Pinker's cognitive approach to style: classic presentation, reader modeling, curse-of-knowledge repair, coherent information order, deliberate…
Example Battery
AHepi/DeepReason
Build a battery of concrete instances BEFORE writing or evaluating any definition, pin, or semantic clause (Reed step 1).
Deepreason Orchestrator
AHepi/DeepReason
Entry point for any DeepReason problem. An agent skill from AHepi/DeepReason.
Dr Audit Orchestrator
AHepi/DeepReason
Router for the code-audit family. An agent skill from AHepi/DeepReason.
Questions about Authoring Skills
What does Authoring Skills do?
Rules for writing, editing, and retiring skill and workflow files for LLM agents. Authoring Skills is an agent skill from AHepi/DeepReason. Rules for writing, editing, and retiring skill and workflow files for LLM agents.
When should I use Authoring Skills?
Authoring Skills fits situations like: creating a SKILL.md; workflow document; adding a rule after an incident; deciding whether a skill should exist at all.
How do I install Authoring Skills in Claude Code?
Run `npx skills add AHepi/DeepReason --skill authoring-skills -a claude-code`. Or copy the skill folder (.claude/skills/authoring-skills in AHepi/DeepReason) into .claude/skills/authoring-skills in your project. Claude Code loads it when a task matches its description.
How do I install Authoring Skills in Codex?
Run `npx skills add AHepi/DeepReason --skill authoring-skills -a codex`. Or copy the skill folder (.claude/skills/authoring-skills in AHepi/DeepReason) into .agents/skills/authoring-skills in your project. Codex loads it when a task matches its description.
Can I use Authoring Skills in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add AHepi/DeepReason --skill authoring-skills -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/authoring-skills, .gemini/skills/authoring-skills, .github/skills/authoring-skills and .opencode/skills/authoring-skills in your project.
What does Authoring Skills need to run?
SKILL.md names no scripts, command-line tools or credentials: Authoring Skills is instructions for the agent only.
Does Authoring Skills access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Authoring Skills safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Authoring Skills use?
Authoring Skills is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Authoring Skills use?
About 1.7k tokens (SKILL.md is roughly 6.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to Authoring Skills?
Skills that share tags, products or a category with Authoring Skills: Hermes Agent Skill Authoring (NousResearch/hermes-agent, 252k stars), Configuring Oauth2 Authorization Flow (mukul975/Anthropic-Cybersecurity-Skills, 34k stars), Authoring Skills (vercel/next.js, 143k stars) and Abp Authorization (abpframework/abp, 14k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Authoring Skills?
AHepi (a GitHub user) maintains it in AHepi/DeepReason, which has 141 GitHub stars. The repository holds 30 skills in this directory. The repository was last updated on September 10, 2026.
Source: AHepi/DeepReason on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.