Agent skill

Agent Development Specification

by SmartSunruiyang in SmartSunruiyang/Agent-development-specification

Bootstraps the Day-0 agent-governance + documentation scaffold for ANY project on ANY stack.

Install Agent Development Specification

skills CLI

$ npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI

$ gh skill install SmartSunruiyang/Agent-development-specification agent-development-specification --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Claude Code skills documentation · loads skills from .claude/skills/

Paste into Claude Code

Install the "agent-development-specification" agent skill from https://github.com/SmartSunruiyang/Agent-development-specification/tree/main into .claude/skills/agent-development-specification/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-development-specification", 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.

skills CLI

$ npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a codex

Project install goes to .agents/skills/; add -g for ~/.codex/skills/.

GitHub CLI

$ gh skill install SmartSunruiyang/Agent-development-specification agent-development-specification --agent codex

Project scope by default (.agents/skills/); add --scope user for a personal install.

Codex skills documentation · loads skills from .agents/skills/

Paste into Codex

Install the "agent-development-specification" agent skill from https://github.com/SmartSunruiyang/Agent-development-specification/tree/main into .agents/skills/agent-development-specification/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-development-specification", 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.

skills CLI

$ npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a cursor

Project install goes to .agents/skills/; add -g for ~/.cursor/skills/.

GitHub CLI

$ gh skill install SmartSunruiyang/Agent-development-specification agent-development-specification --agent cursor

Project scope by default (.agents/skills/); add --scope user for a personal install.

Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/

Paste into Cursor

Install the "agent-development-specification" agent skill from https://github.com/SmartSunruiyang/Agent-development-specification/tree/main into .cursor/skills/agent-development-specification/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-development-specification", 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.

skills CLI

$ npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a gemini-cli

Project install goes to .agents/skills/; add -g for ~/.gemini/skills/.

GitHub CLI

$ gh skill install SmartSunruiyang/Agent-development-specification agent-development-specification --agent gemini-cli

Project scope by default (.agents/skills/); add --scope user for a personal install.

Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/

Paste into Gemini CLI

Install the "agent-development-specification" agent skill from https://github.com/SmartSunruiyang/Agent-development-specification/tree/main into .gemini/skills/agent-development-specification/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-development-specification", 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.

GitHub CLI

$ gh skill install SmartSunruiyang/Agent-development-specification agent-development-specification

Installs 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).

skills CLI

$ npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a github-copilot

Project install goes to .agents/skills/; add -g for ~/.copilot/skills/.

GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/

Paste into GitHub Copilot

Install the "agent-development-specification" agent skill from https://github.com/SmartSunruiyang/Agent-development-specification/tree/main into .github/skills/agent-development-specification/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-development-specification", 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.

skills CLI

$ npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a opencode

OpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.

GitHub CLI

$ gh skill install SmartSunruiyang/Agent-development-specification agent-development-specification --agent opencode

Project scope by default (.agents/skills/); add --scope user for a personal install.

OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/

Paste into OpenCode

Install the "agent-development-specification" agent skill from https://github.com/SmartSunruiyang/Agent-development-specification/tree/main into .opencode/skills/agent-development-specification/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-development-specification", 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: agent-development-specification
GitHub stars: 150
Token cost: ~2.5k tokens
SKILL.md length: 1,117 words
Files: 24 (incl. scripts, references, assets)
Skills in repo: 1
Repo updated: Jun 1, 2026
First seen: Oct 7, 2026
Licence: MIT

At a glance

Bootstraps the Day-0 agent-governance + documentation scaffold for ANY project on ANY stack.

Works in 4 steps: Orient & detect state → Mechanical scaffold (deterministic) → Guided interview (fills the real content) → …
The user runs /ads:init
SKILL.md covers Prime directive: obey the…, The init workflow, Language of generated docs and Resources, plus 1 more section
Calls git and python3

What it does

Agent Development Specification is an agent skill from SmartSunruiyang/Agent-development-specification. Bootstraps the Day-0 agent-governance + documentation scaffold for ANY project on ANY stack. Generates the agent rule system (CLAUDE.md, AGENTS.md, SELFCONSTRAINTS.md, VIBECODINGGUIDE.md) plus upstream docs (PRD, ARCHITECTURE with module cards + dependency DAG, ROADMAP, CONVENTIONS), initializes version control with a zero-point commit, and runs a guided requirements→architecture interview to fill them with real content. Works for greenfield AND retrofitting an existing codebase (reverse-engineers a draft…

Its SKILL.md is about 2.5k tokens, which your agent loads only when the skill is triggered. The skill folder holds 27 other files, including scripts, reference files and assets (for example `README.md`, `assets/templates/AGENTS.md` and `assets/templates/CLAUDE.md`).

It sits in Agent Workflows, covering Agent instruction files and Git workflow. The repository describes itself as: 一个通用、不绑定技术栈的 Claude Code 技能：在写第一行代码之前，先为项目铺好 Agent 规则、文档与架构，并在项目演进中持续把它们当作"源真相"。 The licence is MIT.

When your agent uses it

The user runs /ads:init
/agent-development-specification
Starts a NEW project
Asks to set up agent rules / governance / project spec / self-constraints

Example prompts

“t say”
“: laying down a project”
“Use the agent-development-specification skill to bootstrap the Day-0 agent-governance + documentation scaffold for ANY project on ANY stack”
“/agent-development-specification”

Requirements

Python 3

Workflow steps

4 steps, taken from the step headings in SKILL.md.

Orient & detect state
Mechanical scaffold (deterministic)
Guided interview (fills the real content)
Finalize & hand off

What it can do on your machine

Read from SKILL.md and the folder at commit 44eec9b. 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
Ships 1 file in scripts/, which the agent can run.
Shell commands in SKILL.md call:
- git
- python3
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use git, which can reach the network depending on how they are called.
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

Agent Development Specification loads about 2.5k tokens when it runs, and up to ~12k if it reads all its reference files. Until then it costs about 261 tokens; SKILL.md has 1,117 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it: ~261
When it runs · the whole SKILL.md, loaded when a task matches: ~2.5k
With references · SKILL.md plus every file in references/, read only if the agent opens them: ~12k

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); the scripts in this folder are not scanned.

SKILL.md

The full file from SmartSunruiyang/Agent-development-specification at commit 44eec9b, republished under its MIT licence (© SmartSunruiyang). 1,117 words, ~2,496 tokens.

Download SKILL.mdSave it as .claude/skills/agent-development-specification/SKILL.md (or your agent's skills folder). This skill also uses 23 other files; get the full folder from GitHub.

name: agent-development-specification
description: Bootstraps the Day-0 agent-governance + documentation scaffold for ANY project on ANY stack. Generates the agent rule system (CLAUDE.md, AGENTS.md, SELF_CONSTRAINTS.md, VIBECODING_GUIDE.md) plus upstream docs (PRD, ARCHITECTURE with module cards + dependency DAG, ROADMAP, CONVENTIONS), initializes version control with a zero-point commit, and runs a guided requirements→architecture interview to fill them with real content. Works for greenfield AND retrofitting an existing codebase (reverse-engineers a draft architecture from the code). Use whenever the user runs /ads:init or /agent-development-specification, starts a NEW project, or asks to set up agent rules / governance / project spec / self-constraints, scaffold CLAUDE.md / AGENTS.md / architecture docs, establish module boundaries before coding, or mentions 项目起步 / 起步指南 / Agent规范 / 自我约束 / 架构文档. Trigger even when the user doesn't say "skill" or "ads": laying down a project's rules, docs, and architecture before writing code is this skill's job.

Agent Development Specification (ADS)

This skill installs a project-startup governance system: the agent rules, the upstream documents, and the architecture that must exist before code, so that every later session — yours or another agent's — works against a stable source of truth instead of free-styling on a codebase nobody can fully see.

The full rationale lives in references/spec.md (the canonical, technology-agnostic spec). Read it when you need depth on why a rule exists; this file tells you what to do.

Prime directive: obey the rules you install

This skill's output is a set of hard constraints (SELF_CONSTRAINTS.md) and a practice guide. You must follow them while running init — installing a rulebook you then violate is worse than installing nothing. Concretely:

Version control from day 0. No file gets generated before the repo is under version control.
Non-destructive, always. Never overwrite a file the user has tuned. The scaffold script skips anything that exists; you do the same when filling content — read first, reconcile, never clobber.
Never decide architecture silently. Module splits, dependency direction, naming rules, and MVP scope have long-term consequences. Propose, explain the trade-off, let the user choose. Park assumptions in the PRD's "假设与待澄清项" for sign-off.
One purpose per run. init lands rules + docs. It does not refactor existing code or build features.

The init workflow

The skill's base directory (shown to you when this skill loaded) contains scripts/, assets/, and references/. Paths below are relative to it.

Phase 0 — Orient & detect state

Determine which mode you're in before touching anything:

git status / ls -a               → is there a version-control repo?
ls the source tree               → is there existing application code?
ls .project.agents/ (or chosen)  → is there already an ADS scaffold (full or partial)?

Repo?	Existing code?	Scaffold present?	Mode
no	no	no	Greenfield — fresh start
either	yes	no	Retrofit — see `references/retrofit.md`
either	either	yes (partial/full)	Top-up — fill only what's missing, never overwrite

Tell the user which mode you detected and what you're about to do, then proceed.

Phase 1 — Mechanical scaffold (deterministic)

Version control. If there's no repo, initialize one. If the working tree is dirty (uncommitted changes from prior work), STOP and ask whether to commit/stash first — do not pile the scaffold onto unrelated dirty changes (SELF_CONSTRAINTS §A2/F1).
Decide the structural values AND the project facts, and write a config:
json
```
{
  "PROJECT_NAME": "...", "SOURCE_DIR": "...", "TEST_DIR": "...", "include_uiux": true,
  "PROJECT_TYPE": "...", "TECH_STACK": "...", "ENTRY_POINT": "...",
  "BUILD_COMMAND": "...", "RUN_COMMAND": "...", "TEST_COMMAND": "..."
}
```
The script bakes all of these in deterministically (paths land in ARCHITECTURE §6; the project facts fill CLAUDE.md and AGENTS.md everywhere they appear). Fill in every project fact you can — these are the entry-point files every future agent reads first, and a missing build command is a high-cost error. Don't accept generic defaults: set SOURCE_DIR/TEST_DIR to the stack's real convention (detect them in retrofit; pick per-stack in greenfield — e.g. not "src" for a Go or Swift layout), and the commands to what actually builds/runs/tests this project. Set include_uiux to false for non-UI projects (CLI, library, backend service) so no UIUX references are generated. Agents dir defaults to .project.agents/ and docs to .project.agents/docs/context/ — keep these unless the user differs.
Run the scaffold script — it creates the tree, copies the static governance files, substitutes structural tokens, and skips everything that already exists:
sh
```
python3 <skill-dir>/scripts/scaffold.py --repo <repo-root> --config <config.json>
```
Run with --dry-run first if you want to preview. Read its report: it lists created vs. skipped files.
Zero-point commit (greenfield / clean tree only): stage the new files and commit the skeleton as the reference point, e.g. chore: scaffold agent-governance + doc skeleton (ADS Day 0). On a dirty existing repo, skip the auto-commit and tell the user to review and commit themselves.

After Phase 1 the repo has a complete, valid governance tree — even if you stop here, nothing is broken. The upstream docs still contain {{CONTENT}} tokens and  markers: those are Phase 2's job.

Show full SKILL.md (541 more words)Show less

Phase 2 — Guided interview (fills the real content)

Walk the user through references/interview-flow.md — the operational version of the spec's Steps 1–7 (requirements → domain model → modules → dependency DAG → roadmap → conventions). For each step: ask, write the answer into the corresponding doc (replacing tokens and removing the ADS:FILL comments), show the user, move on. Key discipline:

In retrofit mode, read references/retrofit.md first — you draft ARCHITECTURE from the existing code (prefer CodeGraph MCP if available) and present it for correction, rather than asking from scratch.
Pace it and scale it. One step at a time. A small project gets few modules and milestones; don't manufacture eight layers for a 200-line tool. The card/section format is fixed; the volume isn't.
The "不负责" line is the test of a module. If you can't write what a module does NOT do, its responsibility isn't crisp — re-split before moving on.
Verify the DAG is acyclic by hand. A cycle means two modules are tangled; resolve it now.
Finish CLAUDE.md and AGENTS.md — do not leave them as skeletons. It is tempting to lavish attention on PRD/ARCHITECTURE (the interesting docs) and abandon these two, but they are the FIRST files every future agent reads. After the interview, fill their remaining tokens and ADS:FILL sections — the architecture/module overview, conventions summary, test framework, and anything the config didn't already supply — and CONVENTIONS.md/UIUX.md too. A scaffold whose entry-point files still say {{TECH_STACK}} has failed at its one job.

Phase 3 — Finalize & hand off

Completion gate (run before committing). Grep the whole scaffold for anything left unfilled and resolve every hit — fill it with real content, or delete the line if it genuinely doesn't apply:
sh
```
grep -rn -e '{{[A-Z0-9_]*}}' -e 'ADS:FILL' <agents-dir>/   # expect: no output
```
A surviving {{TOKEN}} or ADS:FILL marker means the doc is half-written. The one exception is VIBECODING_GUIDE.md §8's current-phase note, which you should fill but is low-stakes. Do not commit with residue in CLAUDE.md, AGENTS.md, PRD.md, ARCHITECTURE.md, CONVENTIONS.md, or UIUX.md.
Resolve or explicitly defer each item in the PRD's "假设与待澄清项" with the user.
Update VIBECODING_GUIDE.md §8 "本项目当前阶段" to the real current state.
Commit the filled upstream docs, e.g. docs: define PRD/ARCHITECTURE/ROADMAP/CONVENTIONS (ADS init).
Offer Step 7 (derived docs — implementation spec, asset checklist) only if the project needs them.
Summarize: what was created, what's still ADS:FILL/deferred, and the recommended first milestone.

Language of generated docs

Default to the user's working language for the governance files and doc skeletons; the shipped templates are written in Chinese to match the canonical spec, so render them in another language if the project's working language differs. Keep AGENTS.md in English by convention — it's the cross-vendor contributor guide that non-Claude agents (e.g. Codex) read.

Resources

references/spec.md — the canonical universal spec (the full "why"; read for depth).
references/interview-flow.md — operational Step 1–7 interview guide (Phase 2).
references/retrofit.md — reverse-engineering a draft architecture from existing code.
assets/templates/ — the files written into the repo: governance (SELF_CONSTRAINTS.md, VIBECODING_GUIDE.md, CLAUDE.md, AGENTS.md, settings.json, log-README.md, gitignore-base) and doc skeletons under docs/ (PRD, ARCHITECTURE, ROADMAP, CONVENTIONS, UIUX).
scripts/scaffold.py — deterministic, idempotent, non-destructive tree + static-file creator.

Token conventions inside the templates

{{STRUCTURAL}} tokens (paths, doc names, PROJECT_NAME) — substituted by the scaffold script.
{{CONTENT}} tokens (e.g. {{BUILD_COMMAND}}, {{MODULE_OVERVIEW}}, {{FEATURE_NAME}}) — you fill these during the interview.
 — guidance for a section you must complete; delete the comment once filled.

Files

SKILL.md and 23 other files (scripts, references, assets) in the repository root of SmartSunruiyang/Agent-development-specification.

SKILL.md
LICENSE
README.md
assets/templates/AGENTS.md
assets/templates/CLAUDE.md
assets/templates/SELF_CONSTRAINTS.md
assets/templates/VIBECODING_GUIDE.md
assets/templates/docs/ARCHITECTURE.md
assets/templates/docs/CONVENTIONS.md
assets/templates/docs/PRD.md
assets/templates/docs/ROADMAP.md
assets/templates/docs/UIUX.md
assets/templates/gitignore-base
assets/templates/log-README.md
assets/templates/settings.json
dist/agent-development-specification.skill
evals
… and 7 more

Open the folder on GitHubat commit 44eec9b

Compare with similar skills

Agent Development Specification 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.

Agent Development Specification compared with similar skills
Skill	Stars	Used in	Tokens	Auto-check	Licence	Repo updated
Agent Development Specification this skillSmartSunruiyang/Agent-development-specification	150	—	~2.5k	Automated safety check: Pass	MIT	4 mo ago
Update Claude MdMikalaiBarysevich/CleverSwitch	116	—	~1.2k	Automated safety check: Pass	GPL-3.0	3 days ago
Claude Md Drift Auditalirezarezvani/ClaudeForge	429	—	~559	Automated safety check: Pass	MIT	4 mo ago
Zed Configwcygan/dotfiles	196	—	~930	Automated safety check: Pass	None	yesterday
200 Agents Mdjabrena/plinth	447	—	~820	Automated safety check: Pass	Apache-2.0	2 days ago
MindkeeperLeoYeAI/openclaw-master-skills	2.2k	—	~1.4k	Automated safety check: Pass	MIT	2 mo ago

Similar skills

Update Claude Md
MikalaiBarysevich/CleverSwitch
Analyze all changes in the current git branch and update CLAUDE.md if architectural or important changes warrant it.
116 GitHub stars~1.2k tokensUpdated 3 days ago
Agent WorkflowsAuto-check passed
Claude Md Drift Audit
alirezarezvani/ClaudeForge
Audit every CLAUDE.md in this project for drift against the last week of git history.
429 GitHub stars~559 tokensUpdated 4 mo ago
Agent WorkflowsAuto-check passed
Zed Config
wcygan/dotfiles
Zed editor configuration expert. An agent skill from wcygan/dotfiles.
196 GitHub stars~930 tokensUpdated yesterday
Agent WorkflowsAuto-check passed
200 Agents Md
jabrena/plinth
A skill your agent uses when you need to generate an AGENTS.md file for a Java repository — covering project conventions, tech stack, file structure, commands, Git workflow, and contributor…
447 GitHub stars~820 tokensUpdated 2 days ago
Agent WorkflowsAuto-check passed
Mindkeeper
LeoYeAI/openclaw-master-skills
Time Machine for Your AI's Brain — version control for agent context files.
2.2k GitHub stars~1.4k tokensUpdated 2 mo ago
Agent WorkflowsAuto-check passed
Audit Standards
jeremylongshore/tons-of-skills-marketplace
Audits the current project against the development standards defined in ~/.claude/CLAUDE.md.
2.8k GitHub stars~1.8k tokensUpdated today
Agent WorkflowsAuto-check: notes

Questions about Agent Development Specification

What does Agent Development Specification do?

Bootstraps the Day-0 agent-governance + documentation scaffold for ANY project on ANY stack. Agent Development Specification is an agent skill from SmartSunruiyang/Agent-development-specification. Bootstraps the Day-0 agent-governance + documentation scaffold for ANY project on ANY stack.

When should I use Agent Development Specification?

Agent Development Specification fits situations like: the user runs /ads:init; /agent-development-specification; starts a NEW project; asks to set up agent rules / governance / project spec / self-constraints.

How do I install Agent Development Specification in Claude Code?

Run `npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a claude-code`. Or copy the skill folder (the SmartSunruiyang/Agent-development-specification repository) into .claude/skills/agent-development-specification in your project. Claude Code loads it when a task matches its description.

How do I install Agent Development Specification in Codex?

Run `npx skills add SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a codex`. Or copy the skill folder (the SmartSunruiyang/Agent-development-specification repository) into .agents/skills/agent-development-specification in your project. Codex loads it when a task matches its description.

Can I use Agent Development Specification 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 SmartSunruiyang/Agent-development-specification --skill agent-development-specification -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/agent-development-specification, .gemini/skills/agent-development-specification, .github/skills/agent-development-specification and .opencode/skills/agent-development-specification in your project.

What does Agent Development Specification need to run?

Going by SKILL.md and its folder, Agent Development Specification needs the command-line tools its instructions call (git and python3). Our summary lists: Python 3.

Does Agent Development Specification access the network?

SKILL.md contains no URLs. Its commands use git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Agent Development Specification 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. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Agent Development Specification use?

Agent Development Specification is published under the MIT licence (from the LICENSE file in the skill folder). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Agent Development Specification use?

About 2.5k tokens (SKILL.md is roughly 10k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 9.4k tokens, read only when the agent opens those files.

What are the alternatives to Agent Development Specification?

Skills that share tags, products or a category with Agent Development Specification: Update Claude Md (MikalaiBarysevich/CleverSwitch, 116 stars), Claude Md Drift Audit (alirezarezvani/ClaudeForge, 429 stars), Zed Config (wcygan/dotfiles, 196 stars) and 200 Agents Md (jabrena/plinth, 447 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Agent Development Specification?

SmartSunruiyang (a GitHub user) maintains it in SmartSunruiyang/Agent-development-specification, which has 150 GitHub stars. The repository was last updated on June 1, 2026.

Source: SmartSunruiyang/Agent-development-specification on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.

Install Agent Development Specification

Facts

At a glance

What it does

When your agent uses it

Example prompts

Requirements

Workflow steps

What it can do on your machine

Tool permissions

Runs code

Network

Credentials

Context cost

Safety

SKILL.md

Agent Development Specification (ADS)

Prime directive: obey the rules you install

The init workflow

Phase 0 — Orient & detect state

Phase 1 — Mechanical scaffold (deterministic)

Phase 2 — Guided interview (fills the real content)

Phase 3 — Finalize & hand off

Language of generated docs

Resources

Token conventions inside the templates

Files

Compare with similar skills

Similar skills

Update Claude Md

Claude Md Drift Audit

Zed Config

200 Agents Md

Mindkeeper

Audit Standards

Related

Questions about Agent Development Specification