---

name: add-to-awesome-list description: > Guide a contributor through the full process of adding a new service to the awesome-agent-native-services catalog: verifying criteria, opening an issue, writing the service file in the correct format, and submitting a PR. Use this when the user says "I want to add X to the list" or "how do I contribute?" license: CC0-1.0 compatibility: Works with any agent that can read, write, and use git. metadata: repo: https://github.com/haoruilee/awesome-agent-native-services catalog-version: "2026-03" allowed-tools: WebSearch Read Write Shell

Skill: add-to-awesome-list

Use this skill to walk a contributor through the complete process of adding a new service to the awesome-agent-native-services catalog, from initial eligibility check through a merged PR.

When to activate

Activate when the user says:

• "I want to add [service] to the list."
• "How do I contribute a new service?"
• "I found a service that should be in the catalog."
• "Walk me through the contribution process."

Phase 1 — Pre-flight eligibility check

Before writing anything, verify the service qualifies.

1.1 Check for duplicates

1. Read README.md — search for the service name in all 12 category sections.
2. Search open issues: https://github.com/haoruilee/awesome-agent-native-services/issues?q=<service-name>
3. If found: report the existing entry or issue URL and stop.

1.2 Apply the five criteria

Use the evaluate-agent-native skill (or apply the criteria inline):

For each criterion, find evidence from the official homepage or docs — a direct quote or specific API documentation. Do not infer. Do not use marketing blog posts alone.

If any criterion fails: tell the user clearly which criterion failed, why, and what the correct classification is (agent-adapted or agent-builder). Do not proceed to Phase 2.

1.3 Determine the category

Use this map:

What the service does	Category folder
Gives agents a communication identity (email, messaging)	communication/
Provides remote browser or web extraction for agents	browser-and-web-execution/
Runtime tool discovery, delegated OAuth, managed tool execution	tool-access-and-integration/
Agent-initiated human approval gates	oversight-and-approval/
Agent wallets, KYA identity, autonomous payment execution	commerce-and-payments/
Execution environments, session isolation, secrets, gateway	agent-runtime-and-infrastructure/
Persistent, self-managing agent memory	memory-and-state/
LLM-optimized web search and content retrieval	search-and-web-intelligence/
Secure sandboxes for AI-generated code	code-execution/
Agent trajectory tracing, evaluation, cost attribution	observability-and-tracing/
Fault-tolerant long-running agent workflows	durable-execution-and-scheduling/
Agent presence in voice/video meetings	meeting-and-conversation/

If none fit, consider whether a new category is needed (requires 2+ qualifying services — see the new category issue template).

---

Phase 2 — Open a GitHub issue

Do not write the PR before the issue is approved.

Direct the user to open an issue using the structured template:

Issue URL: https://github.com/haoruilee/awesome-agent-native-services/issues/new?template=01-new-service.yml

Help the user fill in the required fields:

• Service name and official website
• Official tagline — exact quote from the homepage
• Proposed category — from Phase 1
• Criterion evidence — one field per criterion, quote + source URL
• MCP status — is there a published MCP server? Link it.
• Agent Skills status — is there a published SKILL.md? Provide the install command.
• Classification — agent-native
• Generic alternative comparison — name the obvious human-facing alternative and why it fails

After submission, wait for a maintainer ✅ Go before proceeding.

---

Phase 3 — Write the service file

Once the issue has a ✅ Go, create the service file.

File location

services/{category-folder}/{service-name}.md

Naming rules:

• All lowercase
• Hyphens only (no underscores, no spaces)
• Match the service's common name
• Examples: agentmail.md, trigger-dev.md, amazon-bedrock-agentcore.md

Required sections (in this order)

Every service file must contain all 13 sections. Use the template below.

# {Service Name}

> **"{Official tagline from homepage — exact quote}"**

| | |
|---|---|
| **Website** | https://... |
| **Docs** | https://... |
| **GitHub** | https://... (if public) |
| **Classification** | `agent-native` |
| **Category** | [{Category Name}](README.md) |
| **Funding** | (optional: e.g., $6M seed · Y Combinator) |
| **Compliance** | (optional: e.g., SOC 2 Type I) |

---

## Official Website

https://...

---

## Official Repo

https://github.com/...

(If multiple repos, list each with a brief label.)

---

## Agent Skills

**Status:** ✅ Available / ⚠️ Not yet published

<!-- If available: -->
```bash
npx skills add {org/repo}

Skill	What It Teaches the Agent
skill-name	One-line description

Compatibility: Claude Code, Cursor, Codex, and all AgentSkills-compatible tools.

No official skill published yet. Search community skills:

npx clawhub@latest search {service-name}

See https://agentskills.io/specification to contribute one.

---

MCP

Status: ✅ Available / ⚠️ Not yet published

Detail	Value
MCP Repo	https://github.com/...
Transport	stdio / Streamable HTTP
Auth	API key env var name
Compatible Clients	Claude Desktop, Cursor, Windsurf, ...

No MCP server published yet. Integration is via REST API / SDK.

---

What It Does

One to three paragraphs explaining what the service does and why it is agent-native. The core abstraction should be stated clearly: {the agent-specific thing}.

---

Why It Is Agent-Native

Criterion	Evidence
Agent-first positioning	"{exact quote}" — {source URL}
Agent-specific primitive	{description of the primitive with no human equivalent}
Autonomy-compatible control plane	{how agents operate without per-action human confirmation}
M2M integration surface	{SDK name + install command, REST API URL, MCP server, webhooks}
Identity / delegation	{how agent identity, delegated permissions, and audit trail work}

---

Primary Primitives

Primitive	Description
{Name}	{one sentence}

(List 4–10 primitives. Each should be a distinct, named agent-specific abstraction.)

---

Autonomy Model

Step-by-step numbered list showing how an agent uses the service end-to-end without human intervention. Use code blocks for API calls where relevant.

Step 1: Agent calls {endpoint} → receives {result}
Step 2: ...
Step N: {final outcome, no human clicked anything}

---

Identity and Delegation Model

Bullet list explaining:

• How the agent gets its own identity (not shared with a human)
• How permissions are delegated from operator/user to agent
• How the audit trail attributes actions to the specific agent

---

Protocol Surface

Interface	Detail
REST API	{base URL or description}
Python SDK	pip install {package}
TypeScript SDK	npm install {package}
MCP Server	{repo URL or "not yet available"}
Webhooks	{event types}

---

Human-in-the-Loop Support

One paragraph. Is HITL required? Optional? How is it integrated?

---

Why Generic Alternatives Do Not Qualify

Alternative	Why It Fails
{Name}	{specific reason referencing criteria}

(List 3–5 obvious human-facing alternatives and why each fails.)

---

Use Cases

• {Use case title} — {one sentence description of how an agent uses this service}

(List 4–6 concrete, specific use cases.)

### Research checklist for each section

Before writing each section, verify the information from primary sources:

- [ ] **Official Website / Repo**: from the service's GitHub or homepage
- [ ] **Agent Skills**: check skills.sh, agentskills.io, and the service's GitHub for `SKILL.md` files
- [ ] **MCP**: check the service's GitHub for `mcp-server` repos; check mcp.so, glama.ai/mcp, smithery.ai
- [ ] **Why It Is Agent-Native**: direct quotes from docs.{service}.com or {service}.com/docs
- [ ] **Protocol Surface**: check the SDK README for install commands
- [ ] **Use Cases**: sourced from the service's own documentation, not invented

---

## Phase 4 — Update the index files

After creating the service file, update two more files:

### 4.1 Update `services/{category}/README.md`

Add a row to the service table:

```markdown
| [{Service Name}]({service-name}.md) | {tagline} | {key primitives} | ✅/⚠️ |

4.2 Update root README.md

Find the correct category section and add a row to the summary table:

| [{Service Name}](services/{category}/{service-name}.md) | {tagline} | {key primitives} | ✅/⚠️ |

---

Phase 5 — Open the PR

PR title format

[New Service] {Service Name}

PR body checklist

The PR template (.github/PULL_REQUEST_TEMPLATE.md) will auto-populate. Fill in:

• Linked issue: Closes #{issue number}
• Five-criterion evidence table
• All 13 required sections confirmed
• Classification confirmed as agent-native
• Conflict-of-interest declaration

Git commands

git checkout -b add-{service-name}
git add services/{category}/{service-name}.md
git add services/{category}/README.md
git add README.md
git commit -m "[New Service] {Service Name}

Closes #{issue number}

- Category: {category}
- MCP: {✅ available / ⚠️ not yet}
- Agent Skills: {✅ available / ⚠️ not yet}"
git push origin add-{service-name}

Then open a PR on GitHub targeting the main branch.

---

Quality checklist before submitting

• All links are live (clicked each one)
• Official tagline is an exact quote from the homepage (in quotation marks)
• All criterion evidence cites a specific URL
• Service file name is lowercase with hyphens only
• All 13 required sections are present
• Agent Skills section shows real install commands (not invented)
• MCP section shows real repo/transport (not invented)
• No undisclosed financial interest in the service