Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/stacklok/toolhive-registry-server/smoke-testnpx skills add stacklok/toolhive-registry-server --skill smoke-testgit clone --depth 1 https://github.com/stacklok/toolhive-registry-serverWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/stacklok/toolhive-registry-server/smoke-test)<a href="https://agentmods.dev/skills/stacklok/toolhive-registry-server/smoke-test"><img src="https://agentmods.dev/badge/skills/stacklok/toolhive-registry-server/smoke-test.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00040 | $0.08653 |
| Opus 5 | $0.00020 | $0.04326 |
| Sonnet 5 | $0.00008 | $0.01731 |
| Haiku 4.5 | $0.00004 | $0.00865 |
Grade B, and why
smoke-test scanned grade B with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Sends data to an external URLmediumData exfiltration
A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.
curl_post() { curl -s -o /tmp/thv_body -w "%{http_code}" -X POST -H "Content-Type: application/json" -d "$2" "$BASE_URL$1"; } Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
description: Start the Registry Server via docker compose and run a suite of curl-based smoke tests covering system, read-only MCP API, admin API, entry lifecycle, and OAuth/auth enforcement scenarios. How it starts
The opening of the file, as written. The whole thing — 753 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Smoke Tests — Registry Server
Starts the Registry Server stack with docker compose, waits for it to be ready, then executes a suite of curl smoke tests organised as Gherkin scenarios. Pass keep-up as an argument to leave the stack running after the tests complete (useful for interactive exploration).
Embedded Scenarios
Feature: Registry Server smoke tests
Background:
Given the Registry Server is running at http://localhost:8080
And the "default" registry is seeded from the upstream-registry.json file source
# ── System endpoints ──────────────────────────────────────────────────────
Scenario: Health check
When I GET /health
Then the response status is 200
And the body contains "healthy"
Scenario: Readiness check
When I GET /readiness
Then the response status is 200
And the body contains "ready"
Scenario: Version endpoint
When I GET /version
Then the response status is 200
And the body contains "version"
Scenario: OpenAPI spec
When I GET /openapi.json
Then the response status is 200
And the body contains "openapi"
# ── MCP Registry v0.1 — read-only ─────────────────────────────────────────
Scenario: List servers in default registry
When I GET /registry/default/v0.1/servers
Then the response status is 200
And the body contains "servers"
Scenario: List servers with search filter
When I GET /registry/default/v0.1/servers?search=mysql
Then the response status is 200
And the body contains "servers"
Scenario: List servers with limit
When I GET /registry/default/v0.1/servers?limit=2
Then the response status is 200
And the count of servers is at most 2
Scenario: List servers filtered to latest version only
When I GET /registry/default/v0.1/servers?version=latest
Then the response status is 200
And the body contains "servers"
Scenario: List versions for a known server
When I GET /registry/default/v0.1/servers/io.github.stacklok%2Fadb-mysql-mcp-server/versions
Then the response status is 200
And the body contains "servers"
Scenario: Get a specific server version (latest)
When I GET /registry/default/v0.1/servers/io.github.stacklok%2Fadb-mysql-mcp-server/versions/latest
Then the response status is 200
And the body contains "io.github.stacklok/adb-mysql-mcp-server"
Scenario: Unknown registry returns 404
When I GET /registry/nonexistent-registry/v0.1/servers
Then the response status is 404
Scenario: Unknown server version returns 404
When I GET /registry/default/v0.1/servers/com.example%2Fdoes-not-exist/versions/1.0.0
Then the response status is 404
# ── Admin API v1 — registries ─────────────────────────────────────────────
Scenario: List registries
When I GET /v1/registries
Then the response status is 200
And the body contains "registries"
Scenario: Get the default registry by name
When I GET /v1/registries/default
Then the response status is 200
And the body contains "default"
Scenario: Get a nonexistent registry returns 404
When I GET /v1/registries/does-not-exist
Then the response status is 404
Scenario: Create a new registry via PUT
When I PUT /v1/registries/test-registry with an empty source list
Then the response status is 201
And the body contains "test-registry"
Scenario: Update an existing registry via PUT
When I PUT /v1/registries/test-registry again with a description
Then the response status is 200
Scenario: Delete an API-created registry
When I DELETE /v1/registries/test-registry
Then the response status is 204
Scenario: Delete a nonexistent registry returns 404
When I DELETE /v1/registries/does-not-exist
Then the response status is 404
# ── Admin API v1 — sources ────────────────────────────────────────────────
Scenario: List sources
When I GET /v1/sources
Then the response status is 200
And the body contains "sources"
Scenario: Get a known source by name
When I GET /v1/sources/local-file
Then the response status is 200
And the body contains "local-file"
Scenario: Get a nonexistent source returns 404
When I GET /v1/sources/does-not-exist
Then the response status is 404
Scenario: Create a managed source
When I PUT /v1/sources/managed-test with body {"managed":{}}
Then the response status is 201
And the body contains "managed-test"
Scenario: Delete the managed source
When I DELETE /v1/sources/managed-test
Then the response status is 204
# ── Entry lifecycle — publish and delete ──────────────────────────────────
Scenario: Publish a server version
Given a managed source exists
When I POST /v1/entries with a server payload
Then the response status is 201
And the body contains the server name
Scenario: Publish the same server version again returns 409
When I POST /v1/entries with the same server payload
Then the response status is 409
Scenario: Publish a skill version
When I POST /v1/entries with a skill payload
Then the response status is 201
And the body contains the skill name
Scenario: Delete the published server version
When I DELETE /v1/entries/server/com.example%2Ftest-server/versions/1.0.0
Then the response status is 204
Scenario: Delete a nonexistent entry returns 404
When I DELETE /v1/entries/server/com.example%2Fnope/versions/9.9.9
Then the response status is 404
Scenario: Publish with both server and skill in body returns 400
When I POST /v1/entries with both server and skill fields set
Then the response status is 400
Scenario: Publish with neither server nor skill in body returns 400
When I POST /v1/entries with an empty payload
Then the response status is 400
# ── Claims update on edit ──────────────────────────────────────────────────
Scenario: Registry claims are updated on PUT
When I PUT /v1/registries/claims-test with claims {"org":"acme"}
Then the response status is 201
And the body contains "acme"
When I PUT /v1/registries/claims-test with claims {"org":"contoso"}
Then the response status is 200
And the body contains "contoso"
When I GET /v1/registries/claims-test
Then the body contains "contoso"
And the body does not contain "acme"
Scenario: Source claims are updated on PUT
When I PUT /v1/sources/claims-test with file-data and claims {"org":"acme"}
Then the response status is 201
When I GET /v1/sources/claims-test
Then the body contains "acme"
When I PUT /v1/sources/claims-test with file-data and claims {"org":"contoso"}
Then the response status is 200
When I GET /v1/sources/claims-test
Then the body contains "contoso"
And the body does not contain "acme"
Scenario: Entry claims are updated via PUT
When I PUT /v1/entries/skill/test-skill/claims with {"claims":{"team":"eng"}}
Then the response status is 204
# ── Managed source limit ──────────────────────────────────────────────────
Scenario: Second managed source is rejected with 409
Given the managed-test source still exists
When I PUT /v1/sources/second-managed with body {"managed":{}}
Then the response status is 409
And the body contains "at most one managed source is allowed"
# ── List completeness ─────────────────────────────────────────────────────
Scenario: All created sources appear in list
Given three file-data sources are created: smoke-src-a, smoke-src-b, smoke-src-c
When I GET /v1/sources
Then the response status is 200
And the body contains "smoke-src-a"
And the body contains "smoke-src-b"
And the body contains "smoke-src-c"
Scenario: All created registries appear in list
Given three registries are created referencing the file-data sources
When I GET /v1/registries
Then the response status is 200
And the body contains "smoke-reg-a"
And the body contains "smoke-reg-b"
And the body contains "smoke-reg-c"
# ── OAuth / auth enforcement ───────────────────────────────────────────────
# The stack is restarted with auth.mode: oauth before these scenarios run.
# No real OIDC provider is needed: "missing token" and "malformed token"
# cases are rejected before JWKS is consulted.
Scenario: Public paths are accessible without a token in OAuth mode
Given the server is restarted in OAuth mode
When I GET /health
Then the response status is 200
When I GET /readiness
Then the response status is 200
When I GET /version
Then the response status is 200
When I GET /openapi.json
Then the response status is 200
Scenario: OAuth protected-resource metadata is publicly accessible
When I GET /.well-known/oauth-protected-resource
Then the response status is 200
And the body contains "authorization_servers"
Scenario: MCP list-servers requires a token in OAuth mode
When I GET /registry/default/v0.1/servers without an Authorization header
Then the response status is 401
And the response includes a WWW-Authenticate header with Bearer scheme
And the WWW-Authenticate header contains resource_metadata
Scenario: Admin registries endpoint requires a token
When I GET /v1/registries without an Authorization header
Then the response status is 401
Scenario: Admin sources endpoint requires a token
When I GET /v1/sources without an Authorization header
Then the response status is 401
Scenario: Malformed Bearer token returns 401 invalid_token
When I GET /registry/default/v0.1/servers with Authorization: Bearer not-a-real-jwt
Then the response status is 401
And the WWW-Authenticate header contains error="invalid_token"
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 753 lines · 40 tokens per session scan B bd60fc8cb84c
smoke-test is a skill published in the GitHub repository stacklok/toolhive-registry-server (24 stars, last pushed yesterday), licensed Apache-2.0. It adds 40 tokens to every session and 8,653 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 2 findings (sends data to an external url, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
code-review-with-lsp
Code review with LSP-powered code intelligence. Uses MCP tools (diagnostics, hover, references, definition, symbols) for semantic code understanding, not just text grep.
i18n-check
国际化完整性检查。检查翻译 key 是否缺失、硬编码文本、locale 文件一致性。.
vue-best-practices
Vue 2/3 代码规范检查。包括组件命名、Props 校验、Composition API 规范等。.
rust-review
Rust 服务审查：panic、SQL 注入、密钥、错误吞没、遗留标记.
python-review
Python 遗留代码审查：bare except、SQL 注入、反序列化、密钥、调试输出.
ts-review
TypeScript monorepo 审查：XSS、SQL 注入、密钥、any、console.log.