# Inspect live resources and debug a failure

> Inspect live resources and debug a failure: 2 companies fully do it, 6 with a limit, 12 carry the tag. Every listing in Cloud & infrastructure with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/infrastructure.inspect-debug/
- Markdown: https://www.anchorterminal.com/niches/infrastructure.inspect-debug/index.md (~1,150 tokens)
- Slim: https://www.anchorterminal.com/niches/infrastructure.inspect-debug/index.min.md (~330 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/infrastructure.inspect-debug/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

Two or three companies' docs show a full fit.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 2 | 6 | 12 | 26 | 9 |

## What counts as done

The resources and state the question needs, read with read-only calls, in structured output with paging, with no secrets in the response unless asked for, and a written finding.

How agents ask for it: why did the deploy fail; what is running in this account; check the status of this service; show the routing and security groups

## Does the job

The docs show it meets the test above, and nothing we read says otherwise.

- **AWS MCP Servers (Amazon Web Services, Inc.)**: call_aws runs AWS CLI reads with JSON output, READ_OPERATIONS_ONLY=true checks each command against known read operations, and IAM scopes what is reachable; the flag is off by default.
- **Azure MCP Server (Microsoft Corporation)**: --read-only removes write operations and --namespace limits the tools; resource graph, Monitor and per-service list and get tools read resources under the caller's RBAC.

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Cloudflare CLI (cf) (Cloudflare, Inc.)**: Results are JSON on stdout and errors on stderr, but lists return one page with paging options varying by command and no paging metadata, and there is no read-only mode beyond the token's permissions.
- **Cloudflare MCP Servers (Cloudflare, Inc.)**: search and execute return valid JSON capped at about 6,000 tokens with --- TRUNCATED --- markers, under a token or minimal OAuth scopes; but there's no read-only switch, and domain-server list tools return [object Object] (#468, open).
- **DigitalOcean MCP Server (DigitalOcean, LLC)**: Per-service endpoints and --services keep the tool list small, but there is no read-only mode (#273, open), and db-cluster-list and db-cluster-get return database passwords in plaintext (#372, open).
- **Netlify API + MCP (Netlify, Inc.)**: Lists are JSON with page, per_page and Link-header paging, but tokens have no scopes or read-only form, so a read-only session reaches every team its owner belongs to.
- **Railway MCP Server (Railway Corporation)**: The local server's list_, environment_status, get_logs and service_metrics tools read state, but there is no read-only grant, and on the hosted server logs go through the natural-language railway-agent.
- **Render MCP Server (Render Services, Inc.)**: Logs, metrics, deploy history and read-only SQL are available, but there is no read-only mode (#5, open), an API key reaches every workspace, and the docs say sensitive data may still be exposed.
- **Terraform MCP Server (HashiCorp, Inc. (an IBM Company))**: list_workspaces, get_run_details, list_state_versions and get_token_permissions read HCP Terraform state, and --toolsets narrows tools; but this covers only Terraform-managed state, and some writes run without ENABLE_TF_OPERATIONS.

- not this job: neckbeard (David Beale (individual, GitHub account bealesh)): No reads of live cloud resources.

## What stops agents finishing

- no read-only mode, so a read-only session needs a separately scoped key
- list output loses paging metadata or arrives as unparsed text
- tool list or one list call fills the context

## Around it

Before: observability.investigate-alert. After: infrastructure.change-live-resources, code.open-pr.

## Where this comes from

Weight 26 of 100 in Cloud & infrastructure: 9 of 32 accounts; the people who report it working ran the AWS CLI under a read-only credential rather than an MCP listing

As data: https://www.anchorterminal.com/api/v1/use-cases/infrastructure.inspect-debug.json, or the MCP tool get_use_case with id infrastructure.inspect-debug. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
