CollectSEA MCP server
io.github.kelvin-ling/collectseakelvin-lingstreamable-httpv2026.10.17Scanned
Find real software, security and AI-agent failures that resemble your design, with safeguards.
6 tools: 6 only read.
- Reads private data · 1
- Reads text others write · 1
- Can send data out · 0
- 1 finding
Tools listed without sign-in. Remote posture
With Anlyon
Decide which of this server's tools your agent can use.
- Anlyon lists the tools and adopts none.
- Each tool you adopt gets approvals and receipts.
- Your agent never holds the key.
A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.
What the CollectSEA MCP server can do
The CollectSEA MCP server is published by the GitHub account kelvin-ling. It has 6 tools. Every one of them only reads: going by their definitions, none of them creates, changes, sends or deletes anything.
1 tool reads private data (analyze_existing_app); 1 tool brings in text other people can write (analyze_existing_app).
The fixed checks matched 1 finding (credentials-mention). Each one comes with the text it matched.
Its remote endpoint lists its tools without sign-in.
Last scanned Nov 20, 2026, version 2026.10.17.
Tools
Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.
6 tools
Only reads 6Show 6 read-only tools
analyze_designProfile a proposed feature, system or AI-agent workflow and return documented incidents from systems that behaved similarly, their failure patterns, and the safeguards to build in.No findings
Why it is in this group
- The name "analyze_design" says "analyze": a word for reading.
Description, as the server wrote it Profile a proposed feature, system or AI-agent workflow and return documented incidents from systems that behaved similarly, their failure patterns, and the safeguards to build in. Use before writing code for anything that moves money, deletes data, changes permissions, runs code, or acts on external content.
Annotations the server declares
- title: Find failure precedents for a design
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
analyze_existing_appAnalyze an existing app using observed behavior, README, architecture notes, configuration or source snippets.Reads private dataReads text others write1 finding
Why it is in this group
- The name "analyze_existing_app" says "analyze": a word for reading.
What it can reach
- private-dataThe description's first sentence says "analyze" and "note": it reads notes.
- private-dataThe input names a repository (input.repository_url), and the tool reads: it reads repositories.
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The input schema at properties.files.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.files.description: "...ers and the 32 KB request limit. Remove credentials and personal data."
Description, as the server wrote it Analyze an existing app using observed behavior, README, architecture notes, configuration or source snippets. Provide a public GitHub repository_url to import selected context, or inspect a private repository locally and submit relevant text. Code is never executed and the running app is not scanned. Returns source context, behavior signals, historical precedents and safeguards to verify.
Annotations the server declares
- title: Find failure precedents for an existing app
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
get_incidentFull record for one incident by ID (e.g.No findings
Why it is in this group
- The name "get_incident" says "get": a word for reading.
Description, as the server wrote it Full record for one incident by ID (e.g. SEA-POST-005): what happened, source, failure pattern, safeguards and review status.
Annotations the server declares
- title: Get one incident record
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_safeguardsReturn ready-to-save rules (AGENTS.md by default; also CLAUDE.md, Cursor rule or Copilot instructions) for a proposed design, each rule citing the incidents behind it.No findings
Why it is in this group
- The name "get_safeguards" says "get": a word for reading.
Description, as the server wrote it Return ready-to-save rules (AGENTS.md by default; also CLAUDE.md, Cursor rule or Copilot instructions) for a proposed design, each rule citing the incidents behind it.
Annotations the server declares
- title: Get agent rules for a design
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
list_failure_patternsThe 14 recurring failure patterns with triggers, warning signs and engineering principles.No findings
Why it is in this group
- The name "list_failure_patterns" says "list": a word for reading.
Description, as the server wrote it The 14 recurring failure patterns with triggers, warning signs and engineering principles.
Annotations the server declares
- title: List failure patterns
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
search_incidentsKeyword search over documented incidents and security advisories (company, technology, CVE, failure mode).No findings
Why it is in this group
- The name "search_incidents" says "search": a word for reading.
Description, as the server wrote it Keyword search over documented incidents and security advisories (company, technology, CVE, failure mode). Returns titles, summaries, evidence labels and URLs.
Annotations the server declares
- title: Search the incident database
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Exfil surface
Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.
No private-data, untrusted-input, outbound path in this server alone: no tool is tagged outbound.
- Reads private data1
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write1
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out0
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.
Remote posture
What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.
Looked at Nov 20, 2026, 19:07 UTC, unchanged since Oct 22, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 200
- tools/list served without credentials
- yes
- Tools that change things, listed without credentials
- 0
- Protocol version
- 2025-06-18
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- not observed
- Authorization servers
- not observed
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- not observed
- Client ID metadata documents
- not observed
Findings
Matches of fixed checks, with the text they matched.
The input schema at properties.files.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.files.description: "...ers and the 32 KB request limit. Remove credentials and personal data."
History
Changes to its tool definitions and to its remote endpoint, newest first.
No change seen yet. Its tools are rescanned every day, and a remote endpoint is looked at once a week.
Install
Pinned to version 2026.10.17. Values in angle brackets are yours to fill in.
claude mcp add --transport http collectsea https://collectsea.com/mcpclaude mcp add --transport http collectsea https://collectsea.com/mcpRun in a terminal.
Claude Code permissions
Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [
"mcp__collectsea__analyze_design",
"mcp__collectsea__analyze_existing_app",
"mcp__collectsea__search_incidents",
"mcp__collectsea__get_incident",
"mcp__collectsea__list_failure_patterns",
"mcp__collectsea__get_safeguards"
],
"ask": [],
"deny": []
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as collectsea. Review it before you use it.
Capability tags behind 1 rule
- mcp__collectsea__analyze_existing_app private-data, untrusted-input
Pin
Keep the decision: fail CI when the tools change, follow every change, or show a badge.
Pin in CI
Fail a build when the tools you reviewed are not the tools the server serves today.
curl -s "https://radar.anlyon.com/api/v1/servers/io.github.kelvin-ling/collectsea/verify?hash=dc63a67b5d43cf9fd034f6f81497d7e41147c588be3299bd22363e6d994e7a28" | jq -e '.changed == false'Pinned hash dc63a67b5d43, from Nov 20, 2026, 00:17 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.kelvin-ling/collectsea/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.kelvin-ling/collectsea)Source: MCP Registry · repository. Synced Oct 21, 2026, 18:37 UTC.
CollectSEA MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the CollectSEA MCP server safe?
- No scan can promise that. What its definitions show: 6 tools, all of which only read, and 1 finding from fixed checks. Read-only tools can still pass private data to a model.
Can the CollectSEA MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the CollectSEA MCP server do things that cannot be undone?
- No tool is classed as irreversible: none of them sends, pays, publishes or merges, going by their definitions.
What tools does the CollectSEA MCP server have?
- 6 tools: 6 read. analyze_design, analyze_existing_app, search_incidents, get_incident, list_failure_patterns and get_safeguards.
Does the CollectSEA MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the CollectSEA MCP server send data to outside services?
- None of its tools sends, posts or publishes, or takes a recipient, channel or URL.
Has the CollectSEA MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 20, 2026.