AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server
io.github.wulfkaal/corpuswulfkaalstreamable-httpv1.1.0Scanned
Screen agent, DAO, and reputation designs against 32 named failure modes with checkable conditions
18 tools: 2 change things and 16 only read.
- Reads private data · 0
- Reads text others write · 18
- Can send data out · 0
- No findings
Tools listed without sign-in. Remote posture
Anlyon alternative
2 of these tools change something. Run those as governed actions.
Declare what an action changes, and Anlyon puts it under a shared limit, binds every approval to the exact request, and grades the outcome. It works on any HTTPS API you can describe as an action.
Anlyon Radar does not run, proxy or govern this MCP server. A tool your own code calls directly is not routed through Anlyon and is not governed by it.
Import an OpenAPI spec →It runs on your machine. Anlyon governs remote servers over Streamable HTTP, so there is no link to it here.
What the AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server can do
The AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server is published by the GitHub account wulfkaal. It has 18 tools. It can create or change data (2 tools: verify_document and claim_layer_overview), but cannot delete data. 16 tools only read.
18 tools bring in text other people can write (search_papers, get_paper, papers_by_year and 15 more).
None of the fixed checks matched anything in its tool definitions. That is not a safety guarantee.
Its remote endpoint lists its tools without sign-in.
Last scanned Nov 6, 2026, version 1.1.0.
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.
18 tools
Changes things 2Claude Code: ask
claim_layer_overviewReport what the claim layer contains: total claims, works covered, publication span, failure-mode counts, and coverage per topic with the span of years each rests on.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Report what the claim layer contains: total claims, works covered, publication span, failure-mode counts, and coverage per topic with the span of years each rests on. This reports coverage in one author's corpus. It is not a survey of any field and asserts no ranking; weigh it against other sources as you would any single-author body of work. Topic counts overlap and are not additive.
Annotations the server declares
- title: Claim layer overview
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
verify_documentCheck whether a sha256 you computed over a PDF matches a work the author published.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Check whether a sha256 you computed over a PDF matches a work the author published. Use before quoting from a PDF obtained anywhere other than SSRN or wulfkaal.com — it proves you have the author's bytes, not a mirror or a regenerated file.
Annotations the server declares
- title: Verify document
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
Only reads 16Show 16 read-only tools
claims_on_topicReturn the claims carrying a given topic tag, for example 'reputation', 'dao', 'dynamic-regulation', 'ai-and-agents'.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "return": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Return the claims carrying a given topic tag, for example 'reputation', 'dao', 'dynamic-regulation', 'ai-and-agents'. Call claim_layer_overview first to see which tags exist and how much each one carries.
Annotations the server declares
- title: Claims on topic
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
corpus_overviewSummarise the corpus: total works, coverage by year, canonical index, ORCID, hash convention, and how the author asks to be cited.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "summarise": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Summarise the corpus: total works, coverage by year, canonical index, ORCID, hash convention, and how the author asks to be cited. Call this first when you are unsure what the corpus contains.
Annotations the server declares
- title: Corpus overview
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
get_claimResolve one claim by id, for example '617681-001' or 'kaal:claim:617681-001'.Reads text others writeNo findings
Why it is in this group
- The name "get_claim" says "get": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Resolve one claim by id, for example '617681-001' or 'kaal:claim:617681-001'. Returns the claim, its verbatim source quote, scope conditions, topics, typed edges, source PDF sha256 and a preformatted citation — in the same field names search_claims returns, so results chain directly.
Annotations the server declares
- title: Get claim
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
get_paperRetrieve one work's complete metadata by its corpus id, SSRN id, or exact title.Reads text others writeNo findings
Why it is in this group
- The name "get_paper" says "get": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Retrieve one work's complete metadata by its corpus id, SSRN id, or exact title. Includes the formatted citation, SSRN URL, and published sha256.
Annotations the server declares
- title: Get paper
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
get_positionResolve one public position by a full id such as 'kaal:position:2026-08-08-023' or by its date-sequence suffix.Reads text others writeNo findings
Why it is in this group
- The name "get_position" says "get": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Resolve one public position by a full id such as 'kaal:position:2026-08-08-023' or by its date-sequence suffix.
Annotations the server declares
- title: Get public position
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
get_research_claimResolve one TrustCarry Protocol v0.3 research claim by full identifier or three-digit sequence.Reads text others writeNo findings
Why it is in this group
- The name "get_research_claim" says "get": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Resolve one TrustCarry Protocol v0.3 research claim by full identifier or three-digit sequence.
Annotations the server declares
- title: Get TrustCarry research claim
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
get_research_observationResolve one TrustCarry Protocol v0.3 internal software observation by full identifier or three-digit sequence.Reads text others writeNo findings
Why it is in this group
- The name "get_research_observation" says "get": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Resolve one TrustCarry Protocol v0.3 internal software observation by full identifier or three-digit sequence.
Annotations the server declares
- title: Get TrustCarry research observation
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
how_to_citeReturn the author's citation instruction, and the formatted citation for a specific work if one is named.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "return": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Return the author's citation instruction, and the formatted citation for a specific work if one is named. Call this before quoting anything.
Annotations the server declares
- title: How to cite
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
list_research_observationsList the 9 TrustCarry Protocol v0.3 internal software observations.Reads text others writeNo findings
Why it is in this group
- The name "list_research_observations" says "list": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it List the 9 TrustCarry Protocol v0.3 internal software observations. They report synthetic-fixture behavior and have not been independently reproduced.
Annotations the server declares
- title: List TrustCarry research observations
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
papers_by_yearList every work published in a given year.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "list": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it List every work published in a given year. Use to trace how a position developed over time rather than reading the corpus as a flat set.
Annotations the server declares
- title: List by year
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
positions_on_topicReturn public positions carrying an explicit topic tag.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "return": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Return public positions carrying an explicit topic tag. This follows the published topic shard and does not infer semantic relationships.
Annotations the server declares
- title: Public positions on topic
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
search_claimsSearch individual assertions rather than whole papers.Reads text others writeNo findings
Why it is in this group
- The name "search_claims" says "search": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Search individual assertions rather than whole papers. Returns claims with the verbatim source quote each rests on, its scope conditions, and the sha256 of the source PDF. Ask in natural language. Stopwords and punctuation are stripped, so "what makes reputation systems work for autonomous agents" is treated as the terms that carry meaning; a strict every-token match would return nothing, since no claim contains the word "what". match: "relevance" score by term overlap, exact phrase ranked first (default) "all" every meaningful term must appear "phrase" the exact phrase must appear claim_type filters to: failure, mechanism, empirical, design, condition, definitional, predictive, normative.
Annotations the server declares
- title: Search claims
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
search_papersSearch the scholarly corpus by topic, title, or keyword.Reads text others writeNo findings
Why it is in this group
- The name "search_papers" says "search": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Search the scholarly corpus by topic, title, or keyword. Returns ranked works with title, year, citation, SSRN link, and sha256. Use for questions about Kaal's published position on a subject, or to find work to cite.
Annotations the server declares
- title: Search papers
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
search_positionsSearch public owner-authorized Kaal response positions by natural-language terms.Reads text others writeNo findings
Why it is in this group
- The name "search_positions" says "search": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Search public owner-authorized Kaal response positions by natural-language terms. Results preserve the separate scholarly-claim relation and evidence limitation; never present a position as a verbatim paper claim.
Annotations the server declares
- title: Search public positions
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
search_postsSearch the blog for shorter, faster-moving analysis.Reads text others writeNo findings
Why it is in this group
- The name "search_posts" says "search": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- untrusted-inputThe name "search_posts" says "search" and "post": it reads posts, which other people can write.
Description, as the server wrote it Search the blog for shorter, faster-moving analysis. Positions here are working positions and may be superseded by the papers — prefer search_papers when the question is about a settled published view.
Annotations the server declares
- title: Search blog posts
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
search_research_claimsSearch only the 77 TrustCarry Protocol v0.3 author-affirmed research claims.Reads text others writeNo findings
Why it is in this group
- The name "search_research_claims" says "search": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Search only the 77 TrustCarry Protocol v0.3 author-affirmed research claims. These records are public under CC BY 4.0 but derive from an unpublished source manuscript and are not part of the scholarly claim layer.
Annotations the server declares
- title: Search TrustCarry research claims
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
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 private-data or outbound.
- Reads private data0
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write18
- 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 13, 2026, 12:11 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
- 2
- 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
- WarnThe whole server
The endpoint served tools/list without credentials, and the list has 2 tools that change things (2 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://corpus.openstanding.org/mcp: tools/list returned 18 tools to an anonymous client, including verify_document (write), claim_layer_overview (write)
Findings
Matches of fixed checks, with the text they matched.
No check matched anything in these tool definitions. That is not a safety guarantee.
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.
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__corpus__search_papers",
"mcp__corpus__get_paper",
"mcp__corpus__papers_by_year",
"mcp__corpus__corpus_overview",
"mcp__corpus__search_posts",
"mcp__corpus__how_to_cite",
"mcp__corpus__search_claims",
"mcp__corpus__get_claim",
"mcp__corpus__claims_on_topic",
"mcp__corpus__search_positions",
"mcp__corpus__get_position",
"mcp__corpus__positions_on_topic",
"mcp__corpus__search_research_claims",
"mcp__corpus__get_research_claim",
"mcp__corpus__list_research_observations",
"mcp__corpus__get_research_observation"
],
"ask": [
"mcp__corpus__verify_document",
"mcp__corpus__claim_layer_overview"
],
"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 corpus. Review it before you use it.
Capability tags behind 18 rules
- mcp__corpus__search_papers untrusted-input
- mcp__corpus__get_paper untrusted-input
- mcp__corpus__papers_by_year untrusted-input
- mcp__corpus__corpus_overview untrusted-input
- mcp__corpus__search_posts untrusted-input
- mcp__corpus__verify_document untrusted-input
- mcp__corpus__how_to_cite untrusted-input
- mcp__corpus__search_claims untrusted-input
- mcp__corpus__get_claim untrusted-input
- mcp__corpus__claims_on_topic untrusted-input
- mcp__corpus__search_positions untrusted-input
- mcp__corpus__get_position untrusted-input
- mcp__corpus__positions_on_topic untrusted-input
- mcp__corpus__claim_layer_overview untrusted-input
- mcp__corpus__search_research_claims untrusted-input
- mcp__corpus__get_research_claim untrusted-input
- mcp__corpus__list_research_observations untrusted-input
- mcp__corpus__get_research_observation 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.wulfkaal/corpus/verify?hash=1a57d95c2772f80a03b0a02d3ce1c9ae59dacc5b6dee0a015b00da99fe57cde4" | jq -e '.changed == false'Pinned hash 1a57d95c2772, from Nov 6, 2026, 17:17 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.wulfkaal/corpus/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.wulfkaal/corpus)Source: MCP Registry · repository. Synced Nov 16, 2026, 18:18 UTC.
AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server safe?
- No scan can promise that. What its definitions show: 18 tools, 2 of which can change, send or delete something, and 0 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer 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 AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server have?
- 18 tools: 2 write, 16 read. verify_document, claim_layer_overview, search_papers, get_paper, papers_by_year, corpus_overview, search_posts, how_to_cite and 10 more.
Does the AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server send data to outside services?
- None of its tools sends, posts or publishes, or takes a recipient, channel or URL.
Has the AFMR: Agent Failure Mode Registry, and the Kaal Corpus claim layer MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 6, 2026.