agenda-intelligence-mdBLOCK
Deterministic evidence-packet checks for AI-generated claims: references, quotes, numbers, and support gaps. CLI, Python, MCP, and review adapters.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
A deterministic evidence-packet linter for claim-backed AI output. Supply claims, source references, quotations, and source text; receive broken-reference, quote, number, lexical-support, and evidence-gap findings before human review.
Use it to make an AI answer's evidence trail inspectable in a local workflow, agent pipeline, or CI job. Packet completeness is not factual truth, source authenticity, or permission to act.
First run
Python 3.9 or later, from a checkout:
python3 -m venv .venv .venv/bin/python -m pip install -e . .venv/bin/agenda-intelligence check examples/evidence-packet/request.json .venv/bin/agenda-intelligence check examples/evidence-packet/request.json --format json
For the versioned package, use python -m pip install "agenda-intelligence-md==1.16.0" instead of the editable install. The example commands above use files from this checkout.
The bundled synthetic packet reports packet_status=packet_complete and factuality=not_assessed. A stale or inaccurate source can still pass. Add --strict when packet findings should fail a CI step.
For document-based review, start with the local-file review guide and example manifest:
.venv/bin/agenda-intelligence review examples/evidence-review/manifest.json --format html
For a RAG answer with inline citations and retrieved chunk texts, use the Output Verification adapter. It assembles the packet locally and returns line-level findings and repair guidance:
.venv/bin/agenda-intelligence review-answer examples/output-verification/rag-answer.json .venv/bin/agenda-intelligence review-answer examples/output-verification/rag-answer-revised.json --strict
The original fictional answer routes to revision; the corrected answer routes to human review. Both run witho
3d1213a85c98OBSERVED · 2026-10-10Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add agenda-intelligence-md -- uvx agenda-intelligence-md==1.16.0
Exposed tools (17)
17 read · 0 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
CHECK_TRANSACTION_SAFETY | read | Checks supplied transaction evidence and known risk patterns; does not authorize signing. |
agenda-guard | read | Evidence checks for proposed transactions and escrow deliveries; human review is required. |
audit_evidence_claims | read | Audit claims against cited evidence sources |
claims_text | read | Extracted claims to audit |
diagnostics_json | read | Audit or verification failure details |
discovery | read | Machine-readable discovery documents. |
draft_evidence_memo | read | Draft a structured evidence-backed analysis memo for review |
evidence_summary | read | Summary of available evidence |
gates | read | Canonical HTTP endpoints for the triage gates. |
jsonrpc | read | A2A-compatible JSON-RPC endpoint. |
knowledge | read | OKF and entity-map artifacts for retrieval agents. |
monetization | read | Autonomous Base USDC settlement and Bearer key provisioning. |
packet_json | read | Current evidence packet JSON |
self_correct_packet | read | Generate instructions to repair an evidence packet with gaps |
sources_text | read | Source text to verify against |
status | read | Health and status endpoints. |
topic | read | Subject matter of the memo |
Trust audit
BLOCKgrade F · trust 58/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | FAIL |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (6 observation(s))
- Network
- declared (7 observation(s))
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- found
Findings (25)
If the evidence packet is incomplete, ask for the missing source material. Do not send secrets or unrelated private data to the hosted endpoint. Do not invoke Decision Gate tools or treat receipts as
console.log(`ok ${label.padEnd(36)} secret written`);re.compile(r"exfiltrate\s+(keys|balance|private\s+key)", re.IGNORECASE),
console.log(`\nвсего ${rows.length} проверок, провалов ${failed.length}\n`);if paragraph and re.match(r"[A-ZА-ЯЁ0-9]", line):
const token = "portal-issued-token-123456";
.claudeignore
const run=new Function('document','window','crypto','fetch','agendaTelemetryHeaders',show+'; return showWorkedExample;');const paidFetch = new Function('window', 'fetch', PAYMENT_CLIENT_SCRIPT + '; return agendaPaidFetch;')(wallet, fetch);const headers = new Function('window', PAYMENT_CLIENT_SCRIPT + '; return agendaTelemetryHeaders;')(window);const paidFetch = new Function('window', 'fetch', PAYMENT_CLIENT_SCRIPT + '; return agendaPaidFetch;')(wallet, fetch);const pay = new Function("window", "document", "fetch", `${source}; return payWithBaseWallet;`)(window, { getElementById: () => statusDiv }, fetch);import { createHostedMcpCall } from '../../../examples/hosted-mcp/client.mjs';const repositoryRoot = resolve(scriptDirectory, "../../..");
const dir = new URL("../../../schemas/v1/", import.meta.url);? JSON.parse(fs.readFileSync(new URL('../../../examples/output-verification/request.json',import.meta.url),'utf8'))"/profiles/confidential-project-room/index.md": "---\ntype: evidence_readiness.profile_contract\ntitle: Confidential Project-Room Evidence-Readiness Profile\nstatus: build-to-learn\nversion: \"0.1\"\n
curl -sS http://127.0.0.1:8080/v1/audit-claims \
curl -sS http://127.0.0.1:8080/v1/source-coverage \
curl -sS http://127.0.0.1:8080/v1/score \
curl -sS http://127.0.0.1:8080/v1/middle-corridor/deal-risk \
curl -sS http://127.0.0.1:8080/v1/agentic-interaction/trust \
hex"23e9918700000000000000000000000000000000000000000000000000000000000000010000000000000000000000000000000000000000000000000000000000001337000000000000000000000000000000000000000000000000000000000000
hex"0000000000080000000000000000001000000000000000000000000000018000000000000000000000000000000000000000000000000800000000000000000000000000000000000000000000000000000000000000000000000000000000000000
const binary = atob(value.replace(/-/gu, "+").replace(/_/gu, "/") + padding);
Gates applied: critical_finding, no_behavioural_pass, undeclared_transfer.
3d1213a85c98full audit observations/trust-audit/mcp-server/vassiliylakhonin__agenda-intelligence-md.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-10 | 3d1213a85c98 | BLOCK | F | 58 | first audit |
Questions
What is the agenda-intelligence-md MCP server?
Deterministic evidence-packet checks for AI-generated claims: references, quotes, numbers, and support gaps. CLI, Python, MCP, and review adapters.
What tools does agenda-intelligence-md expose?
17 in total: 17 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is agenda-intelligence-md safe to connect to an agent?
No — not without reading the findings first. The audit graded it F (58/100) and found 1 critical or high issue in the source. Each one is listed on this page with the file and line it is on.
What credentials does agenda-intelligence-md need?
It reads AGENDA_INTELLIGENCE_KEY, AGENDA_INTELLIGENCE_MAX_TOKENS, AGENDA_OBSERVABILITY_TOKEN, AGENDA_STATS_TOKEN, ANTHROPIC_API_KEY, OPENSANCTIONS_API_KEY, STATS_TOKEN, VIZIER_API_KEY, VIZIER_DELEGATION_SIGNING_KEY and VIZIER_REVIEW_API_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does agenda-intelligence-md run?
It speaks stdio and streamable-http, so it runs as a local process your client starts. It is published on npm as @agenda-intelligence/guard-mobile at 1.2.4.
How current is this page?
The grade is for one exact copy of the source (3d1213a85c98), read on 2026-10-10. The repository is watched and re-audited when it changes.