agenda-intelligence-md 0.7.5

CLI, MCP server, and JSON schemas for validating and auditing strategic-risk AI agent output

Agenda Intelligence MD

Evidence & eval layer for strategic intelligence agents.

Protocol, JSON schemas, CLI, and MCP server for validating, scoring, and auditing the structure of strategic-risk agent output. The evidence-discipline surface for markdown-first reasoning skills (Global Think Tank Analyst, Central Asia + Caspian, Gulf + Middle East).

What this is

• Markdown protocol — structured reasoning workflow for agents (Agenda-Intelligence.md)
• JSON schemas — validate briefs, evidence packs, audits, signals, memory cards, lenses
• CLI — validate-brief, validate-evidence, source-categories, source-coverage, audit-claims, score, bench, doctor (30+ commands)
• MCP server — stdio server exposing validation, read, and scoring tools
• Eval kit — rubric, LLM-judge prompt, human checklist, benchmark harness
• Source policy — per-claim provenance tags (Axis A/B), source requirements for 12 categories

What this is not

• Not a factuality verifier — checks structure, not truth
• Not an autonomous news agent or source retriever
• Not a source reputation scorer or live news gatherer
• Not a replacement for analyst judgment
• Not a compliance, legal, or financial advisory product

Quickstart

pip install agenda-intelligence-md
# Or pinned wheel:
# pip install https://github.com/vassiliylakhonin/agenda-intelligence-md/releases/download/v0.7.5/agenda_intelligence_md-0.7.5-py3-none-any.whl

agenda-intelligence validate-brief examples/agenda-brief.json
agenda-intelligence score examples/agenda-brief.json --evidence examples/source/evidence-pack.json
agenda-intelligence bench examples/source-backed --strict --min-score 80
agenda-intelligence doctor
agenda-intelligence mcp-config --client cursor

Benchmark baseline

20 source-backed cases, reproduced with agenda-intelligence bench examples/source-backed/:

Metric	Value
Cases	20
Mean score	87.6 / 100
Min / max	84 / 91
Schema-valid	100%
With evidence pack	100%
With claim-level audit	100%
With source category	100%
Mean source coverage	14.8%
Source coverage gap cases	20
Orphan evidence refs	0

Heuristic scores are uncalibrated and not validated against expert judgment. They evaluate structure, evidence labeling, source-coverage diagnostics, and decision-readiness — not factual truth.

Flagship example: examples/source-backed/eu-ai-act.md — brief + evidence pack + claim-level audit using illustrative sources. Before / after pairs: examples/before-after/.

Verification Contract

verify-quotes checks whether a cited quote or excerpt appears in supplied local text, or in text fetched from an already-specified URL when --fetch is used. It does not discover sources, score source reputation, gather live news, or decide whether a claim is true in the world.

Schemas

Schema	Purpose
agenda-brief.schema.json	Brief structure
evidence-pack.schema.json	Evidence pack
evidence-audit.schema.json	Claim-level audit
signal-tracker.schema.json	Signal lifecycle
memory-card.schema.json	AnalysisBank cards
lens-manifest.schema.json	Lens manifest
signal-classification.schema.json	Signal taxonomy

MCP

Stdio MCP server with 11 tools. Full docs and wire-protocol verification: MCP.md. Client setup: docs/integrations/mcp.md.

Tool	What it does
validate_brief	Validate a brief dict against agenda-brief.schema.json
validate_evidence	Validate an evidence-pack dict against evidence-pack.schema.json
audit_claims	Check claim-level audit: support distribution, orphan refs, unsupported claims
score_output	Heuristic score for structure, evidence labeling, decision-readiness
get_protocol	Return the full Agenda-Intelligence.md reasoning protocol
list_source_categories	List source requirement categories before calling source_plan
source_plan	Generate a source plan for a given topic
source_coverage	Diagnose evidence-pack coverage against category source requirements
verify_quotes	Check cited quote fragments in caller-provided text
list_lenses	List available lens packs
get_lens	Return a specific lens pack by name

Status

Component	Status
Markdown protocol, JSON schemas	Stable
CLI (validate, score, bench, audit, doctor)	Stable
MCP stdio server	Stable
Evidence-audit schema (claim-level)	Stable
Signal-tracker schema (lifecycle)	Stable
Heuristic scoring	Stable (uncalibrated)
Live source retrieval	Not implemented
Factual-truth verification	Not in scope

Documentation

Resource	Link
Quickstart	docs/quickstart.md
Tutorial	docs/tutorial.md
Evaluation layers	docs/evaluation.md
Factual verification boundary	docs/factual-verification.md
Source plan coverage boundary	docs/source-plan-coverage.md
Evidence audit	docs/evidence-audit.md
Threat model	docs/threat-model.md
Integrations	docs/integrations/
Use-cases	docs/use-cases/
Agent contract	AGENTS.md
Adoption guide	ADOPTION.md
Changelog	CHANGELOG.md
Roadmap	ROADMAP.md

Repository layout

agenda-intelligence-md/
├─ src/agenda_intelligence/   # Python package (CLI + MCP server)
├─ schemas/                   # JSON schemas
├─ examples/                  # briefs, evidence packs, before/after
├─ skills/                    # OpenClaw skill wrappers
├─ evals/                     # rubric, judge prompt, benchmark
├─ analysis-bank/             # agent persistent memory (memory-card schema, see schemas/memory-card.schema.json)
├─ docs/                      # guides, integrations, use-cases
├─ scripts/                   # dev and CI helpers
└─ tests/                     # pytest suite

Contact

Vassiliy Lakhonin — Almaty, Kazakhstan (UTC+5)

Portfolio · For analysts · Email · LinkedIn · GitHub

Issues, PRs, and eval-case contributions are welcome.

License

MIT.

---

Disclaimer. This toolkit is for informational and educational purposes only. It does not constitute investment, financial, legal, compliance, or trading advice. It does not verify factual truth, predict outcomes, or replace professional judgment. Use at your own risk.