Knowledge Stack Cookbook

Focus on agents. We handle document intelligence.

32 production-style flagship agents + 100+ recipes for enterprise RAG — built on MCP, works with every major agent framework.

Quickstart · Flagships · Recipes · Wiki · Discord

---

Knowledge Stack is the document intelligence layer behind your agents — ingestion, chunking, permissions, versioning, citations — exposed as a stable MCP surface that plugs into LangChain, LangGraph, CrewAI, Temporal, OpenAI Agents SDK, pydantic-ai, Claude Desktop, Cursor, and anything else that speaks MCP. Every flagship runs under every framework — same prompt + schema, framework swaps in.

You focus on agent logic. KS manages the knowledge layer.

→ New here? Jump to the Quickstart below, or browse the flagships by vertical.

Index

• 🚀 Quickstart — git clone → first recipe in ~5 min
• 🧭 Flagships by vertical — 32 production-style agents, grouped by industry
• 📚 Recipes index — 100+ ≤100-LOC single-file patterns
• 🔌 Framework integrations: LangChain · LangGraph · CrewAI · Temporal · OpenAI Agents SDK · pydantic-ai · Claude Desktop · Cursor
• 🛠 MCP tools used — the 10 read-only tools every flagship calls
• ✍️ Contributing · Industries roadmap · Wiki

Quickstart

git clone → first cited recipe in ~5 minutes.

Prereqs: Python 3.11+, uv, a Knowledge Stack API key (sign in at https://app.knowledgestack.ai), and an OpenAI key (gpt-4o; gpt-4o-mini skips grounding and emits empty citations).

git clone https://github.com/knowledgestack/ks-cookbook.git
cd ks-cookbook
cp .env.example .env       # then fill in KS_API_KEY + OPENAI_API_KEY
make setup                 # installs every workspace package + validates env

Run your first recipe:

uv run python recipes/icd10_coder/recipe.py \
    --note-file recipes/icd10_coder/sample_inputs/deid_visit_001.txt

The agent makes ~10–20 MCP tool calls (search_knowledge, read, …), then emits a JSON CodingResult with real chunk_ids pointing into your tenant.

Other quick wins:

uv run python recipes/clause_extractor/recipe.py --contract "Apple 2024 proxy"
uv run python recipes/contract_renewal_checker/recipe.py --contract "Donna Huang software development"
uv run python recipes/aml_sar_narrative/recipe.py --case-id "structuring-cash-deposits"
make help                  # list every demo target

Each recipe folder has its own README.md with a captured live output and troubleshooting. To adapt a flagship to your own tenant, override CORPUS_FOLDER_ID:

CORPUS_FOLDER_ID=your-folder-id make demo-credit-memo

Flagships by vertical

32 flagships. Each links to its own README with the expected corpus, a sample input, and a captured sample output — open one to see exactly what it does before running.

Jump to: Banking · Legal · Accounting & tax · Healthcare · Insurance · Real estate · Sales & revenue · HR · Engineering · Government, pharma, energy

Banking & financial services

• Credit memo drafter — Cited credit memo from your bank's policy + a borrower's financials.
  Tags: banking credit-risk underwriting commercial-lending
• Loan covenant monitor — Flag covenant breaches from quarterly financials.
  Tags: banking covenant-monitoring credit-risk
• KYC onboarding review — CDD checklist + risk tier against your KYC policy.
  Tags: banking kyc aml compliance
• Earnings risk analyzer — Hebbia-style 10-K risk-flag memo with chunk citations.
  Tags: finance sec-filings 10-k investment-research

Legal

• Contract obligation extractor — Every shall/must/will obligation from a contract, categorized + cited.
  Tags: legal contracts msa obligations
• MSA redline vs. playbook — Compare an inbound MSA clause-by-clause against your playbook.
  Tags: legal contracts redline negotiations
• Privacy impact assessment — PIA memo citing GDPR Article 35 + company template.
  Tags: legal privacy gdpr security

Accounting & tax

• Rev-rec memo (ASC 606) — Five-step revenue-recognition memo grounded in policy.
  Tags: accounting asc-606 revenue-recognition memos
• Audit workpaper drafter — Tie a GL balance to source docs with PCAOB AS 1215 cites.
  Tags: accounting audit pcaob workpapers
• Tax position memo — Tax memo citing IRC sections + Treasury Regs.
  Tags: tax irc research memos

Healthcare

• Prior-authorization letter — Cited prior-auth or appeal letter grounded in payer medical policy.
  Tags: healthcare prior-auth payer clinical
• Clinical trial eligibility — Match a patient against I/E criteria from a real protocol.
  Tags: healthcare clinical-trials eligibility ctms

Insurance

• Claim adjudication memo — Coverage-analysis memo for a P&C claim.
  Tags: insurance claims coverage-analysis p-and-c
• Subrogation opportunity review — Recovery potential, citing NAIC Model 902.
  Tags: insurance subrogation claims
• Insurance policy comparison — Side-by-side analysis with explicit coverage gaps.
  Tags: insurance policy-comparison coverage

Real estate

• Lease abstract — One-page cited abstract (term, rent, renewals, CAM, exclusives).
  Tags: real-estate leases commercial
• Zoning compliance check — Proposed use vs. local Land Development Code.
  Tags: real-estate zoning compliance municipal

Sales & revenue

• CSV enrichment — Enrich every CSV row with a short summary from your KB.
  Tags: sales data-enrichment batch operations
• Research brief — Cited .docx research brief from your tenant.
  Tags: research reports analyst
• RFP first draft — RFP responses grounded in past proposals + capability docs.
  Tags: sales rfp proposals go-to-market
• Sales battlecard — Differentiators, objection handlers, win themes.
  Tags: sales competitive enablement
• Compliance questionnaire filler — Auto-complete CAIQ / SIG from your policy docs.
  Tags: security compliance caiq sig questionnaires

HR

• Employee handbook Q&A — Cited answers from the company handbook.
  Tags: hr handbook q-and-a
• Job description generator — JD grounded in leveling + comp bands.
  Tags: hr recruiting job-descriptions

Engineering, product & SRE

• Incident runbook lookup — PagerDuty alert → runbook + cited remediation.
  Tags: engineering sre runbooks incident-response
• API doc generator — Endpoint → developer docs from OpenAPI + style guide.
  Tags: engineering api documentation devex
• Release notes generator — Customer-facing notes from specs + migration guide.
  Tags: product engineering release-notes
• SOW scope validator — SOW completeness vs. template + methodology.
  Tags: proserv sow scope-management

Government, pharma & energy

• Grant compliance checker — Sub-awardee activity vs. NOFO + 2 CFR 200.
  Tags: government grants compliance cfr
• FOIA response drafter — FOIA letter with exemption analysis.
  Tags: government foia public-records
• Adverse event narrative — CIOMS-style AE narrative from drug label + PV SOP.
  Tags: pharma pharmacovigilance cioms safety
• NERC CIP evidence pack — Compliance evidence memo for a NERC CIP requirement.
  Tags: energy nerc-cip compliance utilities

Browse by tag

Tags are first-class metadata declared in each flagship's pyproject.toml [project] keywords. To find every flagship that uses a given tag, search the repo:

grep -l '"banking"' flagships/*/pyproject.toml

To add or change tags on a flagship: edit keywords = [...] in its pyproject.toml, then run make tags.

See INDUSTRIES.md for the broader roadmap and proposed next flagships.

Recipes

Short (≤100 LOC) single-file patterns across LangGraph, raw OpenAI, raw Anthropic, and MCP-only — see recipes/INDEX.md.

MCP tools used

Every flagship and recipe in this repo is read-only against your tenant. The 10 MCP tools the agents are allowed to call:

search_knowledge · search_keyword · read · read_around · list_contents · find · get_info · view_chunk_image · get_organization_info · get_current_datetime

That's the full contract. Writes are intentionally not exposed.

Community

• 💬 Discord — fastest place to get implementation and architecture help.
• 🗣️ GitHub Discussions — long-form questions, propose flagships.
• 🐛 Issues — bugs, flagship requests, recipe requests, framework integrations.
• 🔒 Security: see SECURITY.md. Do not open public issues for vulnerabilities.

Star history

If this repo helps you ship, star it — it's the single biggest signal we use to decide which flagships, frameworks, and verticals to prioritize next.

License

MIT. See LICENSE.