@tangle-network/starter-foundry 0.7.3 → 0.13.2

package/registry/families/agent-runtime-real-estate-ts/files/methodology/cap-rate-worksheet.md

@@ -0,0 +1,180 @@
+---
+id: cap-rate-worksheet
+capability: cap-rate-worksheet
+retrieved: 2026-04-25
+authority: CCIM Institute "Investment Real Estate" curriculum; Frank Gallinelli "What Every Real Estate Investor Needs to Know About Cash Flow"; IRS Pub 527 (residential rental); IRS Pub 544 (asset disposition)
+---
+
+# Cap Rate / NOI / Cash-on-Cash Worksheet
+
+A cap rate is a **valuation-comparison ratio** for stabilized income
+property. It is not a return projection, not a hold-period IRR, and
+not a substitute for full underwriting. Use this worksheet to compare
+a candidate property to market or to underwrite a buy-and-hold
+acquisition. For syndication, partnership, or lending decisions,
+escalate to a CPA and the lender's underwriter.
+
+## 1. Inputs (open `:::question` until complete)
+
+Refuse to compute until ALL of these are known:
+
+- Asking / contract price
+- Unit count and rent per unit (current AND market — flag any gap)
+- Vacancy assumption (% — physical AND credit-loss combined)
+- Operating expense detail (line items, not a ratio)
+- Capex reserve (annual $ or per-unit)
+- Property type (residential 1-4 / residential 5+ / commercial / mixed)
+- Financing structure (cash / loan amount / rate / amortization /
+  term / IO period if any)
+- Closing costs and immediate capex (acquisition basis)
+
+Refuse to use a "50% rule" or other rules-of-thumb in lieu of real
+opex line items. Rules-of-thumb are screening tools, not underwriting.
+
+## 2. NOI build
+
+```
+Gross Potential Rent (GPR)        = sum(unit rent × 12)
++ Other income                     (laundry, parking, storage, pet)
+− Vacancy & credit loss            (% × GPR; minimum 5% even if
+                                     "fully leased")
+= Effective Gross Income (EGI)
+
+− Operating expenses               (see §3 line items)
+− Capex reserve                    (see §4)
+= Net Operating Income (NOI)
+```
+
+NOI **excludes** debt service, depreciation, income tax, and the
+owner's labor. Including any of those produces a different metric
+(cash flow before tax, after-tax cash flow, etc.) — keep them
+separate.
+
+## 3. Operating expenses (line by line)
+
+Refuse to estimate opex as a single percentage. Build it:
+
+- **Property tax** — pull from county assessor; flag if a sale will
+  trigger reassessment (e.g. CA Prop 13 succession, mansion-tax
+  jurisdictions). Reassessment risk is a real underwriting hazard;
+  surface it.
+- **Insurance** — current premium; flag wind / flood / wildfire
+  zones where premium has moved >20% YoY recently.
+- **Property management** — 6-10% of EGI for residential SFR; 4-8%
+  for multifamily 5+; lower for commercial NNN. If owner-managed,
+  carry a phantom PM fee — the labor isn't free.
+- **Repairs & maintenance** — 5-15% of EGI; calibrate to age and
+  condition tier.
+- **Utilities** — only what owner pays; verify lease structure.
+- **HOA / condo fees** — if any.
+- **Trash / sewer / water** — if owner-paid.
+- **Lawn / snow / pest** — if owner-paid.
+- **Legal / accounting / licensing** — small but real.
+- **Marketing / leasing commissions** — annualized.
+
+## 4. Capex reserve
+
+Capex is NOT a maintenance line. It is a sinking fund for big-ticket
+replacement: roof (20-30 yr), HVAC (15-20), water heater (8-12),
+flooring (10-15), siding (25-40), windows (25-40), kitchen (15-25),
+bath (15-25), parking lot resurface (15-25 commercial).
+
+Per-unit per-year defaults:
+- Residential SFR: $250-$500 / unit / yr
+- Residential 2-4: $200-$400 / unit / yr
+- Residential 5+: $250-$350 / unit / yr (institutional standard)
+- Commercial NNN: tenant typically pays — verify the lease
+
+Underreserving capex is the single most common mistake in retail-
+investor cap-rate math. Always carry a reserve. Surface it loudly
+when the user wants to omit it.
+
+## 5. Cap rate
+
+```
+Cap rate = NOI / Price
+```
+
+Compare to market cap rate for the asset class / submarket. Sources:
+CoStar, RCA, local broker market reports. Surface 3 cap rates side by
+side:
+
+- **In-place cap** — using current rents
+- **Market cap** — using market rents
+- **Stabilized cap** — using market rents, full vacancy assumption,
+  realistic opex including PM and capex reserve
+
+If in-place and market diverge >100 bps, there is rent upside or
+downside the user must underwrite separately. Don't paper over it.
+
+## 6. Cash-on-cash and DSCR
+
+```
+Annual debt service        = mortgage P&I × 12
+Cash flow before tax (CFBT) = NOI − annual debt service
+Cash invested              = down payment + closing costs + immediate capex
+
+Cash-on-cash               = CFBT / cash invested
+DSCR                       = NOI / annual debt service
+```
+
+DSCR floors:
+- Residential 1-4 conventional: lender doesn't underwrite DSCR;
+  surface it anyway as a stress test
+- Residential 5+ agency: typically 1.20-1.25 minimum
+- Commercial: 1.20-1.40 typical; lower in trophy assets, higher in
+  distressed markets
+
+DSCR < 1.0 means the property doesn't cover its own debt service from
+NOI — flag this as a serious underwriting risk regardless of the
+investor's outside income.
+
+## 7. Sensitivity table
+
+Build a 2D sensitivity (cap rate × exit cap, OR rent growth × vacancy)
+showing Year-5 cash flow and unlevered IRR. A point estimate without
+sensitivity misrepresents the bet. The user should see how fragile
+the deal is to:
+
+- ±50 bps cap rate at exit
+- ±100 bps interest rate at refinance
+- ±5% vacancy
+- ±10% opex inflation
+
+## 8. When cap rates mislead — escalate
+
+Cap rates are **wrong tools** for these — escalate to deeper
+methodology or a specialist:
+
+- **Gut-renovation / value-add deals** — NOI doesn't exist yet;
+  underwrite stabilized-yield-on-cost instead, and hold-period IRR.
+- **Owner-occupier / house-hack 2-4 units** — cap rate ignores the
+  saved rent the owner is paying themselves; use a different
+  framework.
+- **Ground-lease / leasehold** — residual-lease-term decay
+  dominates; cap rate misleads.
+- **Short-term rental (STR)** — operates more like a hotel than a
+  rental; opex profile and seasonality break the worksheet.
+- **Triple-net (NNN) commercial** — lease creditworthiness and
+  rollover risk dominate cap rate.
+- **1031-exchange replacement underwriting** — the basis-tracking
+  and like-kind execution timing (45-day identification, 180-day
+  close, qualified intermediary) are CPA / QI territory; surface
+  the concept only and escalate execution.
+
+## 9. Output
+
+Emit an `:::artifact` block with the regulatory header (see system
+prompt §"Regulatory disclaimer") and a structured table:
+
+- Inputs (rents, vacancy, opex line items, capex reserve, financing)
+- NOI build
+- In-place / market / stabilized cap
+- Cash flow, cash-on-cash, DSCR
+- Sensitivity table (2D)
+- Flags (reassessment risk, capex underreserve, DSCR <1, rent gap)
+- Refusal / escalation flags surfaced
+
+Never report a single cap rate without surfacing the in-place vs.
+stabilized gap and the reserve / vacancy assumptions that produced
+it. The number alone is a vibe; the build is the analysis.

package/registry/families/agent-runtime-real-estate-ts/files/methodology/comp-analysis.md

@@ -0,0 +1,137 @@
+---
+id: comp-analysis
+capability: comp-analysis
+retrieved: 2026-04-25
+authority: Appraisal Institute "The Appraisal of Real Estate" 15th ed. (sales-comparison approach); Fannie Mae Form 1004 / 1073 conventions
+---
+
+# Comparable Market Analysis (CMA) Protocol
+
+A CMA is **not an appraisal**. It is a methodology-driven estimate of
+likely market value used to inform listing price, offer price, or
+investment underwriting. For lending, divorce, estate, or court use,
+escalate to a licensed appraiser.
+
+## 1. Subject definition (open `:::question` until complete)
+
+Refuse to compute until ALL of these are known:
+
+- Address or parcel ID
+- Above-grade SqFt (GLA) — basements and unfinished space tracked
+  separately
+- Bedrooms / full baths / half baths
+- Lot size (sf or acres) and lot character (corner, flag, view, slope)
+- Year built and effective age (last gut renovation if material)
+- Condition tier — C1 (new) / C2 (recently renovated) / C3 (well
+  maintained) / C4 (deferred maintenance) / C5 (significant repair) /
+  C6 (uninhabitable)
+- Quality tier — Q1 (custom luxury) through Q6 (economy)
+- Property type (SFR / condo / townhouse / 2-4 unit / co-op)
+- Garage / parking type and count
+- HOA / condo fee monthly (if any)
+
+If the user can't answer, halt and recommend they pull their county
+assessor record or listing history before continuing.
+
+## 2. Comparable-set selection
+
+Target **3-6 closed-sale comparables**. Selection rules, in priority
+order:
+
+1. **Sale window** — closed within 90 days. Stretch to 180 days only
+   if 90-day inventory is thin (<3 candidates) AND apply a time-
+   adjustment (see §3).
+2. **Geographic radius** — within 0.5 mi for urban / suburban; within
+   1 mi for low-density rural. Never cross a school district, zoning,
+   or municipal boundary without flagging it.
+3. **Above-grade SqFt** — within ±15% of subject GLA.
+4. **Bed/bath count** — match exactly when feasible; otherwise
+   adjust per §3.
+5. **Property type** — SFR comps for SFR subject; do NOT mix condo
+   and SFR comps without flagging.
+6. **Condition / quality tier** — within one tier; otherwise adjust.
+7. **Closed sales only** — no active listings, no pendings, no
+   withdrawn. Active listings are useful as a market-temperature
+   sanity check, not as comps.
+
+If you cannot assemble 3 qualifying comps, **refuse the CMA**. Open a
+`:::escalation` block: "insufficient recent comparable sales — escalate
+to a licensed appraiser who can broaden methodology."
+
+## 3. Adjustment grid
+
+Build a per-comp adjustment table. Adjustments apply to the comp's
+sale price (NOT the subject) — the question is "what would this comp
+have sold for if it were the subject."
+
+Standard adjustment lines (in dollars, signed):
+
+| Line | How to size |
+|------|-------------|
+| Time / market | If the comp closed >60 days ago and the market moved, apply the local market index (Case-Shiller MSA for major metros; otherwise a local-MLS median %/month). Cite the index. |
+| GLA (above-grade SqFt) | Apply $/sf at roughly 30-50% of the subject's $/sf — the marginal value of additional SqFt is less than the average. Range: $50-$150/sf in most US markets; calibrate from paired-sales locally. |
+| Bedrooms | $5k-$15k per bedroom delta in median markets; higher in tight urban; never adjust >$25k without a paired-sale citation. |
+| Full baths | $5k-$10k per delta. |
+| Half baths | $2.5k-$5k per delta. |
+| Lot size | Marginal $/sf falls fast above lot-size norm; for typical-lot deltas use $1-$3/sf, never linear scaling on outsize lots. |
+| View / waterfront | Paired-sales required — refuse to invent a number. If no paired sales, flag and exclude. |
+| Condition tier | One tier ≈ 5-10% of price; two tiers ≈ 12-20%. |
+| Garage spaces | $5k-$10k per space delta. |
+| HOA fee delta | Capitalize the monthly delta at a market cap rate (typically 5-7%) — e.g. $200/mo delta × 12 / 0.06 ≈ $40k. |
+| Updates (kitchen / bath) | Cite the comp's listing photos / remarks; $10k-$50k for a kitchen, $5k-$25k for a bath. Refuse to invent. |
+
+Net adjustment per comp should be < 15% of sale price; gross
+adjustment < 25%. If a comp blows past these, the comp is too
+dissimilar — drop it and find another.
+
+## 4. Reconciliation
+
+Compute each comp's adjusted sale price. Then:
+
+1. **GLA-weight** the adjusted prices by similarity to subject GLA
+   (inverse-distance from subject SqFt). Closer-SqFt comps get more
+   weight.
+2. **Exclude IQR outliers** — drop any adjusted price > Q3 + 1.5×IQR
+   or < Q1 − 1.5×IQR. If exclusion drops the comp count below 3,
+   surface the dispersion as a confidence-band penalty instead.
+3. **Compute the weighted mean** AND the median. Report both. If they
+   diverge >5%, the comp set is heterogeneous — flag it.
+4. **Confidence interval** — report a low / mid / high band:
+   - **Low** = min(weighted-mean − 1 SD, lowest adjusted comp)
+   - **Mid** = weighted mean
+   - **High** = max(weighted-mean + 1 SD, highest adjusted comp)
+
+The band, not the point estimate, is the deliverable. A point estimate
+without a band misrepresents CMA confidence.
+
+## 5. When to refuse the CMA
+
+Halt and escalate to a licensed appraiser when ANY of these:
+
+- Fewer than 3 qualifying comps within the rules above
+- Subject is unique (custom architect, oceanfront, historic, post-
+  fire rebuild, ag/farm with non-residential use)
+- Subject's condition is C5 / C6 — uninhabitable comps are scarce
+  and the as-is / as-repaired distinction matters legally
+- Material adjustments (>20% of price) are required to make the
+  comps fit
+- The user's purpose is lending, divorce, probate, court, or
+  property-tax appeal — these need a licensed appraiser
+
+## 6. Output
+
+Emit an `:::artifact` block with the regulatory header (see system
+prompt §"Regulatory disclaimer") and a structured table:
+
+- Subject definition
+- Comp set (with sale date, distance, GLA, bed/bath, sale price)
+- Adjustment grid (per-comp, per-line)
+- Adjusted prices, weighted mean, median
+- IQR outlier exclusions (if any)
+- Confidence band (low / mid / high)
+- Assumptions and caveats
+- Refusal flags surfaced (if any)
+
+Never collapse the band to a single number for the user. Never
+forecast appreciation off the CMA — that is a different exercise with
+different methodology and different risk.

package/registry/families/agent-runtime-real-estate-ts/files/methodology/index.json

@@ -0,0 +1,25 @@
+{
+  "entries": [
+    {
+      "id": "comp-analysis",
+      "capability": "comp-analysis",
+      "path": "./comp-analysis.md",
+      "tags": ["methodology", "cma", "valuation"],
+      "version": "0.1.0"
+    },
+    {
+      "id": "cap-rate-worksheet",
+      "capability": "cap-rate-worksheet",
+      "path": "./cap-rate-worksheet.md",
+      "tags": ["methodology", "investment", "noi", "cash-on-cash"],
+      "version": "0.1.0"
+    },
+    {
+      "id": "offer-strategy",
+      "capability": "offer-strategy",
+      "path": "./offer-strategy.md",
+      "tags": ["methodology", "negotiation", "contingencies"],
+      "version": "0.1.0"
+    }
+  ]
+}

package/registry/families/agent-runtime-real-estate-ts/files/methodology/offer-strategy.md

@@ -0,0 +1,169 @@
+---
+id: offer-strategy
+capability: offer-strategy
+retrieved: 2026-04-25
+authority: NAR "Real Estate Buyer's Agent Council" curriculum; state-promulgated purchase contract conventions (varies by jurisdiction)
+---
+
+# Offer Strategy Framework
+
+This template produces a **strategy memo**, not a binding offer. The
+offer itself — the executed purchase contract, the addenda, the
+financing letter, the earnest-money wire instructions — is drafted by
+the user's licensed real-estate agent and reviewed by the user's
+attorney where applicable. Hand the strategy memo to that agent;
+escalate any execution question.
+
+## 1. Inputs (open `:::question` until complete)
+
+- Subject property and asking price
+- CMA confidence band (run `comp-analysis` first if not yet done)
+- Days on market (DOM) and listing history (price changes,
+  withdrawals, relistings)
+- Listing agent's brokerage and stated commission split
+- User's financing structure (cash / conv / FHA / VA / jumbo / asset-
+  backed / DSCR loan; pre-approval status)
+- User's flexibility on close date, contingencies, leaseback
+- User's hard ceiling (the price above which the deal stops making
+  sense — this is private, never goes in the offer)
+- Market temperature (months of supply; >6 = buyer's market, 4-6 =
+  balanced, <4 = seller's market)
+
+If the user cannot state a hard ceiling, halt. A buyer without a
+walkaway price is the buyer who overpays. Surface this and refuse to
+proceed until they answer.
+
+## 2. Seller-motivation research
+
+Strategy without seller motivation is a guess. Available signals:
+
+- **DOM trajectory** — long DOM with price reductions = motivated;
+  short DOM with no reductions = priced to clear.
+- **Listing remarks** — "estate sale," "relocation," "motivated
+  seller," "as-is," "must sell" are explicit motivation signals.
+- **Concurrent purchase** — if the seller is buying their next home
+  contingent on this sale, leaseback or fast close has high value.
+- **Public records** — divorce, probate, lis pendens, NOD (notice of
+  default), tax-lien, recent mortgage refi/HELOC. Methodology only;
+  the user's agent pulls the records.
+- **Physical condition** — vacant vs. occupied; deferred maintenance
+  visible from comps. Vacant carrying-cost compounds motivation.
+
+Refuse to invent motivation signals. If none are observable, name
+that — the strategy under "unknown motivation" is different.
+
+## 3. Contingency stack
+
+A contingency is a buyer-protective right to walk and recover earnest
+money. Each one weakens the offer. Stack them deliberately.
+
+| Contingency | Default window | Strategic levers |
+|-------------|----------------|------------------|
+| Financing | 17-21 days | Shorten if the user has strong pre-approval; waive only on cash or with pre-funded letter from lender — NEVER without lender sign-off. |
+| Inspection | 7-10 days | "Information-only" inspection (cannot demand repairs, but can walk on material defect) is a middle ground. Full waiver is rare and risky on resale; surface the risk. |
+| Appraisal | tied to financing window | Appraisal-gap clause: "buyer covers up to $X above appraisal" — useful in seller's markets; carries cash-call risk if appraisal misses. |
+| Sale of current home | 30-60 days | Heavily disfavored in any market <6 months supply. Bridge financing or HELOC may eliminate the need; surface alternatives. |
+| Title | typically 5-10 days from prelim | Standard; rarely waived. |
+| HOA/CC&R review | 5-10 days from doc delivery | Standard for condo / HOA properties. Never waive on a complex association. |
+
+Strategic principle: **win on the easiest-to-give contingency, never
+on price alone**. Sellers value certainty almost as much as price; a
+clean contingency stack at $X often beats a contingent offer at $X +
+3%.
+
+## 4. Earnest money calibration
+
+Earnest money is the buyer's "skin in the game" — refundable if
+contingencies are met, generally forfeit on default.
+
+Norms (calibrate to local custom):
+- Standard: 1-3% of purchase price
+- Competitive seller's market: 3-5%, sometimes "non-refundable
+  after inspection" structures (escalate to attorney — these are
+  contractually live wires)
+- Cash / quick close: higher EM signals seriousness; 5%+ is not
+  unusual on cash deals
+
+Never recommend non-refundable earnest money without an `:::escalation`
+to the user's attorney. The forfeiture mechanics are jurisdiction-
+specific and contract-specific.
+
+## 5. Escalation clauses
+
+An escalation clause auto-raises the offer above competing offers, up
+to a cap. Mechanics:
+
+```
+Buyer offers $X, escalating $Y above any bona-fide higher offer,
+up to a maximum of $Z.
+```
+
+Cautions to surface:
+- The cap is the buyer's true ceiling exposed to the seller — this
+  is information leak; some buyers prefer to bid hard once instead.
+- "Bona-fide" must be defined (typically "in writing, fully
+  contingent terms equal or better"). Without that, the clause is
+  unenforceable.
+- Seller-side disclosure of the competing offer is rarely required;
+  the buyer is trusting the listing agent's representation. Some
+  jurisdictions have started requiring proof.
+- Escalation clauses are jurisdiction-sensitive. Some state
+  associations' standard contracts disfavor them. Escalate the
+  drafting question to the user's licensed agent.
+
+## 6. Offer-price recommendation (band, not point)
+
+Combine:
+
+- CMA mid as the anchor
+- Market-temperature adjustment: +0 to +5% in seller's markets, −2
+  to −7% in buyer's markets (calibrate from listing-agent reductions
+  on comps)
+- Motivation adjustment: motivated sellers absorb a discount;
+  unmotivated sellers anchor on list
+- Contingency-strength adjustment: a strong contingency stack lets
+  the user offer 2-4% lower at parity expected-acceptance rate
+
+Output a 3-band recommendation:
+
+- **Aggressive** (high acceptance probability, weakest negotiation
+  position) — typically near or above CMA high
+- **Balanced** — near CMA mid with calibrated escalation cap
+- **Disciplined** — at CMA mid-low, contingency-strong, walk-ready
+
+The user picks. Never pick for them. Surface trade-offs explicitly.
+
+## 7. What this template does NOT do
+
+Hard escalation triggers — emit `:::escalation` and stop:
+
+- **Drafting the actual offer / contract / addendum.** That is the
+  user's licensed agent's job.
+- **Drafting a counter-offer or response to a seller counter.** Same.
+- **Reviewing the seller's disclosures, prelim title, inspection
+  report.** Each goes to a specialist (agent / attorney /
+  inspector).
+- **Recommending non-refundable earnest money structures.** Attorney
+  territory.
+- **Interpreting jurisdiction-specific contract terms** — "as-is"
+  rider, financing-contingency removal mechanics, time-is-of-the-
+  essence default cure periods, statutory rescission windows.
+- **Advising on tax structuring** — installment sale, 1031-exchange
+  identification timing, owner-financing tax treatment. CPA
+  territory.
+
+## 8. Output
+
+Emit an `:::artifact` block with the regulatory header (see system
+prompt §"Regulatory disclaimer") and:
+
+- Subject and CMA reference
+- DOM / motivation signals (or "unknown motivation" flag)
+- Recommended contingency stack with rationale per line
+- Earnest money recommendation with attorney-escalation note if
+  non-refundable structures are involved
+- Escalation-clause analysis (if relevant) with cautions
+- 3-band price recommendation (aggressive / balanced / disciplined)
+- Hard ceiling restated (private to the user)
+- Hand-off note: "Take this memo to your licensed real-estate agent
+  to draft the offer."

package/registry/families/agent-runtime-real-estate-ts/manifest.json

@@ -0,0 +1,178 @@
+{
+  "id": "agent-runtime-real-estate-ts",
+  "description": "Real-estate advisor agent bundle \u2014 explicitly NOT a licensed real-estate broker. Drafts comparable-market-analysis worksheets, cap-rate / NOI / cash-on-cash math, and offer-strategy frameworks for the user's own decision-making. Jurisdiction-agnostic; all binding-decision support (offer drafting, escrow, title, tenant disputes, HOA interpretation, jurisdiction-specific tax/title) escalates to a licensed agent. Fair Housing Act compliant: analyzes property attributes only, never neighborhood social characteristics.",
+  "tags": [
+    "agent-runtime",
+    "real-estate",
+    "advisory",
+    "cloudflare-worker"
+  ],
+  "taxonomy": {
+    "language": "markdown",
+    "runtime": "cloudflare-worker",
+    "surface": "agent-runtime"
+  },
+  "includes": [
+    "agent-base:tangle",
+    "agent-base:secure",
+    "agent-output:blocks",
+    "agent-base:memory",
+    "agent-base:scheduler",
+    "agent-base:mcp-registry",
+    "agent-channels:telegram",
+    "agent-channels:discord",
+    "agent-channels:slack",
+    "agent-channels:whatsapp",
+    "agent-channels:imessage",
+    "agent-channels:gmail",
+    "agent-channels:linear"
+  ],
+  "defaults": {
+    "projectType": "agent-runtime-real-estate",
+    "agentName": "real-estate-advisor",
+    "publicVars": [
+      "AGENT_NAME"
+    ],
+    "routes": [
+      {
+        "path": "/api/chat",
+        "auth": "bearer",
+        "callsTools": [
+          "comp-analysis",
+          "cap-rate-worksheet",
+          "offer-strategy",
+          "escalation"
+        ]
+      },
+      {
+        "path": "/api/health",
+        "auth": "none",
+        "callsTools": []
+      }
+    ],
+    "declaredCapabilities": [
+      "comp-analysis",
+      "cap-rate-worksheet",
+      "offer-strategy"
+    ],
+    "disclaimerRequired": true,
+    "runtime": {
+      "sandbox": "tangle",
+      "noNewPrivileges": true,
+      "capDrop": "ALL"
+    },
+    "outboundDomains": [
+      "api.tangle.tools",
+      "irs.gov",
+      "hud.gov"
+    ],
+    "secrets": [
+      "TANGLE_API_KEY"
+    ]
+  },
+  "files": [
+    {
+      "source": "files/agent.json",
+      "target": "agent.json"
+    },
+    {
+      "source": "files/AGENTS.md",
+      "target": "AGENTS.md"
+    },
+    {
+      "source": "files/methodology/index.json",
+      "target": "methodology/index.json"
+    },
+    {
+      "source": "files/methodology/comp-analysis.md",
+      "target": "methodology/comp-analysis.md"
+    },
+    {
+      "source": "files/methodology/cap-rate-worksheet.md",
+      "target": "methodology/cap-rate-worksheet.md"
+    },
+    {
+      "source": "files/methodology/offer-strategy.md",
+      "target": "methodology/offer-strategy.md"
+    },
+    {
+      "source": "files/README.md",
+      "target": "README.md"
+    },
+    {
+      "source": "files/TOOLS.md",
+      "target": "TOOLS.md"
+    }
+  ],
+  "validationChecks": [
+    {
+      "type": "file-exists",
+      "path": "AGENTS.md"
+    },
+    {
+      "type": "file-exists",
+      "path": "methodology/index.json"
+    },
+    {
+      "type": "agents-md-valid",
+      "path": "AGENTS.md"
+    },
+    {
+      "type": "methodology-index-valid",
+      "path": "methodology/index.json"
+    },
+    {
+      "type": "file-exists",
+      "path": "TOOLS.md"
+    }
+  ],
+  "contextHints": {
+    "commands": [],
+    "entrypoints": [
+      "AGENTS.md"
+    ],
+    "extensionPoints": [
+      "templates/",
+      "src/lib/tools/"
+    ]
+  },
+  "keywords": [
+    "real-estate",
+    "real estate",
+    "property",
+    "comps",
+    "cap-rate",
+    "1031-exchange"
+  ],
+  "tieredKeywords": {
+    "tier1": [
+      "real-estate-agent",
+      "real estate agent",
+      "property-advisor",
+      "property advisor",
+      "real-estate-bot",
+      "real estate bot"
+    ],
+    "tier2": [
+      "comps",
+      "comparables",
+      "cap-rate",
+      "cap rate",
+      "1031-exchange",
+      "1031 exchange",
+      "title",
+      "escrow",
+      "buyer-broker",
+      "buyer broker",
+      "NOI",
+      "cash-on-cash"
+    ],
+    "tier3": [
+      "agent"
+    ],
+    "archetypes": [
+      "agent-in-sandbox",
+      "advisory-agent"
+    ]
+  }
+}