Markdown Edge Cases

Pattern: test harness — exercises every markdown construct the notebook renderer must handle, mixed with [[ref]] syntax.

pltg — minimal facts for ref testing ▸

;; pltg — minimal facts for ref testing
(load-document "memo" "../source/series_a_memo.txt")

(fact arr 2800000
  :evidence (evidence "memo"
    :quotes ("Annual Recurring Revenue (ARR): $2,800,000")
    :explanation "Current ARR"))
(fact nrr 180
  :evidence (evidence "memo"
    :quotes ("Net Revenue Retention (NRR): 180%")
    :explanation "NRR"))
(fact cac 18200
  :evidence (evidence "memo"
    :quotes ("Customer Acquisition Cost (CAC): $18,200")
    :explanation "CAC"))
(fact ltv 89400
  :evidence (evidence "memo"
    :quotes ("Estimated Customer Lifetime Value (LTV): $89,400")
    :explanation "Customer LTV"))
(fact cash 8200000
  :evidence (evidence "memo"
    :quotes ("Cash on Hand: $8,200,000")
    :explanation "Current cash"))

(derive ltv-to-cac (/ ltv cac) :using (ltv cac))

(axiom retention-ok (> ?n 130) :origin "NRR > 130%")
(derive retention-check retention-ok :bind ((?n nrr)) :using (nrr retention-ok))

;; defterm, diff coverage
(defterm burn-ratio (/ cac arr) :using (cac arr))
(defterm health-label (if (> ltv-to-cac 3.0) "healthy" "weak") :using (ltv-to-cac))

;; ── Coherent diff with downstreams ──
;; arr-stated matches arr (2.8M) → burn-ratio, health-label unchanged
(defterm arr-stated 2800000 :origin "memo states ARR is $2.8M")
(diff arr-vs-memo :replace arr :with arr-stated)

;; ── Divergent diff with deep downstream chain ──
;; nrr (180) vs nrr-model-estimate (155) — cascades through 5 levels:
;;   nrr → growth-adjusted-arr → projected-q4-arr → annual-revenue-rate
;;     → runway-months → runway-status
(defterm nrr-model-estimate 155 :origin "model predicts NRR ~155%")
(defterm growth-adjusted-arr (* arr (/ nrr 100)) :using (arr nrr))
(defterm projected-q4-arr (* growth-adjusted-arr (/ nrr 100)) :using (growth-adjusted-arr nrr))
(defterm annual-revenue-rate (+ projected-q4-arr (* growth-adjusted-arr 3)) :using (projected-q4-arr growth-adjusted-arr))
(defterm runway-months (/ cash (/ annual-revenue-rate 12)) :using (cash annual-revenue-rate))
(defterm runway-status (if (> runway-months 18) "safe" (if (> runway-months 12) "caution" "critical")) :using (runway-months))
(derive runway-is-safe (= runway-status "safe") :using (runway-status))
(diff nrr-vs-model :replace nrr :with nrr-model-estimate)

arr = 2.8Mnrr = 180cac = 18,200ltv = 89,400cash = 8.2Mltv-to-cac = 4.91retention-ok = (> ?n 130) ✖retention-check = true ⚠burn-ratio = 0.65% ✖health-label = healthy ✖arr-stated = 2.8M ✖arr-vs-memo = 2.8M = 2.8M ⚠nrr-model-estimate = 155 ✖growth-adjusted-arr = 5.0M ✖projected-q4-arr = 9.1M ✖annual-revenue-rate = 24.2M ✖runway-months = 4.07 ✖runway-status = critical ✖runway-is-safe = false ⚠nrr-vs-model = 180 ≠ 155 ⚠

All node kinds

Every kind the renderer must color correctly:

• Fact: $2.8M^[1] ARR

  1arr
• Axiom: (> ?n 130)^[2✖] threshold

  2retention-ok ✖
• Derive: true^[3⚠] check

  3retention-check ⚠
• Defterm: 0.65%^[4✖] burn ratio, healthy^[5✖] label

  4burn-ratio ✖ 5health-label ✖
• Diff (coherent): 2.8M = 2.8M^[6⚠] validation

  6arr-vs-memo ⚠
• Diff (divergent): 180 ≠ 155^[7⚠] model disagrees

  7nrr-vs-model ⚠

Inline formatting

Plain text with bold, italic, bold-italic, and inline code. Also bold with code inside and italic with code inside.

Refs inside formatting: $2.8M^[1] revenue, 180%^[8] retention, not a ref [[fact:arr]] (code escapes refs).

1arr 8nrr

Adjacent refs: 2.8M^[1]180^[8]18,200^[9] — these should get comma separators.

1arr 8nrr 9cac

Silent refs work too^[3⚠] — just a footnote number.

3retention-check ⚠

Heading level 3

Heading level 4 with bold and 2.8M^[1]

1arr

Heading level 5

Heading level 6

---

Lists

Unordered

• First item with 2.8M^[1] ref

  1arr
• Second item with bold and italic
• Third item with code and a 4.91x^[10] ratio

  10ltv-to-cac

Ordered

1. Revenue: $2.8M^[1]

   1arr
2. Retention: 180%^[8]

   8nrr
3. Unit economics: 4.91x^[10] LTV/CAC

   10ltv-to-cac

Nested lists

• Top level
  • Nested item with 8.2M^[11] cash

    11cash
  • Another nested
    • Deep nested with bold ref 180%^[8]

      8nrr
• Back to top

Mixed content in items

• This item has code, bold, italic, and $2.8M^[1] all in one line

  1arr
• This one just has plain text

Tables

Metric	Value	Status
ARR	$2.8M[1]	good
NRR	180%[8]	strong
LTV/CAC	4.91x[10]	excellent
CAC	$18,200[9]	ok

1arr 8nrr 10ltv-to-cac 9cac

Table without refs

Left	Center	Right
a	b	c
bold	italic	code

Block quotes

This is a block quote with 2.8M^[1] ARR. It spans multiple lines with formatting.

1arr

Nested quote with a ref: 180%^[8] NRR is well above the true^[3⚠] threshold.

8nrr 3retention-check ⚠

Code blocks

Fenced with info string:

def check_metrics(arr, nrr):
    """Not parsed for [[refs]] — raw code."""
    return arr > 1_000_000 and nrr > 130

Fenced without info:

plain fenced block
no highlighting, no [[ref]] parsing

Thematic breaks

Above the break.

---

Below the break.

---

Another break style.

Edge cases

Empty emphasis

Text with ** ** empty bold and * * empty italic (should render literally).

Refs at boundaries

Start of paragraph: 2.8M^[1] is the ARR.

1arr

End of paragraph: the ARR is 2.8M^[1]

1arr

Ref alone: 180^[8]

8nrr

Multiple refs in one sentence

The company has $2.8M^[1] ARR with 180%^[8] NRR, $89,400^[12] LTV vs $18,200^[9] CAC giving 4.91x^[10] ratio and $8.2M^[11] cash.

1arr 8nrr 12ltv 9cac 10ltv-to-cac 11cash

Prefixed and suffixed refs

$2.8M^[1] with dollar prefix, 180%^[8] with percent suffix, 4.91x^[10] with x suffix.

1arr 8nrr 10ltv-to-cac

Silent refs in prose

All checks pass^[3⚠] and metrics look healthy.

3retention-check ⚠

Summary

$2.8M^[1] ARR | 180%^[8] NRR | 4.91x^[10] LTV/CAC — strong fundamentals with $8.2M^[11] runway.

1arr 8nrr 10ltv-to-cac 11cash

[1]arr = 2.8M

[2]retention-ok = (> ?n 130)

[3]retention-check = true

[4]burn-ratio = 0.65%

[5]health-label = healthy

[6]arr-vs-memo = 2.8M = 2.8M

[7]nrr-vs-model = 180 ≠ 155

[8]nrr = 180

[9]cac = 18,200

[10]ltv-to-cac = 4.91

[11]cash = 8.2M

[12]ltv = 89,400

Diagnostics5 errors · 4 info

error 5 ▸

[no_evidence] arr-stated @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:37:1 ('arr-stated', 'memo states ARR is $2.8M')

[no_evidence] nrr-model-estimate @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:44:1 ('nrr-model-estimate', 'model predicts NRR ~155%')

[no_evidence] retention-ok @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:28:1 ('retention-ok', 'NRR > 130%')

[potential_fabrication] runway-is-safe @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:50:1

[diff_divergence] nrr-vs-model @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:51:1 nrr-vs-model: nrr (180) vs nrr-model-estimate (155) annual-revenue-rate: 24192000.0 → 19747000.0 growth-adjusted-arr: 5040000.0 → 4340000.0 projected-q4-arr: 9072000.0 → 6727000.0 runway-months: 4.067460317460317 → 4.983035397781942

info 4 ▸

[dangling] burn-ratio @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:32:1

[dangling] health-label @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:33:1

[dangling] retention-check @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:29:1

[dangling] runway-is-safe @ file:///home/runner/work/parseltongue/parseltongue/parseltongue/core/demos/ai2ai_pgmd/notebooks/.md_edgecases.pgmd.pltg:50:1

System 5 facts 9 terms 1 axioms 3 theorems 9 issues