Practical answers to the ten questions that come up most often in primary care, drawn strictly from the AUA/SUO Early Detection of Prostate Cancer Guideline (2023, amended 2026).
| # | Question | Answer | Guideline Statement(s) | Strength Evidence |
|---|---|---|---|---|
| 1 | Who needs screening, and when? | Begin regular screening at 50–69 (see row 3). Below 50, screening is discretionary: clinicians may offer a baseline PSA at 45–50 in average-risk patients, and should offer screening from 40–45 only for those at increased risk. Always after SDM; PSA is the first-line test. | 1, 2, 4, 5 | Clinical Principle; Strong Rec · A; Conditional · B; Strong Rec · B |
| 2 | What makes someone high risk? | Three named factors: Black race, germline mutations, strong family history — the last of which has no standard definition. | 5 | Strong Rec · B |
| 3 | How often should I screen? | Every 2 to 4 years, ages 50–69. Interval may then be personalized on preference, age, PSA, risk, life expectancy, and health. | 6, 7 | Strong Rec · A; Conditional · B |
| 4 | Is the DRE still needed? | Not as a screen and never as a PSA substitute. Strongly consider it at PSA ≥ 2 ng/mL to establish risk. | 8 | Conditional · C |
| 5 | When should I refer to urology? Are there age-specific cut points? | No referral threshold is issued by the Guideline. Age-varying PSA values are reported (2.5 / 3.5 / 4.5 / 6.5 by decade) but not as a rule. Calculators may inform the decision; low calculated risk may justify forgoing biopsy; PSA > 50 is urgent. | 10, 11, 18 | Conditional · B; Clinical Principle; Expert Opinion |
| 6 | Should the PSA be repeated before referral? | Yes. Repeat before any secondary biomarker, imaging, or biopsy — 25% to 40% normalize on retesting. | 3 | Expert Opinion |
| 7 | Should I give antibiotics to bring the PSA down? | No. Empiric antibiotics should not be used to treat an elevated PSA in an asymptomatic person. Repeat the PSA instead. | 3 | Expert Opinion (Choosing Wisely) |
| 8 | Can I act on a rising PSA alone? | No — velocity is not a trigger. PSA velocity should not be the sole indication for a secondary biomarker, imaging, or biopsy. Confirm the absolute value first (Stmt 3). | 9 | Strong Rec · B |
| 9 | My patient already had a negative workup — now what? | Keep screening on the normal 2-to-4-year interval — a negative biopsy is not a reason to stop. When the PSA is elevated again, what should drive the next step is overall risk, assessed with a tool that credits the prior negative biopsy, rather than the PSA number on its own. | 19–25, 30 | Strong Rec · B/C; Clinical Principle; Conditional · C |
| 10 | When do I stop screening, and how does life expectancy factor in? | No fixed stop age. May discontinue or lengthen at ≥ 75 with PSA < 3 ng/mL. Ten years of life expectancy is the threshold below which screening is unlikely to benefit — use SSA tables or MUSIC rather than clinical gestalt. | 7, 20 | Conditional · B; Strong Rec · C |
Definition that drives everything else: the Guideline defines clinically significant prostate cancer as Grade Group 2 or higher (GG2+), and uses "clinically significant" and "GG2+" interchangeably. The stated rationale is that mortality risk from GG1 disease is extremely low. An abnormal MRI, for Guideline purposes, is PI-RADS 3 to 5 — though it notes some centers restrict that to PI-RADS 4 to 5 based on local reading expertise.
Screening is offered, not ordered reflexively. The Guideline explicitly discourages both drawing a PSA without telling the patient and failing to mention that PSA screening exists (Statement 1, Clinical Principle). The AUA's four SDM elements are: both parties involved in the decision; both share information; consensus built by expressing preferences; and agreement on the plan.
The verb in each statement carries the weight. Only one age band is backed by a directive, Grade A statement — 50 to 69. Everything below 50 is either a conditional option (average risk) or a directive limited to a defined high-risk group.
| Population | What the Guideline says | Statement / level | Supporting rationale in the Guideline |
|---|---|---|---|
| Any patient, ages 50–69 | Should offer regular screening every 2 to 4 years | Stmt 6 · Strong Rec · A | The anchor recommendation, and the only Grade A statement on timing. ERSPC and Goteborg-1 showed reduced metastasis and prostate cancer mortality at 16 to 22 years. |
| Increased risk, ages 40–45 Black race, germline mutation, or strong family history |
Should offer screening beginning at 40–45 | Stmt 5 · Strong Rec · B | Earlier start plus a shorter re-screening interval may be appropriate, particularly where risk of metastatic disease is increased. |
| Average risk, ages 45–50 | May begin screening and offer a baseline PSA — an option to raise in conversation, not a test to order routinely | Stmt 4 · Conditional · B | No randomized evidence supports routine screening before 45. The trials showing benefit (Goteborg-1, ERSPC) began at 50 and 55. The case for a baseline rests on observational data: midlife PSA is a robust predictor of later aggressive disease, metastasis, and death — stronger than race or family history in one review. Median PSA is roughly 0.4–0.7 ng/mL in the 40s and 0.7–1 ng/mL in the 50s. |
| Average risk, under 45 | Not supported | — | Prevalence at 40–45 is low. In PROBASE, screen-detected prostate cancer prevalence at age 45 was 0.2%, with only 4 aggressive (GG3+) cancers; participation was 20% and 35% of those with a biopsy indication declined it. Modeling shows lowering the start age slightly increases lives saved but substantially increases the number of PSA tests. |
What "Conditional" actually means here. In AUA nomenclature a Conditional Recommendation is explicitly non-directive: benefits and risks/burdens appear balanced, and the best action depends on individual patient circumstances. So Statement 4 (Conditional Recommendation, Grade B) is not an instruction to screen 45-year-olds. It is permission to offer a baseline PSA to an average-risk 45-to-50-year-old who, after an SDM conversation, wants one. A patient who declines, or a clinician who defers the conversation to age 50, remains within the Guideline.
Note for the visit: the Guideline is written for all patients with a prostate and refers to them as "people" or "patients" throughout. It separately flags that data are lacking on screening preferences, biomarker accuracy, and management priorities for transgender women and non-binary patients.
Statement 5 (Strong Recommendation, Grade B) names three risk factors that change the age at which screening is offered: Black race, germline mutations, and strong family history of prostate cancer. The narrative adds further modifiers of risk — polygenic risk scores, and calculator inputs such as total PSA, PSA density, percent free PSA, and age.
| Factor | Guideline detail |
|---|---|
| Black race | Disproportionate cancer burden and a two-fold higher risk of prostate cancer death compared with White individuals. Modeling found the risk of fatal prostate cancer, if undiagnosed, reaches general-population levels three to nine years earlier, informing the proposal to begin screening roughly five to ten years earlier than for average-risk patients. More frequent (e.g., annual) screening may address this, but overdiagnosis risk in older Black patients is considerably higher than average, so SDM and personalization matter especially here. |
| Germline mutations | BRCA2 is the strongest signal: IMPACT showed an eight-fold increased risk of aggressive cancer in carriers, for whom systematic PSA screening is indicated. The role of screening in BRCA1 carriers needs further study. Mismatch-repair carriers (MSH2, MSH6) had higher prostate cancer risk than age-matched non-carriers after one screening round. ATM, MLH1, PMS2, HOXB13, NBS1, CHEK2 are listed as needing further study. These patients may benefit from both earlier initiation and shorter intervals. |
| Strong family history | The Guideline states plainly that there is no standard definition, then sets out the criteria common to guideline and consensus statements — detailed immediately below. Two or more first-degree relatives confers roughly a four-fold relative risk versus no family history. |
| Polygenic risk score | The 2026 amendment is deliberately restrained: PRS should not independently dictate imaging or biopsy decisions outside structured screening programs. Most PRS endpoints have been any prostate cancer rather than clinically significant or lethal disease; PRS panels derive largely from European-ancestry GWAS with variable cross-ancestry performance. BARCODE-1 had a 22% participation rate, limiting generalizability. |
Since Statement 5 (Strong Recommendation, Grade B) turns on this phrase but the Guideline declines to fix a single definition, these are the criteria it reports as common across guideline and consensus statements:
| Criterion | Specifics |
|---|---|
| 1. Prostate cancer in relatives | One brother or father, or two or more male relatives, with any one of the following:
|
| 2. Family history of other cancers | Two or more cancers in the hereditary breast and ovarian cancer syndrome spectrum, or the Lynch syndrome spectrum. |
Caveats from the same section:
Genetics referral trigger: the Guideline states that patients with a strong family history should ideally be genotyped to determine whether a pathogenic variant is present (e.g., BRCA1/2, Lynch syndrome, ATM, CHEK2). Absent that information, they may be screened earlier and/or more frequently, similar to known carriers.
The anchor is every 2 to 4 years between ages 50 and 69 (Statement 6, Strong Recommendation, Grade A), supported by ERSPC and Goteborg-1, which showed reduced metastasis and prostate cancer mortality at 16 to 22 years versus no or opportunistic screening.
Statement 7 (Conditional Recommendation, Grade B) then permits risk-stratified tailoring of that interval. The Guideline's own worked examples:
| Patient profile | Suggested interval |
|---|---|
| PSA 1–3 ng/mL, ages 45–70 | Every 1 to 4 years |
| PSA < 1 ng/mL, or PSA below the age-specific median, ages 45–70 | Interval may be prolonged |
| Age 40–59 with PSA below the age-specific median, no strong family history, no known pathogenic germline mutation | Very low risk of metastasis or long-term prostate cancer mortality; risk-stratified (less frequent) screening should be considered at ages 45–59 |
| Age 60 with PSA < 1 ng/mL | 25-year risk of metastasis 0.5% and of death 0.2% in a largely unscreened cohort; reasonable to significantly lengthen the interval or discontinue after SDM, absent other risk factors |
| Age 60 with PSA ≥ 2 ng/mL | Continuing biennial screening was favorable: NNS 23 and NND 6 to prevent one prostate cancer death at 15 years |
One modeling comparison in the Guideline is worth quoting to patients in concept: lengthening the interval to 8 years for a 45-year-old with PSA < 1.0 ng/mL, versus biennial screening from 45 to 69, halved the number of tests while preserving more than 95% of the lives saved.
| Study | Screen ages | Follow-up | Protocol | NNS | NND |
|---|---|---|---|---|---|
| ERSPC | 55–69 | 16 years | Every 2–4 y; biopsy PSA > 3 ng/mL | 570 | 18 |
| ERSPC (2009) | 50–74 | 9 years | Every 2–4 y; biopsy PSA > 3 ng/mL | 1,410 | 48 |
| Goteborg-1 | 50–64 | 22 years | Every 2 y; biopsy PSA 2.5–3+ ng/mL | 221 | 9 |
| ERSPC modeling | 55–69 | Lifetime | Annual, biopsy PSA 3+ ng/mL Every 4 y | 98 129 | 5 5 |
| U.S. modeling | 50–69 | Lifetime | Every 2 y, biopsy PSA 4+ ng/mL Every 2 y, biopsy PSA 2.5+ ng/mL | 243 204 | 3 4 |
On PLCO: the Guideline addresses the negative U.S. trial directly — more than 80% of the control arm received at least one PSA during the trial, the biopsy cut-off was higher than ERSPC (4 vs 3 ng/mL), only 34% of those with elevated PSA were biopsied, and screening stopped after 6 years. A modeling study reconciling PLCO and ERSPC found screening reduces prostate cancer mortality by roughly 30% at 11 to 13 years.
The short answer: not as a screening test, but it retains value once the PSA is elevated.
| Setting | Finding |
|---|---|
| PROBASE (age 50) | PPV of a suspicious DRE 0.87%, versus 4.9% among patients aged 55–59 in PLCO. Of 57 participants with suspicious DRE, 37 were biopsied and 2 had cancer — both GG1. |
| Meta-analysis | Pooled PPV 0.21 for DRE, similar to PSA at 0.22 (p=0.9); combining DRE and PSA gave PPV 0.19 (p=0.5). Adding DRE did not significantly improve PPV. |
| PLCO, absolute 10-year risk of clinically significant cancer, suspicious vs non-suspicious DRE | PSA < 2 ng/mL: 1.5% vs 0.7% (small difference). PSA 2–3 ng/mL: 6.5% vs 3.5% (modestly relevant). PSA ≥ 3 ng/mL: 23.0% vs 13.7% (clinically relevant). All statistically significant. |
| U.S. cohort | With PSA < 4 ng/mL and abnormal DRE, risk of finding cancer was only 3%, but adding DRE improved detection of higher-grade disease. |
| Patient acceptability | Survey data suggest nearly a quarter of patients may forgo prostate cancer screening altogether when it includes up-front DRE with PSA testing. |
The practical rule the Guideline lands on: among patients with PSA ≥ 2 ng/mL, clinicians should strongly consider supplementary DRE to establish risk of clinically significant prostate cancer. In patients going to biopsy for an elevated screening PSA, an abnormal DRE improves the PPV for both any prostate cancer and GG2+ detection.
Read this first. The Guideline does not issue a numbered recommendation defining a PSA threshold for urology referral, and it does not endorse a single cut point. It states that clinicians "may tailor the definitions of elevated risk and elevated PSA to the clinical situation at hand," and it repeatedly steers away from thresholds in isolation — most explicitly cautioning against using PSA density threshold values in isolation for management decisions. The figures below are the age-varying thresholds the Guideline reports from the literature, not a referral rule.
| Age band | Threshold cited in most studies |
|---|---|
| 40s | 2.5 ng/mL |
| 50s | 3.5 ng/mL |
| 60s | 4.5 ng/mL |
| 70s | 6.5 ng/mL |
The Guideline's own framing of why a flat 4 ng/mL fails: PSA rises with age in people without cancer, so the threshold for "elevated" should rise with age. Four is too high for people in their 40s and 50s, and too low for those in their 70s and 80s, where overdiagnosis risk is high. The commonly cited 4 ng/mL derives from early studies of the highest levels seen in patients presumed cancer-free; the 3 ng/mL figure comes from the ERSPC biopsy referral threshold.
What does not by itself justify referral: PSA velocity — see Question 8, where Statement 9 is covered in full. Conversely, Statement 11 (Clinical Principle) allows forgoing near-term biopsy when calculated risk is low, even with a mildly elevated PSA, provided the patient is informed of the underdiagnosis risk and the need for follow-up.
5-ARIs decrease PSA after at least 6 months of use. Older studies suggested doubling the serum PSA as an adjusted baseline, but the Guideline cautions that kinetics vary — one trial found only about one-third of patients on 5-ARI therapy experienced a 40% to 60% decline at one year.
Yes. Statement 3 (Expert Opinion) directs that for a newly elevated PSA, the PSA be repeated before proceeding to a secondary biomarker, imaging, or biopsy. This is arguably the single highest-yield primary care action in the document.
| Exposure | Effect per the Guideline |
|---|---|
| DRE | Does not appreciably alter PSA |
| Bicycle riding | Does not appreciably alter PSA |
| Ejaculation | Most controlled studies: no significant impact or a modest increase of about 10% |
| Urinary tract infection | Transient increase — repeat after an appropriate interval to allow return to baseline |
| Instrumentation: recent catheterization, prostate biopsy, cystoscopy, urinary retention | Transient increase — repeat after an appropriate interval |
| 5-ARI use ≥ 6 months | Decreases PSA; see the caveat under Question 5 |
No. The Panel strongly supports the AUA Choosing Wisely position that empiric antibiotics should not be used to treat an elevated PSA in an asymptomatic person. This is backed in the Guideline by two randomized trials — a multi-institutional RCT of empiric antibiotics for elevated PSA, and an RCT of six weeks of fluoroquinolone therapy that found no clinical benefit.
The correct move instead is Statement 3 (Expert Opinion): simply repeat the PSA after an appropriate interval, since 25% to 40% normalize without any intervention. Where a genuine urinary tract infection or recent instrumentation is present, treat the infection on its own merits and allow time for the PSA to return to baseline before repeating.
Note that antibiotics do appear elsewhere in the Guideline, in a different context — periprocedural prophylaxis around transrectal versus transperineal biopsy (Statement 35, Conditional Recommendation, Grade B). That is a urology decision, not a screening one, and does not change the answer above.
No. This is one of the few places the Guideline issues a Strong Recommendation against a common practice: for people undergoing screening, PSA velocity should not be used as the sole indication for a secondary biomarker, imaging, or biopsy (Statement 9, Strong Recommendation, Grade B).
| Finding | Detail |
|---|---|
| It adds nothing once you know the basics | With knowledge of age, PSA, DRE, percent free PSA, family history, and whether there was a previous biopsy, large-scale studies in Europe and the U.S. showed that adding PSA velocity at various thresholds does not add value in predicting clinically significant prostate cancer. |
| The steepest rises point away from cancer | The Guideline calls this paradoxical: a very high PSA velocity (> 3 ng/mL/year) is more closely associated with inflammation on biopsy than with cancer. |
| Much of what looks like a trend is noise | PSA fluctuates year to year, and a newly elevated value returns to normal on retesting in 25% to 40% of people. Among STHLM-3 patients with PSA 3–10 ng/mL retested 8 weeks later, 17% fell below 3 ng/mL. A two-point “rise” may simply be two points of biological variation. |
| The same logic applies after a negative biopsy | Statement 21 (Strong Recommendation, Grade B) separately directs that a PSA threshold alone should not decide whether to repeat a biopsy, since the original biopsy was usually prompted by an elevated PSA in the first place — the result is repeated unnecessary biopsies. |
The one place velocity does appear: in the repeat-biopsy setting, the Guideline lists a PSA velocity of 0.27 ng/mL/year or greater among the factors that may identify patients likely to harbor clinically significant cancer after a negative biopsy and a negative MRI (Statement 31, Conditional Recommendation, Grade B), alongside PSA density > 0.15 ng/mL and PHI density > 0.44. That is a urology-side consideration in an already-worked-up patient — not a screening threshold, and not a contradiction of Statement 9 (Strong Recommendation, Grade B).
A prior negative biopsy or a prior negative MRI lowers risk — it does not close the question. The Guideline devotes seven statements to this scenario, and the two most relevant to primary care are both Strong Recommendations against reflexes that feel intuitive.
| Do | Don’t |
|---|---|
| Resume screening on the normal interval. Re-evaluate within two to four years, or sooner depending on risk and life expectancy, typically with a PSA (Stmt 22, Clinical Principle). | Don’t stop screening because the biopsy was negative (Stmt 20, Strong Rec, Grade C). The systematic review found 5% to 25% of patients undergoing a subsequent biopsy in the short term are diagnosed with GG2+ disease, and prostate cancer mortality over a 20-year horizon after an initial negative biopsy ranges from 1.4% to 5.2%. |
| Use a risk tool that credits the negative biopsy. At re-evaluation, use a risk assessment tool incorporating the protective effect of the prior negative biopsy (Stmt 23, Strong Rec, Grade B) — a previous biopsy genuinely reduces the risk of finding GG2+ disease on a subsequent one. | Don’t treat a PSA back above threshold as an automatic repeat biopsy (Stmt 21, Strong Rec, Grade B). The first biopsy was usually prompted by that same elevated PSA, so applying the same threshold again just produces repeated unnecessary biopsies. Re-engaging urology is reasonable — what is not reasonable is assuming the number alone settles the question. |
| Close the loop after the biopsy. Review results, reassess the risk of undetected or future GG2+ disease, and decide together whether to stop screening, continue, or pursue adjunctive testing (Stmt 19, Clinical Principle). | Don’t order biomarkers reflexively when the probability of GG2+ disease is low (Stmt 24, Clinical Principle). A low PSA density (≤ 0.10) at initial biopsy already predicts low likelihood of GG2+, including with a negative or equivocal MRI. |
One thing to check on the prior pathology report. If the biopsy was reported as anything other than benign — atypical glands or high-grade PIN in more than one core, for example — that patient belongs back with urology rather than in routine screening. The Guideline assigns each of these findings its own statement (26 to 29) with a distinct follow-up pathway.
Statement 7 (Conditional Recommendation, Grade B) makes discontinuation a legitimate, SDM-based decision rather than an age cut-off. There is no single stop age in the Guideline. The specific guidance:
| Scenario | Guideline position | Supporting data |
|---|---|---|
| Age ≥ 75 with PSA < 3 ng/mL | Clinicians may discontinue or substantially lengthen the interval | In the Baltimore Longitudinal Study of Aging, patients ≥ 75 with PSA < 3 ng/mL were unlikely to be diagnosed with aggressive cancer, and no patient aged 75–80 with PSA < 3 ng/mL died of prostate cancer during their remaining lifetime. |
| Ages 70–74, previously screened, no cancer diagnosis | Discontinuation could be considered at PSA < 3.0 ng/mL | ERSPC Rotterdam cumulative prostate cancer–specific mortality by age 85: 0.54% overall (95% CI 0.40–0.70), 0.11% with PSA < 2 ng/mL (0.05–0.27), and 0.85% with PSA 2–3 ng/mL (0.47–1.5). |
| Age 60–61 with baseline PSA < 2 ng/mL | Further screening unlikely to be beneficial after age 68–70 if PSA remains < 2 ng/mL | Long-term ERSPC follow-up; actuarial probability of clinically significant cancer at 16 years was 1.2–1.5% for ages 55–69 with baseline PSA < 1.0 ng/mL. |
| Significant comorbidity | Earlier cessation is reasonable | A modeling study found stopping at ages 66 and 72 for severe and moderate comorbidity respectively produced harms and benefits similar to screening an average-health person to age 74. |
| Ages 70–80 generally | The window where individualization matters most | Higher risk of competing mortality; overdiagnosis risk rises with increasing age. |
One thing that never justifies stopping: a single negative biopsy. Statement 20 (Strong Recommendation, Grade C) states that screening should not be discontinued based solely on a negative prostate biopsy. The systematic review found 5% to 25% of patients undergoing a subsequent biopsy in the short term are diagnosed with GG2+ disease, and over a 20-year horizon prostate cancer mortality after an initial negative biopsy ranges from 1.4% to 5.2%. Statement 22 (Clinical Principle) directs re-evaluation within the normal two-to-four-year interval, or sooner.
Life expectancy is one of the six named inputs in Statement 7 (Conditional Recommendation, Grade B), and it is the input that most often decides the question. The threshold the Guideline uses is ten years.
| Tool | Note |
|---|---|
| Social Security life tables | The simple approach. Based on current SSA data, American patients older than 77 have less than a 10-year life expectancy. |
| MUSIC life expectancy tool | A paper-based tool from the Michigan Urological Surgery Improvement Collaborative that incorporates comorbidities. |
| Insurance company calculators | Many are online and incorporate tobacco, alcohol, physical activity, and comorbidities. |
Worth internalizing: the Guideline states that for estimating life expectancy, these tools are likely more reliable than individual clinician judgment. It also endorses AHRQ's SHARE approach for structuring the conversation: Seek participation, Help explore options, Assess values and preferences, Reach a decision together, Evaluate the decision.
Useful for setting expectations, and for reassuring patients who fear that referral means an inevitable biopsy.
| Step | Guideline position | Stmt · Level |
|---|---|---|
| Pre-biopsy counseling | Patients should be told a biopsy may identify a cancer with a sufficiently low mortality risk that it could be safely monitored on active surveillance rather than treated. | 12 · Clinical Principle |
| MRI before initial biopsy | May be used to increase detection of GG2+ cancer. | 13 · Conditional · A |
| PI-RADS reporting | Radiologists should use PI-RADS for mpMRI reporting. Detection of clinically significant cancer by score: PI-RADS 1–2, 7%; 3, 11%; 4, 37%; 5, 70%. | 14 · Moderate · C |
| Suspicious MRI lesion, biopsy-naïve | Targeted biopsies of the lesion; systematic template biopsy may also be performed. | 15 · Moderate / Conditional · C |
| Negative MRI but elevated GG2+ risk | Proceed with systematic biopsy. NPV of a negative MRI (PI-RADS 1–2) for GG2+ was 91% in a review of 42 studies — roughly 1 in 10 still harbor GG2+ disease. | 16 · Moderate · C |
| Urine or serum biomarkers | May be used when further risk stratification would change the biopsy decision — typically for PSA in the 2.5–10 ng/mL range. Not for use when risk is already so low or so high that the result would not change management. | 17 · Conditional · C |
| Biopsy route | Either transrectal or transperineal is acceptable. | 35 · Conditional · B |
A number that helps in counseling: a meta-analysis within this Guideline found secondary biomarkers would reduce the number of biopsies by 35% (95% CI 26–44%), at the cost of 9% (95% CI 6–11%) of clinically significant cancers not being detected. Referral is not a one-way street to biopsy.
Grade C supporting a Strong Recommendation is described in the Guideline as rarely used — Statement 20 is one such instance.
Flagged so this summary is not over-read:
The Guideline also observes that clinicians tend to discuss the potential benefits of screening far more often than the potential harms, and identifies an unmet need for decision aids in multiple languages and at varying levels of health literacy.
Guide by @michael_gorin using Claude Opus 5