Read docs/editorial-style.md before writing. Its universal rules and the named technical-evidence register govern all Pharos-owned prose; this skill adds only factual, sourcing, schema, and format requirements.

Yield Coverage Audit Drain

Turn the yield-coverage-audit queue into reviewed Yield Intelligence changes. The queue is a lead: verify live source data, local identity, safety/publication gates, and current methodology docs.

Required Context

Read routed yield docs and scoped instructions. Inspect worker/src/cron/yield-coverage-audit.ts, the relevant resolver/config source under worker/src/cron/yield-sync/ or worker/src/lib/yield-config/, and shared/types/status.ts when the payload changes. Verified docs own durable policy; /agents/ is scratch only.

Queue Snapshot

Prefer a read-only production/cache snapshot. The cache key is yield-coverage-audit; status also exposes a bounded view through yieldHealth.coverageAudit.

Use an already authenticated read-only path. Run this example from worker/, where the Wrangler configuration lives:

npx wrangler d1 execute stablecoin-db --remote --command "select key, value, updated_at from cache where key = 'yield-coverage-audit';"

If remote D1 is unavailable, use the latest fixture or prior run output. The audit runs from worker/src/handlers/scheduled/monthly-yield-audit.ts and the authenticated POST /api/trigger-yield-coverage-audit route, implemented in worker/src/handlers/scheduled/yield-coverage-audit-manual.ts. After deploying reviewed coverage changes, use the manual replay procedure in docs/runbooks/yield-health.md; it uses the normal audit lease. Keep production D1 inspection SELECT-only; do not manually mutate production D1 or dispositions.

Record date, reportedAt, counts, and whether evidence is production, local, or fixture-derived.

For queue accounting, suppression, display caps, and operator-note decisions, read Coverage Audit Queue; for budget interpretation and unavailable supply, use Yield Health first checks and remediation. Record the full pre-cap totals and their basis, not bounded samples. The queue is display-only; never write dispositions or mutate D1 to drain it.

Decision Workflow

1. Triage queue items

Handle queue items by kind:

• stale-auto-lending-override: verify the mapped DeFiLlama pool still exists, is single-exposure, is stablecoin-marked, remains in an allowlisted project, clears the APY and TVL floor, and is not collision-blocked. Remove stale pins or repoint to a verified replacement.
• native-exact-pool: verify tracked stablecoin identity by chain/address or a documented wrapper variant. Promote only exact native holder-yield rows.
• lending-allowlist: verify DeFiLlama protocol category is Lending, CDP, RWA Lending, or Uncollateralized Lending. Add allowlist slugs only from the queue or a clearly equivalent candidate, not broad protocol hunting.
• source-family-adapter: require a reusable protocol-native API/on-chain adapter shape. Do not add one-off scrapers for pages without stable machine payloads.
• Historical quarantine-ready-to-restore: follow retired-reader boundaries and coverage-audit remediation before interpreting retained items.
• unmatched-high-tvl-pool / missing-protocol: a high-TVL stablecoin pool or protocol has no local coverage mapping. Decide venue map, allowlist, or adapter candidacy. Record a documented intentional gap; do not leave the row unaddressed just because it is watch-severity.
• venue-risk-config-missing: an active source's venue protocol has no reviewed entry in YIELD_RISK_CONFIG (shared/lib/yield-source-risk-registry.ts). Author a reviewed risk entry from source evidence; do not guess sub-scores.
• stale-venue-risk-score: a YIELD_RISK_CONFIG entry's reviewedAt has exceeded the review cadence. Re-review the venue and refresh the entry (sub-scores and reviewedAt); do not bump the date without a re-review.
• manifest-missing / ranking-missing: check whether this is a source gap, publication gate, safety gate, stale benchmark, or intentional gap. Do not force coverage around a guard without written rationale.

For retired BIMA, CETES, or vaults.fyi candidates, read the protocol-native source owner. Prove stale-item clearance through recompute after a coverage drain, not an hourly publication.

The kind list above mirrors YieldCoverageAuditQueueItemKind in shared/types/status/yield-liquidity.ts. The source file wins; triage any kind not listed here from its emitting code in worker/src/cron/yield-coverage-audit.ts.

2. Verify source identity

For each candidate:

1. Pull live DeFiLlama Yields data from https://yields.llama.fi/pools or the candidate chart endpoint.
2. Match the candidate to local stablecoin metadata by contract address, wrapper mapping, chain alias, and symbol. Symbol-only matches are acceptable only when address data is unavailable and no same-symbol collision exists.
3. For CoinGecko-derived or contract-derived identity, use existing Pharos chain IDs from shared/types/chain-identity.ts; do not invent unsupported chain identifiers.
4. Add same-symbol false-positive guards to AUTO_LENDING_COLLISION_BLOCKLIST when a live pool is valid but belongs to a different tracked asset.
5. For safety-bypass candidates, add a code comment explaining why the bypass is limited and why the exact pool is acceptable. Use bypasses sparingly.

3. Promote, reject, or defer

Promote only when the candidate passes identity, source, APY/TVL, and publication gates:

• Exact stablecoin yield-bearing pools usually belong in YIELD_POOL_MAP or an explicit source-family config.
• Non-yield-bearing venue opportunities belong in AUTO_LENDING_POOL_MAP only when the pinned DeFiLlama UUID is safer than generic discovery.
• Reusable lending protocols belong in the authored LENDING_PROTOCOLS map in worker/src/lib/yield-config/yield-config-lending-protocols.ts; its slugs and labels derive LENDING_PROTOCOL_ALLOWLIST and LENDING_PROTOCOL_LABELS. Do not edit the derived exports.
• Benchmark-derived assets require a benchmark fetcher, cache schema, yield benchmark type update, rate-derived config, tests, and docs/source roster updates.

Reject or defer when:

• The pool is multi-exposure, below floor, reward-only/no base yield, or same-symbol but wrong asset.
• The product page shows APY but no stable public machine-readable endpoint.
• The source is gated by login/session headers, legal accreditation, or undisclosed off-chain reporting.
• A native module exists but lacks a verified adapter path. Document the candidate and next review date instead of adding speculative code.

Documentation and Versioning

Update routed owners for source/config, queue, or publication changes. Follow Methodology Versioning and ADR-3 only when methodology changes; new providers also update docs/about-page.md.

Validation

Run focused tests for changed surfaces first, then the broader gates:

npx vitest run worker/src/cron/__tests__/yield-config-registry.test.ts worker/src/cron/__tests__/yield-coverage-audit.test.ts worker/src/cron/__tests__/yield-helpers.test.ts worker/src/lib/status/__tests__/yield-health.test.ts
npm run check:stablecoin-data

If the reviewed changes will be published, use pharos-release-runner after the focused checks. GitHub's required PR gate is authoritative; the heavy local merge gate is optional rehearsal work.

Add worker/src/cron/__tests__/fetch-tbill-rate.test.ts when benchmarks change, and frontend status/methodology tests when public status or UI copy changes.

Output

If the queue is empty, report source/date and zero counts, then stop; do not hunt outside the queue. Otherwise report promotions by config, rejections and deferrals with next trigger, checks, and remaining queue-health risk.

Hard Stops

• Do not add manual/on-chain/CMC/DEX supply overrides.
• Do not multiply DefiLlama list-endpoint circulating values by price.
• Do not lower global lending APY/TVL/safety gates to win coverage.
• Do not add broad protocol allowlists outside queue/category-gated evidence.
• Do not scrape unstable product pages when a durable API/source contract is not confirmed.