feat(review): fleet-wide gaming-pattern detector for the self-host orb fleet (#2350)

[JSONbored]

Summary

• Extends computeFleetAnalytics (src/orb/analytics.ts) with gamingPatternFlags: a more targeted detection signal than the existing single-metric outlier check. An instance whose decided-PR volume is unusually high (>2x fleet median), whose merge-precision is unusually high (>fleet median + 0.25), and whose reversal-rate is unusually low (<0.5x fleet median) — all three simultaneously — fits the "mass-submitting only trivially-safe PRs to inflate merge-precision" signature the issue describes. High precision alone isn't suspicious (could be a genuinely careful team); combined with abnormal volume and suspiciously few reversals, it is.
• Surfaced read-only on the existing operator dashboard tile set and the operator-only MCP gittensory_get_fleet_analytics tool summary. Detection only — nothing acts on it automatically, nothing touches the live gate.
• instanceId is the same opaque, HMAC-derived handle already used throughout this pipeline (orb-collector.ts) — never a login or anything more identifying.

Scope (verified against the actual pipeline before implementing)

• Instance-level, not per-miner. The fleet pipeline carries no per-actor identity by deliberate, repeatedly-documented design — review_audit has no login column, and predicted_gate_calibration_ledger is explicitly documented as never-exported, citing this issue by number as the reason why. A genuine per-miner detector would require adding a new anonymized per-actor signal to the export pipeline, a separate privacy-sensitive design decision that deserves its own focused issue/PR rather than a rushed addition here.
• "Duplicate-claim-election win-rate skew" is not implemented. isDuplicateClusterWinnerByClaim's outcome is never persisted anywhere in this pipeline — only the losing side of a duplicate cluster produces a finding (duplicate_pr_risk), with no cluster id and no actor linkage. There is no winner data to measure a win-rate from, and a per-instance duplicate_risk rate would measure something different (how often this instance's own PRs lose a local collision) than "identities farming wins." No proxy is implemented — a misleading one would be worse than none.

Given these two structural gaps, this uses "Advances #2350" rather than closing it.

Test plan

• npm run typecheck clean
• test/unit/orb-analytics.test.ts — 14 new tests: normal-distribution fixture (no flag), inflated-trivial-volume fixture (flags, with exact expected numbers), high-precision-alone / high-volume-alone / high-volume-and-precision-but-normal-reversal (each individually insufficient), null-mergePrecision instance never flags even alongside a real farmer, no-merge-verdict-at-all fleet, unregistered instance never flags, below-MIN_DECIDED instance never flags, empty store, DB-error fail-safe
• test/unit/operator-dashboard.test.ts — 2 new tests: empty-fleet tile shows "no gaming pattern detected", populated fixture surfaces the flagged instance by name
• Measured coverage directly (not whole-file): src/orb/analytics.ts scoped to its own test file — 100% statements/branches/functions/lines
• Confirmed the literal word "farming" cannot appear anywhere in the operator dashboard payload (this repo's own FORBIDDEN_EXPORT_TERMS check, mirroring src/signals/redaction.ts's public-safe vocabulary) — renamed the field/labels to "gaming-pattern" throughout; "anti-farming" stays in source comments only, never serialized

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.20%. Comparing base (74afe7d) to head (53f761e).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5083   +/-   ##
=======================================
  Coverage   94.20%   94.20%           
=======================================
  Files         470      470           
  Lines       39705    39719   +14     
  Branches    14489    14492    +3     
=======================================
+ Hits        37404    37418   +14     
  Misses       1645     1645           
  Partials      656      656           

Flag	Coverage Δ
shard-1	46.40% <93.33%> (+<0.01%)	⬆️
shard-2	33.62% <66.66%> (+0.05%)	⬆️
shard-3	31.36% <93.33%> (+0.03%)	⬆️
shard-4	32.63% <26.66%> (-0.16%)	⬇️
shard-5	33.47% <20.00%> (-0.10%)	⬇️
shard-6	45.10% <26.66%> (+0.19%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/mcp/server.ts	95.62% <ø> (ø)
src/orb/analytics.ts	100.00% <100.00%> (ø)
src/services/operator-dashboard.ts	100.00% <100.00%> (ø)

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.