Murathanx12
UserFree, open-source market intelligence platform. Institutional-grade analysis, accessible to everyone.
Categories
Indexed Skills (5)
lane-integrity-check
Run before AND after any change that goes near the paper lanes, lane YAMLs, the scheduler's lane path, rebalance logic, or NAV/positions tables. Verifies the track record's integrity invariants — config hashes byte-stable, no spurious segments, registry untouched, positions consistent with NAV. The paper_nav write-path is sacred (CANON §5).
pre-register-trial
Run BEFORE any new signal, strategy, factor, overlay, or hypothesis starts accruing data or gets evaluated. Creates the tamper-evident commitment (hypothesis + primary metric + decision rule + earliest decision date) in docs/TRIALS/ and the experiment registry. If it isn't pre-registered, it didn't happen (CANON §6).
seed-a-lane
The attended procedure for bringing a NEW paper lane live (new strategy arm, overlay trial, event-driven lane). Use whenever a new lane, lane variant, or lane seed flag is requested. Seeding is env-gated and attended — Murat flips the flag; sessions never improvise a seed.
silent-fragility-audit
Scan a change (or a subsystem) for the project's house failure mode — code that runs green and silently does nothing. Use after adding any collector, fetcher, model loader, or try/except; when a subsystem's output looks plausible but unverified; or on request ("audit X for silent fragility"). Wrong math gets caught by tests; silence doesn't.
verify-prod-after-deploy
Run after every push that deploys to Railway — verifies the deploy actually landed AND the changed surface actually works live. Use when a push to main completes, when a deploy is claimed done, or before reporting any shipped feature as live. Green tests are not a live verification (the insider collector passed 12 tests while 403-ing on 100% of prod fetches).
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.