adam-s

Market data integrity rules — gap filling, look-forward bias prevention, and stock split handling. Use when writing or reviewing code that loads prices, fills missing data, computes returns, backtests strategies, or makes trading decisions. Applies across Python, TypeScript, SQL, and TimescaleDB.

Analyze Volatio debug logs from /tmp/volatio-debug/. Use when the user asks to check debug logs, analyze problems, or diagnose simulation issues.

Apply bottom-up systematic testing to a multi-layer system. Use when asked to test, debug, or verify a pipeline that spans multiple layers (e.g., Python worker → bridge → processor → API → UI).

Drizzle Kit workflow for Volatio using idempotent migrations (generate + edit + migrate), never use push. Idempotent patterns prevent partial failure issues, enable safe re-runs, and work cleanly in CI. Use when schema changes, migration errors, or database sync issues occur.

Template for creating Volatio research experiments with Optuna optimization and portfolio backtesting. Includes proper train/test splits, lookahead bias prevention, results saving, and validated patterns from Exp 051-053.

Financial research look-ahead bias verification for Volatio ML pipelines, mandatory baseline comparisons, timezone handling, feature leakage detection. Use when model accuracy seems too good or validating prediction experiments.

Creating PROGRESSION.md documents for research experiment series in Volatio. Captures the evolution of ideas, learnings, and breakthroughs across numbered experiments (001, 002, ...). Use when documenting experiment series or reviewing research evolution.

Run Volatio CI locally before pushing (ci-local.sh), mirrors GitHub Actions exactly, catches failures fast. Use when preparing to push, debugging CI failures, or verifying changes will pass.

Discover any website's API and create domain plugins with proxy routes. Use when the user wants to create an API for a website, discover a web service's data transport, add a new domain, capture browser traffic, build typed API clients, or integrate with a third-party site. Also use when the user mentions a website name and wants to interact with it programmatically.

Build a complete application from a short description. Asks the developer clarifying questions, generates data requirements, launches discovery agents, and builds a dashboard. Use when the developer describes what they want to build in plain language — "compare tickets", "track prices", "search jobs across sites".

Deploy to EC2 production server. Use when deploying new code, rebuilding Docker images, wiping and reseeding the database, checking container health, or managing the EC2 instance via AWS CLI.

Run local CI checks and verify GitHub Actions status. Use before committing, after pushing, or when asked to check/fix CI.

Build Next.js dashboard pages that consume domain proxy APIs. Use when the user wants to create a dashboard, build a UI page, add a search interface, display data from captured APIs, create comparison views, or build any frontend that calls /api/<domain>/ endpoints.

Iterative debugging with targeted logs. Use when browser connections fail, traffic capture returns empty, proxy routes return errors, WebSocket issues, or any problem that can't be solved on the first attempt. Add logs, read output, narrow the search, repeat until fixed, then clean up.

Use sub-agents to iteratively improve dashboard-building instructions. Three-phase pipeline — discover APIs, build dashboard matching a wireframe, review code + UI with a SOTA reviewer agent. The dashboards are throwaway; instruction improvements and framework code fixes are the product.

Use sub-agents as test subjects to iteratively improve .claude/ instruction files. Run agent → inspect → fix instructions → re-run until agents follow the protocol correctly without hints.

Compare two eval runs and report what changed. Reads both runs' events, transcripts, and produced artifacts. Writes a short markdown summary classifying differences as regression, improvement, or neutral.

Generalized recursive iteration loop. Runs parallel sub-agents against a target, scores deterministically, diagnoses instruction gaps, applies fixes, and recurses until the stop condition is met or max depth is reached.

Run an evaluation against an eval with a specific config

Write a handoff document so a new chat can continue the work

Show evaluation results and comparisons

Test-driven development of a Hono/Bun WebSocket application. Read requirements, read tests, build server, verify, iterate until all tests pass.

Scaffold a new evaluation

Stop all agent-spec processes, clear ports, remove sandboxes, verify clean state.

Audit every user-facing legal citation in apps/web against the canonical OBL register at docs/case-management-research/enforcement-map.md. Enforces the three-tier legal-leakage taxonomy from CLAUDE.md — verbatim regulation text lives only in `DisclosureCard.citation`; eyebrows, headings, button labels, and form copy stay plain-language; acronyms expand on first use on family surfaces. Flags fabricated cites (a §-reference that doesn't resolve to a real OBL entry), Tier-2 leakage (legal jargon in chrome), and Tier-3 leakage (raw acronyms in family chrome). Runs inline. Use when the user says "audit citations", "check grounding", "legal-leakage check", or after lifting a prototype page that mentions any of `§ / OBL- / CFR / IDEA / PWN / ARD / IAES / REED / manifestation`.

Adversarial review of design-system fidelity in the plugin/ subtree (plugins/*/src/pages, plugins/*/src/components, packages/ui). Hunts editorial drift (facts surfaced twice, sections that dump tables instead of answering questions, lists that should be summary stats), atomic-unit drift (the same person row / item row / section header rendered with different markup), and proportional drift (arbitrary spacing values, ad-hoc letter-spacing, hand-rolled cards next to the Card primitive). Pairs with packages/ui/src/boundary.test.ts which enforces the deterministic patterns. Runs inline in the current chat (no subagent fork). Use when the user says "design review", "design audit", "consistency check", "is the design system holding", or after adding any UI surface.

Drive one or more control-room workflows end-to-end and diff the capture against the prior run for the same workflow. Surfaces which (persona, step) cells changed and why, so a cascade regression caused by a recent change becomes a structured finding instead of an after-the-fact "I think something broke." Runs inline in the current chat (no subagent fork). Use when the user says "run the harness", "drive a workflow", "verify the cascade", "check for regressions across personas", "run the journey loop", or after a non-trivial change to rights derivation / state-machine guards / Drizzle queries / persona shells.

Inline mutation testing for the apps/ monorepo — inject targeted regressions into @iep/domain (machine guards, rights derivation), @iep/data (Drizzle queries, schema), @iep/services (mutation contracts, composition), or apps/web (server actions), then run `pnpm -w typecheck` + `pnpm -w test` and report which mutations SURVIVED. Surviving mutations are direct evidence of validation gaps. Runs inline in the current chat (no subagent fork). Use when the user says "trickster", "mutation test", "break the code", "can the tests catch regressions", "grade the validation", or after adding code whose coverage is uncertain.

Adversarial bug-hunting code review of the plugin-first framework in plugin/ — @plugin/{core,runtime,ui}, the host app, and plugin packages. Hunts real bugs in manifest validation, merger floor enforcement, mutator atomicity, registry collision detection, auditor declared-emitter discipline, state-machine actor lifecycle, realtime subscriber isolation, and the plugin-scoped service injection seam. Not style nits. Runs inline in the current chat. Use after non-trivial framework changes or before a phase commit.

Adversarial bug-hunting code review of the apps/ monorepo — apps/web, @iep/domain, @iep/data, @iep/services, @iep/auth. Hunts real bugs in state-machine guards, rights derivation, Drizzle queries, server actions, persona/audit emission, and idempotent seed/migration code. Not style nits. Runs inline in the current chat (no subagent fork). Use when the user asks for a "red team", "bug hunt", "adversarial review", or at checkpoints during long lifts.

Adversarial review of the TEST AND VALIDATION SURFACE of the apps/ monorepo — vitest unit suites in @iep/domain + @iep/data + @iep/services + @iep/ui + apps/web, the Drizzle integration tests under @iep/data/queries, the boundary.test.ts pattern enforcers in @iep/ui and @iep/data, and the OBL enforcement-map. Hunts tautological assertions, integration tests that don't reach the DB, missing regression coverage for known-deferred obligations, and surfaces that document an invariant without enforcing it. Runs inline in the current chat (no subagent fork). Use when the user says "red team the tests", "audit the tests", "are our tests any good", or after adding tests/validation that needs scrutiny.

Find a real upstream PrairieLearn bug with no open PR, reproduce + fix it in the harness, and publish it as an issue + pull request ON THE USER'S OWN FORK (never upstream) — packaged as a human-readable proposal a maintainer could copy over. Wraps debug-issue with a discovery front-end, a dual-test proof (PL's own tests + harness evidence), and a fork-only SHARE phase. Runs inline. Use when the user says "contribute a fix", "open a PR on my fork", "propose a fix to PrairieLearn", "find an issue to fix and PR it", or "/contribute".

Drive one PrairieLearn GitHub issue or PR end-to-end through the harness — PRE-FLIGHT → GATHER → SCAN → CLASSIFY → BUILD. Spawns an isolated worktree + Docker journey, reproduces the bug with Playwright, injects the trace spine closest to the suspect code, classifies root cause from the merged timeline, then verifies a fix with before/after evidence. Runs inline. Use when the user says "debug issue N", "repro this PR", "investigate

Drive a PrairieLearn Playwright journey — persona switching (student/instructor/admin), render-gating (never screenshot a blank/500 and call it fine), numbered before/after screenshots, socket-safe waits, and cold-vs-warm diffing. Spins the journey's containers, runs scenario.ts, captures evidence. Runs inline. Use when the user says "run the journey", "run the scenario", "screenshot this flow", "drive it as instructor/student", "check for regressions across personas", or after editing a fix to re-verify.

Isolate one file/function/module and exercise it in isolation with full observability — instrument its inputs, intermediate values, and return with timestamped trace logs (Python or TS/JS), drive it with crafted cases, and assert expected results. The fastest empirical layer: between forbidden static reasoning and a full Docker/Playwright boot. Drives a tight Observe→Orient→Decide→Act loop. Runs inline. Use when the bug lives in a specific function/module, the user says "isolate", "exercise the slice", "test it in isolation", "instrument inputs/outputs", or when the full stack is overkill or won't boot.

Add precise timestamped debug logging across PrairieLearn using the concentric/adaptive model — start at the code CLOSEST to the issue (ring 0), read the merged cross-process timeline, widen the field only if the narrow ring didn't reveal it. Covers the TS⇄Python boundary, SQL, sockets, and the zygote via @pldebug/trace. Runs inline. Use when the user says "add logging", "trace this", "instrument", "where is time going", "why is this swallowed", or when debug-issue reaches CLASSIFY and the cause isn't obvious.

Run one eval-driven instruction-tuning iteration on the harness itself. A blind sub-agent solves a CLOSED PrairieLearn issue (no access to the merged fix or this conversation), a blind judge compares its answer to the accepted PR, and you diagnose where the harness/.claude failed it and apply GENERALIZED improvements. The product is the instructions, not the PL fix — the code is thrown away. Runs inline (parent-orchestrated). Use when the user says "tune", "improve the harness", "run an eval iteration", "look back how can we improve .claude", or after picking a closed-issue case.

Adversarial bug-hunting review of a PrairieLearn diff, PR, or worktree change — hunts REAL bugs (not style) in the seams where PL breaks: the TS⇄Python code-caller boundary, zygote serialization, SQL/Drizzle edge cases, assessment/grading state, socket event isolation, auth/CSRF, and migrations. Severity-grouped findings with file:line, each with a concrete trigger. Read-only, runs inline. Use when the user says "red team", "bug hunt", "adversarial review", "what could break", or before posting a fix to a PR.

Launch an adversarial mutation-testing agent (Opus) that injects targeted regressions into production code, runs the test suite, and reports which mutations SURVIVED — surviving mutations are direct evidence of test-coverage gaps. Use when the user says "trickster", "mutation test", "break the code", "can the tests catch regressions", "grade the tests", or after adding production code whose test coverage you're not sure about.

Run a bounded live audit of the extension against multiple HN handles in parallel — Playwright + real Chromium + real HN, time-series snapshots, deterministic divergence analysis. Use when the user asks to "audit", "sit and observe", "live test", or "check whether the extension is correctly tracking real HN activity across multiple users". Bounded by wall time AND request budget — never unbounded.

Launch a Jony Ive–style product design adversary (Opus) to critique UI/UX *within* the project's existing aesthetic constraints. Finds what is graceless, disproportionate, clumsy — in the designer's own voice. Use when the user wants a "design critique", "Jony Ive review", "UI review", or says the interface "needs improvement".

Launch an adversarial bug-hunting code reviewer (Opus) to find real bugs, correctness issues, concurrency hazards, and server/resource abuse risks — not style nits. Use when the user asks for a "red team", "bug hunt", "adversarial review", or at checkpoints during long coding tasks.

Launch an adversarial reviewer (Opus) to find bugs IN THE TEST SUITE — tautological assertions, weak oracles, shim/mock lies, stale fixtures, coverage gaps, clock-mock pitfalls. Use when the user says "red team the tests", "audit the tests", "are our tests any good", "find weak tests", or after adding a large batch of tests that needs scrutiny.

Deploy to EC2 production server (volatio-prod). Use when deploying new code, rebuilding Docker images, wiping and reseeding the database, checking container health, or managing the EC2 instance via AWS CLI.

Visual development with Patchright screenshots. Use when building, fixing, or reviewing UI — takes screenshots, analyzes layout and functionality, iterates until correct, then cleans up.

Reverse-engineer any web API by capturing browser traffic with Patchright route interception. Use when building typed API clients for sites without public documentation.

Iterative debugging with DEBUG() logs. Use when you can't solve a problem on the first attempt — add targeted logs, read output, narrow the search, repeat until fixed, then clean up.

Bottom-up systematic validation for the deep-research pipeline. Use when asked to test, debug, or verify any layer from Python worker through to the dashboard UI.

The 5-step build pipeline (PRE-FLIGHT, DATA, STRATEGY, OPTIMIZE, file structure). Use when creating a new experiment — pre-flight documentation, building datasets, implementing strategies.

Self-Improving Instruction Tuning

Emergency stop — kill all agents, clean APC, restore references, prune worktrees, verify DB clean. Use when stopping a cycle or recovering from a crash.

Launch an adversarial mutation-testing agent (Opus) that injects targeted regressions into goldseam's production code, runs the suites, and reports which mutations SURVIVED — surviving mutations are direct evidence of test-coverage gaps. Use when the user says "trickster", "mutation test", "break the code", "grade the tests", or after adding production code whose coverage is unproven.

Launch an adversarial bug-hunting code reviewer (Opus) to find real bugs, correctness issues, and trust hazards in the goldseam plugin/engine — not style nits. Use when the user asks for a "red team", "bug hunt", "adversarial review", or at checkpoints after a milestone lands.

Launch an adversarial reviewer (Opus) to find bugs IN THE TEST SUITE — tautological assertions, stub lies, jsdom-shim divergence, vacuous system checks, coverage gaps. Use when the user says "red team the tests", "audit the tests", "are our tests any good", or after adding a large batch of tests.

Launch a mutation-testing agent (Opus) that injects a targeted regression into production code, runs the checks, and reports whether it SURVIVED — a survivor is direct evidence of a test-coverage gap. Use when the user says "mutation test", "break the code", "grade the tests", or after adding production code whose coverage is unproven.

Launch a skeptical bug-hunting code reviewer (Opus) to find real bugs, correctness issues, and trust hazards in this project's production code and data pipeline — not style nits. Use when the user asks for a "red team", "bug hunt", "break-it review", or at checkpoints after a milestone lands (especially before any publish/deploy).

Launch a skeptical reviewer (Opus) to find bugs IN THE TEST SUITE — tautological assertions, mock/stub divergence from reality, blindly-regenerated golden snapshots, vacuous invariance checks, coverage gaps. Use when the user says "red team the tests", "audit the tests", "are our tests any good", or after adding a batch of tests.