Results for “regime-verdict”

实验判决门：Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

Outcome Eval

Structured visual QA verdict for screenshot-to-reference comparisons

Record a decision: finalized → ADR (optionally rule + guide or spec + plan); open proposal → RFC. Use for 'we decided', 'record this decision', 'make it a standard', 'draft an RFC', 'should we switch to Y'. Not for feature planning or documenting existing code.

Debugging constant Markov regime features in RL observations - when HMM probabilities show uniform values instead of dynamic regime estimates

Resolves legacy references to the vercel-react-expert capability by routing to the current runtime agent, plugin, or narrower skill.

Detect structural macro regime transitions (1-2 year horizon) using cross-asset ratio analysis, analyzing RSP/SPY concentration, yield curve, credit conditions, size factor, equity-bond relationship, and sector rotation to identify regime shifts.

Track legislation, regulatory actions, and legal developments affecting prediction markets, crypto, and AI agents, with stage, impact, affected protocols, and operator actions.

Explains how Agent Script instruction resolution works across three phases, with patterns for pre-LLM setup, LLM reasoning, and post-action loops.

Audit-ready decision artifacts for LLM outputs — assumptions, risks, recommendation, and review gating (schema-valid JSON).

research-synthesis

The single highest-leverage decision is recognizing that most of what teams call

Evaluate and rank agent results by metric or LLM judge for an AgentHub session.

Check the status and result of an NVIDIA NIM inference job

Evaluate and rank agent results by metric or LLM judge for an AgentHub session.

Drafts a Motion for Judgment Notwithstanding the Verdict (JNOV) under FRCP 50(b) or state equivalents, with alternative new-trial request. Builds element-by-element evidentiary insufficiency arguments using transcript citations and preserves the appellate record. Use when drafting JNOV motions, post-trial motions, renewed judgment as a matter of law, or challenging jury verdicts for insufficient evidence.

Structured visual QA verdict for screenshot-to-reference comparisons

Produces a machine-validated, auditable JSON decision record with assumptions, risks, recommendation, and review gating for high-stakes decisions.

Analyse experiment results, run validity checks (SRM, exposure parity, data integrity, novelty/primacy), interpret causally, make a ship/iterate/kill decision against the pre-declared rule, and append to cumulative learnings. Forces honest readouts — strips significance claims from underpowered or peek-violating tests; never lets directional results masquerade as causal wins. Load when results exist, or when the user says "read out this experiment", "analyse the test", "did the test win", "interpret the results", "what did we learn", "ship or kill", or when the experimentation orchestrator routes here.

Verify a Jira issue against the current repository state, then post a Jira comment with a PASS, PARTIAL, FAIL, or BLOCKED verdict. Works in two modes: (a) feature-branch mode, comparing a checked-out branch to its base ref; or (b) main/trunk mode, verifying that an issue is already implemented on the current default branch (typical for stories that have already merged). Use when a user asks whether a branch or merged change completes a Jira ticket, or needs a Jira-visible verification comment.

Evaluates an AI judge system's pairwise ranking accuracy on generated commit messages against a heuristic ground truth from five automatic text metrics, using the MCMD dataset.

Reconcile canonical declarative documents under docs/ with verified implementation discoveries after a FULLY_IMPLEMENTED moonspec-verify verdict. Use when an orchestration run must decide whether verified discoveries show the owning canonical document is impossible, unclear, or inconsistent, apply the smallest correct doc update, or escalate ambiguous authority conflicts and deliberate divergences instead of editing.

Create and maintain lightweight Architecture Decision Records as agent-readable decision memory — what was decided, why, and which alternatives were rejected. Use when "record this decision", "set up ADRs", "the agent keeps suggesting Y again". Docs vs code drift → plan-docs-sync. Session state → handoff.

Produces auditable, schema-valid JSON decision records with assumptions, risks, recommendations, and review gating for high-stakes choices.

ATA RAI Remnant Ablation / Adjuvant Therapy Decision Tool

Drafts U.S. internal legal memoranda using IRAC structure to analyze issues, synthesize authority, assess risks, and recommend strategy. Use when asked to draft a research memo, internal memo, issue analysis, case strategy memo, or any IRAC-based legal analysis.

Evaluate and rank agent results by metric or LLM judge for an AgentHub session. Use when the user runs /hub:eval or asks to score, compare, or pick a winner among completed AgentHub agents.

Evaluate and rank agent results by metric or LLM judge for an AgentHub session.

Decide

Perform a non-destructive post-implementation verification gate validating the implementation against spec.md, plan.md, tasks.md, and constitution.md.

PR Fleet

Prioritize product work with explicit criteria, scoring, tradeoffs, and decision rationale.

Processes a PR's review comments (Copilot or human), classifies them, applies the fixes that belong, replies to each thread, and pushes. Use whenever the user says "address the review", "fix what Copilot said", "handle the comments", mentions Copilot already reviewed a PR, or passes a PR number/URL with pending reviews — or in Spanish "revisá los comentarios", "arreglá lo que dijo Copilot", "atendé el review".

Draft privacy policies, terms of service, disclaimers, and legal notices with GDPR, CCPA, and other regulatory compliance.

Evaluate and rank agent results by metric or LLM judge for an AgentHub session. Use when the user runs /hub:eval or asks to score, compare, or pick a winner among completed AgentHub agents.