mechanist

The claim agent of /auto. Runs the `/auto-claim` skill under two orthogonal axes — `BEHAVIOR_SOURCE` (given / given-validation / discovery) sets where the behavior comes from and whether it is validated; `MECHANISM` (given / discovery) sets who picks the mechanism method. `discovery` generates ranked, novelty-checked ideas; `given` / `given-validation` faithfully enumerate the behavior/claims in task.md. All combinations emit a refined proposal + experiment plan that downstream stages consume. Use this agent when the orchestrator wants the claim stage executed end-to-end with isolated context.

The experiment agent of /auto. Wraps the /auto-experiment skill, which folds mechanism-family routing inline before implementing, code-reviewing, and deploying the experiment suite. Supports two-step invocation — first call returns candidate families for the orchestrator's mini-prompt, second call (with chosen_family) commits the routing and runs the full pipeline.

The iteration agent of /auto. Runs the /auto-iteration-loop skill — an autonomous review loop that consumes /auto-verify's four-state output (PASS / FAIL / INCONCLUSIVE / ZERO_ELIGIBLE_VARIANTS) plus the orthogonal deferred bucket and routes each claim to the right back-edge (① variant-only fix / ② baseline-script fix / ③ claim-stage re-entry / ⓪ narrative-only) under a unified MAX_ITERATIONS budget with a MAX_CLAIM_REENTRIES sub-budget. Semantics are owned by skills/auto-iteration-loop/SKILL.md — this wrapper only forwards flags, relays the awaiting_upstream handoff, and reports the result.

The verify agent of /auto. Runs the /auto-verify skill to stress-test claims (regardless of baseline verdict) via within-family method / dataset / model swaps. Two mandatory integrity gates — Phase 2 per-claim baseline audit (runs for every target claim) and Phase 9 per-claim variant audit on Phase 3 step 0's top-K picked claims (each = /experiment-audit + /mechanism-audit combined by max_severity). Each claim ends in one of five states: PASS / FAIL / INCONCLUSIVE / ZERO_ELIGIBLE_VARIANTS / INTEGRITY_ONLY. Sonnet by default since the task is more mechanical than creative. Semantics are owned by skills/auto-verify/SKILL.md — this wrapper only forwards flags and reports the result.

Use when main results pass result-to-claim (claim_supported=yes or partial) and ablation studies are needed for paper submission. The external LLM reviewer (via llm-chat MCP) designs ablations from a reviewer's perspective, CC reviews feasibility and implements.

Analyze ML experiment results, compute statistics, generate comparison tables and insights. Use when user says "analyze results", "compare", or needs to interpret experimental data.

Search, download, and summarize academic papers from arXiv. Use when user says "search arxiv", "download paper", "fetch arxiv", "arxiv search", "get paper pdf", or wants to find and save papers from arXiv to the local paper library.

Workflow 1: Claim-stage pipeline, controlled by two orthogonal axes. BEHAVIOR_SOURCE selects the behavior stage: `given` (default; behavior taken from task.md and assumed to hold — no ideation, no novelty, no M0), `given-validation` (behavior taken from task.md but the experiment plan opens with an M0 phenomenon-validation gate), or `discovery` (mine a NEW behavior via /mechanism-behavior-discovery + full ideation: research-lit → idea-creator → novelty-check → impact-check → research-review → research-refine-pipeline, impact-first ranking; plan opens with M0). MECHANISM selects the mechanism stage: `discovery` (default; system routes — claim loads /mechanism-explore to shape direction) or `given` (user named the mechanism method/family in task.md — capture it and forward as CHOSEN_FAMILY, no routing). resource_fidelity:strict (exact models/datasets, no downscaling) is stamped iff BEHAVIOR_SOURCE=given AND MECHANISM=given (the reproduction combination). Always emits `idea-stage/IDEA_REPORT.md` + `refine-logs/{FINAL_PROPOSAL,EXPERIMENT_PLAN}.md`. Use when user says "idea discovery pipeline", "找idea全流程", "从零开始找方向", "复现 task.md", or wants the claim stage of /auto.

The single place for every data constraint an experiment must satisfy — dataset provenance (existing → adapted → constructed), clear train / validation / test splits, labels that reflect the target behavior, and the minimum data amount. Use whenever an experiment chooses, adapts, or constructs a dataset, defines splits, or sets a sample size — for phenomenon validation (M0), mechanism exploration, intervention, or tuning. Domain-general: no assumption about model family, modality, or task.