「ground-truth」の検索結果

Extract exactly what the person typed in a Claude Code session (script, no agent context spent), have a fresh agent write a faithful structured summary under it, and optionally have an adversarial agent judge whether the working agent is still on track. Use at a checkpoint, before a handoff, after a compaction, when a long session may have drifted from what the owner asked, or when told to 'summarize what I said' or 'check we're on track'.

日本語の概要は準備中です。原文の説明を表示しています。

Provides adversarial code comprehension for security research, mapping architecture, tracing data flows, and hunting vulnerability variants to build ground-truth understanding before or alongside static analysis.

日本語の概要は準備中です。原文の説明を表示しています。

Declare bankruptcy on a long-lived AI agent's accumulated memory — audit what it currently believes, separate ground truth from stale and wrong, purge deliberately, restate the truths that survive, and log what was lost. Use when an agent keeps acting on outdated facts, contradicts itself across sessions, 'remembers' things wrong, or after a reorg/pivot makes its worldview obsolete. Produces a belief audit, a keep/correct/purge ledger, a restated ground-truth file, and the bankruptcy record.

日本語の概要は準備中です。原文の説明を表示しています。

Interactive cell annotation and image QC for IMC/MIBI using napari, napari-imc, Mantis Viewer, and cytomapper, covering the pixels-to-cell-table bridge, overlaying masks to catch segmentation/spillover artifacts, inter-annotator variability as the accuracy ceiling, contrast-as-threshold, and building class-balanced ground-truth label sets. Use when manually labeling cells, generating training data for a classifier, QC-ing segmentation on the image, confirming clusters are spatially real, or choosing an annotation viewer.

日本語の概要は準備中です。原文の説明を表示しています。

Use when the user wants to run a full ad-campaign pipeline on top of Higgsfield Marketing Studio — 'create a campaign', 'build a content plan', 'run the content pipeline', 'generate 100 UGC videos', 'plan and schedule a launch', 'make a batch of ads from my product', or 'how much did this save vs traditional production'. Covers the 5-stage orchestration (Research → Plan → Generate → Publish → Report), the UGC-first 5-format campaign mix (UGC Entertainment, Street Interview, Unboxing, Product Review, ASMR), the even-split allocation math, button-driven onboarding, the per-batch generation gate, and the publish + cost-report tail (satellite). Defers all Marketing Studio API ground-truth (presets, params, hooks, avatars) to higgsfield-marketing-studio.

日本語の概要は準備中です。原文の説明を表示しています。

Run the weak-agent adversarial test harness against docx-cli. Spawns weak exercise agents (Haiku by default, Sonnet to probe, or a local agent harness's pre-produced runs) to perform real document tasks over six scenarios — five editing (MNDA form-fill + font fidelity, invoice table-edit/restructure + logo replace, résumé styling, contract redlining + commenting, contract finalize via accept/reject + comment reply/resolve) and one authoring (T. S. Eliot poetry journal: multi-column, verse, footnotes, links, figure) — renders every result with Word, has opus judge them against ground-truth rubrics, measures each exercise's tool economy, token cost, wall-clock, and correctness (from transcripts for Claude, the exercise.json ledger for the local harness), and synthesizes a prioritized ergonomics report. Use when the user says 'adversarial review', 'test docx-cli with weak agents', 'run the haiku harness', 'weak agent test', or wants to re-run yesterday's adversarial process.

日本語の概要は準備中です。原文の説明を表示しています。

Determine the TRUE cause of a failed retort run before attributing it. Ground-truth every failure (run its tests, read its agent logs, inspect its workspace) and classify it as an infrastructure false-fail, a genuine model miss, or an environment issue — never trust the gate verdict or a log signature alone. Use whenever a run is recorded failed (test_coverage=0 / gate fail), a pass rate looks low, or you're deciding whether a "failure" is real before reporting or proceeding.

日本語の概要は準備中です。原文の説明を表示しています。

Think and work like an expert Digital Pathology Scientist. Use when a task calls for Digital Pathology Scientist judgment. Reasons from whole-slide pixels, pathologist-ground-truth label levels, and stain/scanner batch effects through QuPath, CLAM/TIAToolbox MIL, Macenko/Vahadane normalization, and MI-CLAIM/TRIPOD+AI standards while treating patch-level leakage, scanner-ID shortcuts, attention-on-necrosis artifacts, and inter-pathologist-kappa ceilings as first-class failure modes.

日本語の概要は準備中です。原文の説明を表示しています。

Audit whether a paper's EVALUATION DESIGN actually measures what it claims and whether its reporting is complete — the validity layer family D (experiment-forensics) cannot reach. Three patterns: train/test leakage means the reported score may not measure generalization (HP-EVAL-LEAKAGE — adopts the Kapoor & Narayanan 8-type / 3-category leakage taxonomy; the illegitimate-proxy / sampling-bias / pretraining-contamination subtypes hand off as needs_external_check, naming but NEVER running Oren-2023 exchangeability / Shi-2023 Min-K% / Golchin-2023 Time-Travel / BIG-bench canary); a load-bearing LLM judge is conflicted (same model/family as a compared system) or unvalidated (no human-agreement, no bias control) (HP-JUDGE-VALIDITY); a declared condition/metric is dropped or switched to favor the method, or 'best' is chosen with no held-out set (HP-SELECTIVE-REPORTING). Verdict-bearing at L0/L1 from the DESCRIBED protocol — NOT repo-gated like experiment-forensics; L2 only CONFIRMS against split/preprocessing/result files. A fresh cross-model reviewer (gpt-5.6-sol xhigh, read-only, fresh thread per pass) PROPOSES findings, each span-anchored to a ledger claim_id; tools/adjudicate_findings.py DECIDES the verdict. Leakage and under-reporting are usually HONEST methodological errors — every finding describes a discrepancy to CHECK, never an accusation. An LLM generating GROUND-TRUTH labels is HP-FAKE-GT (experiment-forensics) — routed there, not here. Emits eval-design-forensics.findings.json; computes NO verdict. Detect-only. Triggers: "eval design audit", "evaluation validity", "train/test leakage", "data leakage", "is the score measuring generalization", "LLM judge bias", "is the judge validated", "selective reporting", "cherry-picked results", "评估设计审计", "评测有效性", "数据泄漏", "训练测试集泄漏", "裁判模型有没有验证", "选择性报告".

日本語の概要は準備中です。原文の説明を表示しています。

Use before committing, opening a PR, or reporting a task done — when you want ground-truth confirmation there are no conflict markers, live-looking credentials, debug leftovers, untracked TODOs, or oversized files, without spending a model turn re-reading every file yourself.

日本語の概要は準備中です。原文の説明を表示しています。

Use before publishing documentation, merging a docs PR, or reporting a Markdown-heavy repo's navigation as complete — when you want ground-truth confirmation that internal links, heading anchors, and directory links resolve, without opening every file yourself.

日本語の概要は準備中です。原文の説明を表示しています。

Report ground-truth Claude token consumption and estimated cost by parsing JSONL session files directly. Use when checking API spend, auditing token usage by project/session/model, generating daily/weekly/monthly cost reports, or diagnosing ccusage undercounting. Includes subagent files that ccusage ignores.

日本語の概要は準備中です。原文の説明を表示しています。

Scan the current market landscape and produce a ground-truth ecosystem snapshot. Chains /last30days for live research, validates with user, and writes brand/landscape.md with a Claims Blacklist that hard-gates all content generation. Use when: "landscape", "ecosystem", "market snapshot", "ground truth", "what's happening", "refresh landscape", "market trends", or before any content campaign to verify claims.

日本語の概要は準備中です。原文の説明を表示しています。

Experimental/manual harness canary procedure. It records benchmark manifests and scores but does not provide a fully automated runner.

日本語の概要は準備中です。原文の説明を表示しています。

Audit datasets, joins, labels, and ground truth before analysis or modeling. Use when data meaning, row grain, time semantics, source-of-truth reliability, or label construction may invalidate conclusions; not for general model evaluation after the evidence base is already trusted.

日本語の概要は準備中です。原文の説明を表示しています。

Interactive cell annotation and image QC for IMC/MIBI using napari, napari-imc, Mantis Viewer, and cytomapper, covering the pixels-to-cell-table bridge, overlaying masks to catch segmentation/spillover artifacts, inter-annotator variability as the accuracy ceiling, contrast-as-threshold, and building class-balanced ground-truth label sets. Use when manually labeling cells, generating training data for a classifier, QC-ing segmentation on the image, confirming clusters are spatially real, or choosing an annotation viewer.

日本語の概要は準備中です。原文の説明を表示しています。

Interactive cell annotation and image QC for IMC/MIBI using napari, napari-imc, Mantis Viewer, and cytomapper, covering the pixels-to-cell-table bridge, overlaying masks to catch segmentation/spillover artifacts, inter-annotator variability as the accuracy ceiling, contrast-as-threshold, and building class-balanced ground-truth label sets. Use when manually labeling cells, generating training data for a classifier, QC-ing segmentation on the image, confirming clusters are spatially real, or choosing an annotation viewer.

日本語の概要は準備中です。原文の説明を表示しています。

Run the full on-track check on a conversation, project, or keyword the owner names: find the sessions, extract exactly what he typed, dispatch independent judges who go to the live product as him and report what was missed, plan the remedy, and return one page — what he will see, what is missing, the plan. Use when told to check whether work is on track, to audit a project that has drifted, on 'run the on-track check on X', or when a long campaign is about to freeze, hand off, or claim done.

日本語の概要は準備中です。原文の説明を表示しています。