「human review」の検索結果

Editing English prose that reads as machine-written. Use when reviewing or rewriting a draft for AI tells — a README, a changelog entry, a comment, a user-facing string, a review. French text is `humaniseur-fr`.

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。

Maintains four living project files (overview and progress log, decision log, RAID log, links) from the mail, channel posts, documents and meeting notes the user attaches or the agent can reach, returns each touched file complete in the chat for the user to save, and builds a weekly DRAFT status report from those files. Use when the user asks to "set up tracking for project X", "catch the project files up", "log this decision against the project", "add this risk to the RAID log", "save this link to the project" or "write this week's status report". Do not use for reviewing an existing RAID log for stale or ownerless entries, use raid-log-review instead; for an HTML status page, use project-dashboard-builder; for minutes of one meeting, use meeting-minutes-writer. Drafts for human review; never approves, authorises or signs off.

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。

Reviews the text and structure of a document, slide deck or web page against the accessibility rules the user supplies, or a default set covering heading structure, image alternative text, colour used as the only cue, link text, reading order, tables and language, and returns numbered findings with location, rule, severity and a ready-to-apply fix for the author, plus proposed alt text and a list of checks that need a tool or a person. Use when the user asks to "check this document for accessibility", "review the alt text in this deck", "is this page accessible", "run an accessibility check on these slides", "find accessibility issues before I publish" or "check my headings and link text". Do not use for brand or template compliance of slides, use branded-deck-builder instead. Drafts for human review; never approves, authorises or signs off.

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。

Compares one policy against one requirement set (standard, regulation, contract schedule or customer requirement) clause by clause and returns a draft gap review: each requirement element, the policy clause that appears to address it, an apparent match state, and every gap as a question with the clause reference on both sides. Never concludes that the policy complies; the owner decides. Use when the user asks to "gap our policy against the standard", "map this policy to the regulation clause by clause", "check the security policy against the customer's contract schedule", "compare our policy with the new edition" or "where does our policy fall short of the requirements". Do not use for mapping a requirement to the operating controls that implement it, use controls-gap-pack instead; for what a regulatory change means for the organisation, use regulatory-change-impact-note. Drafts for human review; never approves, authorises or signs off.

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。

Compares one clause (liability, termination, confidentiality or any other) across several contracts, templates or successive versions and returns a DRAFT comparison table: verbatim clause text per document, the clause split into elements with every difference from the baseline highlighted and categorised, and neutral questions for counsel. Never says which wording is better or enforceable. Use when the user asks to "compare the liability caps across these five supplier contracts", "line up the termination clause in these three drafts", "show how the confidentiality clause changed between versions", "contrast this clause with our template" or "benchmark the indemnity wording across our customer agreements". Do not use for a whole-contract review, use contract-review-pack instead; for screening one NDA against standard positions, use nda-triage. Drafts for human review; never approves, authorises or signs off.

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。

Drafts the instructions field of a declarative agent that orchestrates a set of custom skills, under a character cap (default 8,000), from the skills' names and descriptions: role and scope, one routing rule per skill, workflow order, gates written as workflow holds, global rules, output format and what the agent never does. Returns the text in a fenced block with its character count labelled counted or estimated, plus a skill inventory, routing table, gate table, rules trace and open questions for the owner. Use when the user asks to "write the instructions for my agent", "draft a system prompt that ties these skills together", "tighten these agent instructions to fit the cap", "how should the agent route between these skills" or "revise my agent's orchestration text". Do not use to review a single SKILL.md, use skill-file-reviewer instead. Drafts for human review; never approves, authorises or signs off.

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。

Drafts a blameless incident postmortem (summary, impact, timeline, detection and response, contributing factors as candidates for review, action items) from an incident channel export, ticket notes and any alert or change records, with UNKNOWN for every fact the inputs do not contain. Never assigns blame, names a root cause or re-rates severity. Use when the user asks to "write the postmortem for this incident", "build the incident timeline from the chat export", "draft the post-incident review" or "turn this ticket and channel log into an incident report". Do not use for a knowledge article from the resolved ticket, use knowledge-article-drafter instead; for project retrospectives, use lessons-learned-synthesis; for data-breach impact and notification questions, use data-incident-impact-brief. Drafts for human review; never approves, authorises or signs off.

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。

Designs a pre-launch evaluation plan for a declarative agent from its instructions text, skills, capabilities, knowledge sources and sample requests: a behaviour contract of testable statements quoted from the text, test prompts by scenario family and tier with setup, expected behaviour, pass and fail criteria and red flags, a red-flag catalogue, a blank scoring sheet, a coverage table and a go or no-go checklist the owner completes and decides. Use when the user asks to "write test cases for my agent", "how do I evaluate this agent before launch", "build a test plan for the agent", "what should I check before publishing this agent" or "give me a go no-go checklist". Do not use for finding injection or leakage exposures in the instructions, use agent-instructions-red-team instead; do not use for reviewing one SKILL.md, use skill-file-reviewer instead. Drafts for human review; never approves, authorises or signs off.

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。

Triages an incoming non-disclosure agreement against the organisation's written standard NDA positions and returns a DRAFT triage pack: a clause-by-clause deviations table with a severity band, a suggested response per clause quoted from the positions document, open questions and a governing-law flag, all for counsel to decide. Never gives legal advice or an acceptability call. Use when the user asks to "triage this NDA against our standard positions", "screen this confidentiality agreement before it goes to legal", "red-flag this mutual NDA", "first-pass the counterparty's NDA" or "is this NDA standard or does it need counsel". Do not use for a full review of an MSA, SOW, licence or any whole contract, use contract-review-pack instead; for one clause lined up across several documents, use clause-comparison-table. Drafts for human review; never approves, authorises or signs off.

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。

Reads a purchase order extract (spreadsheet, CSV or pasted table) and returns a DRAFT question sheet for the buyer: possible split orders, round amounts, amounts just under an approval threshold, first-seen or near-duplicate vendors, duplicate orders and missing or self approvals, each a neutral question naming the PO numbers, never a finding. Use when the user asks to "review these purchase orders", "screen the PO log", "spot anomalies in this spend extract", "sanity-check the PO register" or "prepare buyer questions before the audit". Do not use for an expense claim or travel request checked against policy, use expense-policy-precheck instead. Drafts for human review; never approves, authorises or signs off.

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。

Synthesises a batch of customer feedback (survey answers, reviews, ticket and chat comments) into a DRAFT themed report: coded themes with record counts and shares, anonymised verbatim quotes cited to record codes, contradictions and unverified claims, and the follow-up questions to answer next. Produces no sentiment score or percentage positive as fact; ratings appear only as distributions with n. Use when the user asks to "theme this feedback", "what are customers saying", "summarise these survey answers", "analyse these reviews", "find the top complaints in these comments" or "code this voice of the customer data". Do not use for exit interviews, use exit-interview-synthesis instead; for sorting a live queue and drafting replies, use ticket-triage-pack; for project retrospectives, use lessons-learned-synthesis. Drafts for human review; never approves, authorises or signs off.

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。

Reads a cash forecast (spreadsheet, model export, pasted table or the narrative that accompanies it) and returns a DRAFT assumptions sheet: every assumption the forecast rests on, stated or implied by the figures, with its value as read, its location in the model, the cash lines it drives, its stated source or "not stated", the evidence that would test it, an owner, and one neutral challenge question, so a reviewer can work through the assumptions one by one. Use when the user asks to "list the assumptions behind this cash forecast", "what is this cash flow forecast assuming", "prepare the challenge questions for the treasury review", "pull out the drivers of the 13-week cash forecast" or "build an assumptions register for the liquidity plan". Do not use for actuals against budget by line, use budget-variance-explainer instead; do not use for the case behind a capital request, use capex-request-pack instead. Drafts for human review; never approves, authorises or signs off.

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。

Maps every proposed P&ID object from an accepted enrichment draft to the generic family, catalogue class and required attributes the project's intelligent P&ID authoring tool would need, as a readiness table with UNKNOWN wherever the project catalogue extract is not provided, plus connectivity readiness per line and a numbered list of import blockers. Mapping only: it never writes to the tool, never produces an import file and never claims a class exists in the catalogue. Use when the user asks to "build the readiness map", "show how far the draft is from a sandbox import", "list what would block the import", "map the objects to the tool catalogue" or "check attribute readiness against the catalogue extract" after gate G3. Do not use for validating the discipline sections or assembling the review package, use validation-and-review-package instead. Drafts for human review; never approves, authorises or signs off.

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。

Proposes tag and number structures for the new lines and instruments in a P&ID draft according to the project's numbering procedures, checks every tag extracted from the PFD against the quoted procedure format, and lists cross-section identifier mismatches and the registry checks a human must run. Never invents a numbering convention, never alters a PFD tag and never assigns a sequence number the procedure reserves for the registry. Use when the user asks to "run the tagging and numbering section", "check the tags against the numbering procedure", "propose the tag structure for the new lines and instruments" or "list the registry checks before validation" after the process safety flags. Do not use for validating the four sections, use validation-and-review-package instead; do not use for instrument content, use instrumentation-and-control-enrichment. Drafts for human review; never approves, authorises or signs off.

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。

Prepares a DRAFT management-of-change intake record from a change description (email, meeting note, marked-up drawing or request form): what changes from what to what and where, the identifiers named, the candidate affected documents matched against the register, the disciplines to consult with the reason for each, and the numbered questions reviewers must answer, with UNKNOWN wherever the description is silent. Prepares the record only: never approves, classifies the change, decides replacement in kind or makes any hazard or safety judgement. Use when the user asks to "raise an MoC for this", "prepare the change request intake", "which documents does this change touch", "who needs to review this change" or "draft the MoC form from this email". Do not use for tracking who owes information to whom across disciplines, use interface-register-builder instead. Drafts for human review; never approves, authorises or signs off.

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。

Answer reviewers on your own pull requests in other people's repos: find threads where the ball is on your side, tell humans from bots, do exactly what was asked and reply with evidence. Use at the start of any GitHub session or on a new-review alert. Triggers: /pr-reply, /prr, answer the reviewers, who is waiting on us.

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。

Remove signs of AI-generated code from a codebase while preserving behavior. Use when editing or reviewing code to make it read like a human engineer on this team wrote it. Works through a test-guarded detect-modify-verify loop: pin down the functional contract and edge cases first, write or run tests as a safety net, then rewrite AI tells (redundant comments, emoji, verbose names, reflexive try/except, defensive null-guards, over-decomposition, speculative abstraction, dead code, non-idiomatic constructs, hallucinated dependencies) and re-verify after every change. The tests decide whether a defensive block is slop or a real safeguard.

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。

Keep a reviewable decision trail for long-running or unattended work: a TSV log with one row per decision (what, why, evidence, result). Local by default; commit it when a reviewer needs the trail to trust the result. Use for /show-me-your-work, autonomous or multi-phase runs, or work a human reviews after stepping away.

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。

Verify an open factory PR before human review - re-runs gates, checks the test proves the fix, checks scope and test tampering. Use when reviewing a factory PR, running the PR review routine, or asked to independently check a change.

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 workflow for summarizing work that still needs human judgment across configured pull requests, issues, feedback, errors, and delivery signals. Use for a human decision queue or bounded review.

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。

Conducts multi-axis code review. Use before merging any change. Use when reviewing code written by yourself, another agent, or a human. Use when you need to assess code quality across multiple dimensions before it enters the main branch.

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。

Influencer/creator marketing command desk (Busabase App-in-Skill) — the agent sweeps and fit-scores creator candidates, drafts outreach and briefs, and gates content before it ships; the human reviews creator cards, approves outreach/briefs/contracts, and tracks campaign ROI. Use when the user invokes $kelly-creators or /kelly-creators, mentions influencer or creator marketing, creator discovery, fit scoring, outreach DMs, campaign briefs, UGC, sponsorship or partnership deals, content review / FTC disclosure gating, or campaign ROI, or wants to review/approve agent-drafted creator outreach before it is sent through other channels.

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。

Personal finance anomaly review desk (Busabase App-in-Skill) that imports the three business tables — orders, invoices, payments/receipts (回款) — from CSV/JSON exports, flags anomalies with deterministic rules, and runs a human review queue with auditable decisions. Use when the user invokes $kelly-audit or /kelly-audit, mentions finance audit, order invoice payment reconciliation, 对账, 应收, 回款, 发票异常, receivables aging, overdue receivables, duplicate payments, missing invoices, amount mismatches, or anomaly review.

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。

Adversarial triage of open review findings on a PR. Every finding (Codex, Claude bot, Gemini) is a hostile hypothesis to REFUTE — fix only what is proven real, reachable, in the PR's scope, and above the round's severity bar. The bar rises every review round. Everything else is declined or filed as an issue. Use for 'fix the PR findings', '/fix', '/fix 123', or before any fix commit in a review loop.

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。