Compare an agent's reported work with authoritative before-and-after evidence to recommend whether progress is real, stale, blocked, or needs review. Use for agent supervision; do not use to mark work complete or grant approval.
Compare an agent's reported work with authoritative before-and-after evidence to recommend whether progress is real, stale, blocked, or needs review. Use for agent supervision; do not use to mark work complete or grant approval.
Classify supplied content into a caller-approved finite set of destinations and return a confidence-aware routing recommendation. Use for support triage, inbox tagging, document queues, or analytics schema-domain selection when Relay's structured-decision tool is available; do not use for open-ended answers or unapproved actions.
Classify supplied content into a caller-approved finite set of destinations and return a confidence-aware routing recommendation. Use for support triage, inbox tagging, document queues, or analytics schema-domain selection when Relay's structured-decision tool is available; do not use for open-ended answers or unapproved actions.
Exports skills for OpenAI, Anthropic, Gemini, Copilot, Cursor, and .agents-compatible clients.
Checks imported skills for source repo, commit pin, license, checksums, maintainer status, and trust.
Find, evaluate, and activate version-pinned procedural skills from Relay without waiting for a human to name one. Use at the beginning of a substantive task, when the goal materially changes, when the work enters an unfamiliar or specialized domain, or when a reusable procedure could improve accuracy or safety. Treat no-match as a valid result and surface relevant suggestions when policy does not allow automatic activation.
Reviews skill trigger quality, scope clarity, output contract, examples, resource structure, and tests.
Screen factual claims against supplied, versioned source passages and produce a citation-preserving review register. Use for research, reports, and agent output verification; do not treat a model score as proof or search for missing evidence without authorization.
Screen factual claims against supplied, versioned source passages and produce a citation-preserving review register. Use for research, reports, and agent output verification; do not treat a model score as proof or search for missing evidence without authorization.
Use when the user wants to know what will break if they change something, or needs safety analysis before editing code. Examples: \"Is it safe to change X?\", \"What depends on this?\", \"What will break?\"
Review code changes with GitNexus from a GitHub PR URL or number, a branch/ref or commit range, or local staged, unstaged, and untracked changes. Use when the user asks for a code review, merge-risk assessment, regression hunt, missing-test analysis, or a verdict on whether a PR, branch, commit range, or local diff is safe.
Recommend an approved analytics data domain for a question using a versioned domain catalog. Use before query planning when several authorized domains could fit; do not use to reveal schemas, write SQL, or grant data access.
Recommend an approved analytics data domain for a question using a versioned domain catalog. Use before query planning when several authorized domains could fit; do not use to reveal schemas, write SQL, or grant data access.
Turn a desired outcome or an explicit sequence of Agent Skills into one reusable master skill and, when execution needs a governed graph, a Relay workflow. Use when a recurring task requires several skills, repeated passes, typed handoffs, gates, agents, subworkflows, or a single callable front door; do not use merely to run one existing skill once.
Design and calibrate a typed AI decision before adding it to a Relay workflow. Use when a workflow needs a bounded choice, score, or yes/no gate with a measured fallback; do not use for open-ended generation.
Design and calibrate a typed AI decision before adding it to a Relay workflow. Use when a workflow needs a bounded choice, score, or yes/no gate with a measured fallback; do not use for open-ended generation.