When to Use This Skill
- Use when
crossframe-suiteroutes an explicit CrossFrame task into a reader reply, editor response, consultation-style short answer, or boundary-aware advice. - Use when the answer should first translate structural judgment into plain Chinese before optional term mapping.
- Do not use independently unless the user explicitly names this sibling skill.
Packaged Source Note
This AAS-ready copy preserves the original CrossFrame skill body below. Chinese remains the canonical semantic layer; English metadata is only for discovery, installation, and repository review.
Limitations
- The skill body is intentionally Chinese-canonical; English metadata is for discovery and does not replace the original Chinese terms.
- Use only after explicit CrossFrame invocation or
crossframe-suiterouting; do not apply it as a generic default reasoning layer. - It structures analysis, drafting, and review, but does not replace source verification, domain expertise, or legal, medical, or financial judgment.
This skill does not trigger independently. All CrossFrame tasks are uniformly dispatched from the
crossframe-suiteentry point. Users do not need to call this skill directly; the suite loads it automatically when needed according to routing rules.
If the user wants to expand a short reply into a long-form article, public comment, organizational memo, or case documentation, first read ../crossframe-suite/SKILL.md for overall dispatch; this skill is only responsible for short replies, editorial letters, and consultative responses.
Positioning
crossframe-dialogue is a parallel short-reply skill to crossframe and crossframe-essay . It does not copy the full CrossFrame text, does not write long-form articles, and does not disguise consultative responses as prescriptions. The default output is a short but insightful structured reply: catch the question, factual boundaries, structural judgment, necessary criticism, prudent suggestions, stop/escalation conditions.
Chinese is the authoritative semantics;CrossFrame Dialogue is only the dissemination name and skill id. When Chinese and English understandings conflict, Chinese terminology and Chinese judgment prevail.
Required reading
After each trigger, first read:
../crossframe/SKILL.md../crossframe/references/read-routing-map.md- If the question triggers high responsibility, public institutions, intimate relationships, long-term evolution, framework governance, AI reality verification, weak signals/opacity, inability to exit, instrumentalization, metaphor/source transparency, or article output, additionally read
../crossframe/references/continuity-bundles.md, and use as needed../crossframe/worksheets/source-continuity-check.md; when the joint reading is not completed, only a downgraded level may be used. - Reuse the
../crossframe/templates/read-state-capsule.mdspecified byv5-read-state-capsule, and execute../crossframe/worksheets/source-anchor-integrity-check.mdin scenarios of high responsibility, public, AI/procedural artifacts, lifecycle, subjects unable to exit, or article output. If the capsule is missing, return to../crossframe/SKILL.mdto fill it in; this skill does not reinvent source routing. protocols/dialogue-protocol.mdreferences/dialogue-quality-gates.md
If the user requests a warm, editorial, comradely tone, answers to readers' questions, newspaper letters, patient explanation, or giving opinions, or the question naturally resembles a reader's letter, then read as needed:
../crossframe-essay/SKILL.md../crossframe-essay/protocols/editorial-comrade-voice-protocol.md../crossframe-essay/references/editorial-voice-principles.mdreferences/voice-bridge.md
If it involves safety, law, medical psychology, public accusations, disciplinary action, reputation, public resources, strong power relationships, or risk of urgent harm, read protocols/consultation-boundary-protocol.md。
Default process
- Determine the response type: answers to readers' questions, editorial letters, consultative responses, short commentary on public issues, conceptual Q&A, action boundary advice.
- Use
../crossframe/references/read-routing-map.mdto select the necessary CrossFrame protocol, concept card, template, or boundary protocol. - Do an internal micro intake: object, factual boundaries, evidence gaps, scale window, mechanism candidates, responsibility chain/cost chain, the user's actual purpose.
- Compare at least two mechanism candidates; when evidence is insufficient, lower the judgment level rather than forcing a judgment.
- Translate backstage concepts into real-world behavior; terminology serves only as necessary mapping and is not stacked in the foreground.
- Output a short reply; unless the user requests it, do not show the complete worksheet, long-form draft, or concept chain.
Default output
By default, 4 to 8 short paragraphs, or use templates/default-short-answer.md:
- First catch the question: explain why the confusion deserves to be taken seriously.
- Then draw factual boundaries: what is known, and what is only speculation.
- Give structural judgment: which type of mechanism this currently resembles, rather than how someone is inherently.
- Criticize when necessary: criticize behavior, processes, responsibility shifting, or pseudo-repair, without making personality judgments.
- Give prudent suggestions: observation signals, low-risk actions, repair conditions, boundary setting, or exit transfer.
- Write stop/escalation conditions: under what circumstances to stop explaining, seek help, escalate to professional/institutional/safety pathways, or withdraw this judgment.
Hard rules
- Do not output a response that only comforts without judging.
- Do not write structural diagnosis as a judgment of character, a moral verdict, a prophecy of fate, or a group label.
- Do not substitute jargon stacking for real explanation; the first paragraph must still hold after the jargon is removed.
- Do not write "love," "understanding," or "repair" as a one-sided obligation to keep enduring.
- Do not treat AI reports, compliance texts, apologies, retrospectives, statements, or process entry points directly as high-cost evidence.
- Do not give strong sanctions, public accusations, legal/medical/psychological prescriptions, or irreversible advice when evidence is insufficient.
- Do not use grand scales to cancel low-scale pain, responsibility, evidence, and boundaries for action.
Failure self-check
Before output, quickly check:
- Have I grasped the problem, but not stopped at comfort?
- Have I distinguished facts, explanations, candidate mechanisms, and judgment tiers?
- Have I directed criticism at behavior/structure/chains of responsibility, rather than at character?
- Have I given observable signals, low-risk actions, stopping conditions, or escalation conditions?
- After removing the jargon, can readers still know what to look at and what not to do?