Route complex analysis tasks through structured reasoning workflows

crossframe-suite is the umbrella router for the whole CrossFrame family and does exactly three things: decides which workflow a user's task belongs to, sequences the reading order across sibling skills, and outputs a short dispatch outline before handing off to the matched skill or skills. It never copies the body text of crossframe, crossframe-essay, or any other sibling skill into its own output. When a task triggers the core crossframe skill, the suite also factors crossframe/references/runtime-read-policy.md and crossframe/references/continuity-closure-map.md into its dispatch decision - specifically whether this task needs the full v5.0 continuity-bundle reading rather than a single concept card; v5.0 is the default canonical source structure, with v3.0 and v2.0 kept only as historical baselines. At the suite's own entry point it confirms exactly two things with the user: output mode and voice/role - article type is deliberately not chosen at that point. If the task is headed into crossframe-essay and the article layer hasn't been explicitly closed, the suite finishes proposition decomposition and the structural-insight groundwork first, and only crossframe-essay/templates/article-type-selection-dialog.md confirms the article type later, right before the essay itself is written. Default output for any CrossFrame content task routed through this suite is a "5.0 mixed long-form" result at output tier full-visible-v5-longform - a fully visible structural-insight groundwork plus a full-length essay - following the pipeline crossframe -> [needed sibling skills] -> crossframe-essay(full-visible-v5-longform) -> crossframe-review. This default exists because an article format is better suited to hand structural judgment, sibling-skill output, and v5.0 fidelity checks to an ordinary reader, not to inflate every answer's length. The fixed full interaction order is /crossframe-suite -> mode/role selector -> suite routing and sibling decomposition -> structural-insight groundwork -> article-type selector -> writing-technique reading -> article body -> quality-gate closure. Voice for the essay is decided jointly by role, output mode, and topic_sensitivity in references/output-mode-selector.md: an academic-expert or critical-reflector role defaults to a neutral analytical voice, a general-audience or futures-explorer role can enable an editorial undertone, and the user's explicit request for a warmer or reply-style voice overrides either default - a vulnerable topic still has to meet the person first even in the neutral register. crossframe-review is the quality gate, not the default final author: as long as the article layer isn't explicitly closed, the final deliverable must still contain both the structural-insight groundwork and the full article body, and passing the gate only appends a short verdict, never a review report as the sole output.

Use it only when the user explicitly names CrossFrame Suite, crossframe-suite, /crossframe-suite, or $crossframe-suite, as the umbrella router for a task that may need multiple sibling skills working in sequence - diagnosis plus public analysis, essay writing, and review, for example. It is the preferred entry point when: the user wants CrossFrame but hasn't named a specific sub-skill; the user only says "analyze this / what do you think / write something" and clearly wants a readable passage rather than a diagnosis printout; the task is an essay, commentary, opinion piece, or organizational-recovery piece; it's a reader Q&A, an editor reply, or a consultation-style response with a structural diagnosis behind it; raw material needs to become a case study before analysis; a reading/theory/article note needs comparison against CrossFrame; a proposition debate needs to become an essay, conclusion, or rebuttal; output needs to be generated first and then reviewed for whether the reasoning actually holds; or the user says these skills should be used together, in sequence, under one overall entry point. If the task is very single-purpose, go directly to the matching sibling skill instead of detouring through the router: pure structural diagnosis to crossframe, pure essay to crossframe-essay, pure review to crossframe-review, pure short replies to crossframe-dialogue, pure case-library work to crossframe-casebook, pure public-proposition evidence work to crossframe-public, pure organizational-repair memos to crossframe-org, pure concept teaching to crossframe-teach, pure proposition argumentation to crossframe-debate, and pure reading/theory research notes to crossframe-notebook.

On every trigger it reads references/output-mode-selector.md, references/workflow-routing-map.md, protocols/suite-dispatch-protocol.md, and templates/suite-reasoning-outline.md, then reads whichever sibling skill the routing map points to - basic structural judgment usually starts with crossframe/SKILL.md and crossframe/references/read-routing-map.md. Dispatch principles govern what happens next: complex tasks establish fact boundaries, scale windows, mechanism candidates, and judgment position through crossframe first; only the sibling skills this specific task needs are triggered, never the whole family at once; writing an article always comes after the diagnosis; and the suite passes only selection_state, workflow_state, voice_mode, and the article-layer on/off switch into the essay skill - it is barred from reading v5 source indexes itself, unpacking a continuity bundle itself, or generating its own v5-read-state-capsule, which only the crossframe core layer produces after actually completing the required companion-reading closure. Unless the user asks only "which skill should I use," the suite's final output always opens with a short dispatch outline - task type, output mode and role, workflow, required and conditional reading, continuity bundle, topic sensitivity, article voice, article type, output tier, what was not read, and the quality gate - before the matched sibling skill's normal output, and that outline must never run longer than the task itself.

Ships as the router skill inside the same repository as crossframe, crossframe-essay, crossframe-review, crossframe-dialogue, crossframe-casebook, crossframe-public, crossframe-org, crossframe-teach, crossframe-debate, and crossframe-notebook; it only ever loads the sibling skills a given task actually needs, never the whole family at once.

Anyone running a CrossFrame task that isn't a single clean diagnosis - someone who needs several CrossFrame skills chained together (diagnosis, then public-evidence work, then an essay, then review) and wants one entry point deciding the order instead of manually invoking each skill. It is not for a user who already knows exactly which single sibling skill they want; the suite explicitly tells them to call that skill directly instead.