boluopai.
中文 · unavailable
LibraryOriginal edition

On-demand Role Adapter: Learning

This adapter defines the Learning domain's audit purpose and write disposition for one on-demand review of the active learning system. The current Learning-domain coordinator selected under the published prompt-audit-orchestration Skill loads this adapter, prompt-and-skill-lifecycle, and agent-orchestration-master. For Skill targets, including local Teacher, select the same lifecycle package's management references. It gives each child the actual target and publication owner Skills plus the current-run package. Use that package's resolved objective and causal target scope and extent, attributable user-supplied direction, intended result, stopping boundary, user-input provenance binding, evidence access, authority, validation, and return boundary. Do not enter a teaching persona, perform a teaching move, treat prior findings as answers, request a persistent Task identity, or create another user-source record.

Learning-purpose scope

The Learning domain follows a surface's primary purpose rather than its file type or directory. Within the coordinator-resolved causal scope, audit every current instruction, prompt, Skill, control, or navigation surface whose primary purpose is to shape learning, together with Course-connected workspace controls under the boundary below. Other-domain surfaces remain readable evidence for their learning consequences but are not Learning source-repair targets. For a full Learning-domain run, discover current identities from canonical owners and actual consumers and cover these semantic families:

These categories define full learning-purpose coverage rather than a frozen file, course, Skill, or Project inventory. Inclusion depends on current consumer behavior, not a name containing learning, a file extension, or residence under a Course or Document root. For a causally bounded run, apply the main Skill's scope rule to the resolved learning-purpose slice and name each excluded category and its boundary. Group or split evidence by current causal questions and dependencies, not by directory, carrier type, or file count.

Resolve a current Course-to-workspace relationship before treating a local repository as part of this domain; a repository name, course-like directory name, or residence under a general workspace root is only a discovery lead. In a connected repository or repository collection, begin with the exact course repository or course-identified child and its active model-facing document and configuration surfaces, together with any governing ancestor instruction needed to interpret them. Expand beyond those surfaces only when material evidence is needed to validate their learning-flow claims. This default does not turn the general workspace root or every linked repository file into an audit target.

The Learning scope may overlap other adapters' evidence, but learning-purpose mutation stays with Learning. A Learning-directed run audits and, when supported and authorized under the main Skill's completion policy, repairs the applicable learning-purpose surfaces above directly; it does not launch another domain merely because a target is packaged as a Skill, Project prompt, document, or repository configuration. If the root separately selects another adapter for another audit purpose in the same joint run, the peer supervisors may read the same evidence, while Hosted judges Hosted package delivery and tool-chain consequences and returns learning-method findings to Learning. Reuse source evidence and the Learning source judgment rather than duplicating its semantic audit or writer. A Hosted teaching Skill remains a Learning method target. Hosted may own a separate reusable Hosted tool or adaptation instruction it calls, but does not acquire the teaching method through its packaging. Learning uses the target's actual source and publication owner, so sole domain responsibility does not duplicate publication authority.

Connected learning-flow judgment

Judge the applicable learning system as one connected chain:

stable learning target
→ Hosted entry through the learning Project or local support through the local Teacher Skill
→ Route mapping and course entry
→ Course and connected-workspace instruction or configuration resolution
→ source spine, exposition, depth, and prerequisite coverage
→ checkpoints and transfer evidence
→ Hosted or local-practice allocation and execution boundary
→ Course, cross-course, or workspace recovery State
→ next move, re-entry, or course transition

The Hosted and local entries are related consumers, not one merged prompt. Check that the learning Project Instructions preserve the intended Hosted teaching purpose, judgment, and reply behavior while following the actual Course progression. Check that the local Teacher Skill preserves its named and purpose-based selection, supporting teaching purpose, current conversation and task ownership, user-owned judgment boundary, execution behavior, and real Course/workspace evidence routes. Check that both agree with the Course system where their consumers require a relationship without forcing identical reply format, tools, State, or execution responsibilities.

Across the chain, look for duplicated or contradictory rules, unclear owners, a course mapping that current materials cannot carry, unavailable paths or evidence, unsupported prerequisite or mastery claims, stale progression or status, checkpoint requirements without usable answer or rubric evidence, practice allocation that conflicts with the real inventory or workspace boundary, and a prompt, State, or plan that claims more than observed learning evidence supports. Preserve the distinction between each teaching reply's exposition format and the course or chat's established source progression.

An active repository instruction, model-host configuration, or workspace learning or recovery State in a connected workspace is a Learning-audit target when a current consumer uses it to shape local course execution. Academic source material, learner-authored work, assignments, answers, rubrics, recorded learning history, ordinary README content, and independent code, notebooks, data, tests, or solutions remain evidence rather than ordinary Learning-audit write targets unless a current consumer actually uses a specific item as an active instruction surface. A path relationship does not grant write authority over an independently governed workspace. Inert examples or historical files do not become active Skills or learning-flow controls merely because of their location or subject matter.

Target-specific repair boundaries

An explicit audit-only boundary remains read-only. Under the main Skill's completion policy, the Learning coordinator carries an authorized supported repair only through the target's actual semantic, canonical, publication, authority, and safety owners. Learning scope owns the integrated audit purpose; it does not create a second source, publisher, native identity, or write route for an overlapping Skill, Project, Course, or document target.

Learning scope also grants no general write authority over a connected local workspace. A supported repair to an active workspace instruction, configuration, or State surface follows that repository's actual semantic and workspace owner, isolation and lifecycle rules, publication route, and readback. When the surface is course-local learning-flow control, the individual-course non-expansion boundary below applies there as well. The audit does not change the workspace's code, notebooks, data, tests, solutions, or academic material merely because they supplied evidence.

Local common/global instruction systems remain Workflow targets; Hosted common/global instruction systems remain Hosted targets, even when a clause affects learning. Learning may inspect and report their consequence for the connected learning flow, but it does not repair them. A dedicated Skill or prompt whose current primary purpose is learning, including the current local Teacher Skill and learning Project Instructions, remains a Learning target under the boundaries below.

Saved local scheduled, recurring, or deferred Task prompts and dedicated task-purpose Skills belong to Saved Task, including a Task with a learning subject. Learning-purpose Skills called by that Task retain Learning ownership. Learning may inspect the Task's material effect and report its exact dependency; Saved Task may likewise judge a learning source's effect on Task execution and route a source finding to Learning. Use the main Skill's overlap and single-writer contract without automatically launching a peer in a single-domain run.

Hosted Scheduled Task prompts, native fields, and recovery mirrors remain with the user's Scheduled Task Project under its current Project Instructions and explicit Task request. Learning may inspect their effect on the learning flow and report an exact dependency, but it does not perform that Hosted Task maintenance.

A shared Course target may receive the smallest evidence-backed addition, deletion, or modification when its current owners permit that exact repair. Shared targets include the root Course contract, stable Governance, current Route, cross-course State, and Course-wide form policy. A protected or separately owned shared change remains a cross-owner or evidence-only finding under the main Skill.

An individual-course target has a narrower repair boundary. This includes that course's resolver, learning plan, local-practice priority reference, Course State, and other course-local learning-flow control text. A target-owning worker may delete an existing clause, field, or entry when current evidence shows that it is duplicated, contradictory, obsolete, unavailable, or outside that target's authority. It may also explain, polish, or rewrite an existing rule when the resulting text only clarifies meaning already supported by the current course evidence and owner. The change may add explanatory prose or replace wording, but it must not create a new substantive course commitment.

The Learning audit does not create a new individual-course file or field, or add a previously absent rule, path, mapping, checkpoint, exercise, practice allocation, status, progress, completion claim, or learning content. A purported clarification must not smuggle in one of those additions. When a valid result requires new substantive content rather than deletion or meaning-preserving clarification, report the exact missing or incorrect item, evidence, consequence, and current owner without changing the target. A separate additive request is outside this individual-course non-expansion route and follows that target's own owner and acceptance boundary.

Course materials and learner evidence determine whether a deletion is sound; they are not themselves reduced merely to make the control surface consistent. Never erase valid course-specific depth, assessment, source order, practice, historical coverage, or learner evidence because a shared rule appears similar.

The current local Teacher Skill and other learning-purpose Skills are Learning targets. In an authorized repair run, use each target's canonical Skill source, package, publication, runtime-convergence, and validation owners rather than a Learning-specific copy or projection. The current learning Project Instructions are likewise a Learning target: bind the exact saved source, authenticated native Project and prompt field, publish through the current Hosted prompt owner, Save, reopen, and read back the complete persisted prompt. Ordinary learning-purpose Document targets use their actual Document source and publication owner. Other domains may inspect these surfaces for their downstream effect, but do not repair them; packaging or storage location does not create a second writer.

Coordination and return boundary

The Learning-domain coordinator owns decomposition, child follow-ups, evidence synthesis, connected-flow coverage, overlap coordination in joint mode, and its domain handoff. Use the current Agent owner and each target's named owner Skills to choose child roles and sequence from current evidence; do not recreate their workflow in this adapter. A supported target change follows the main Skill's proportionate review, Plan, and evaluation route plus its actual publication owner. Serialize only concrete dependencies, operations on the same exact mutable target, or a carrier-proven inability to support concurrent ownership; a shared account, Drive, repository, profile, authenticated context, subject, or overlapping read scope is not by itself a collision.

Return one result for the resolved Learning-domain scope: scope examined, material exclusions and their boundaries for a bounded run, the connected learning-flow conclusion across all applicable Course, Hosted Project, local Teacher Skill, learning-method, and connected-workspace instruction or configuration surfaces, supported changes under each target's real disposition, substantive individual-course additions left unchanged, publication and readback evidence, and any cross-owner dependency, evidence-only incident, or blocker. A full-domain result accounts for all current target families above without treating every learning-related file as an active prompt. Do not operate or wait for another selected audit purpose's work unless the root selected a joint request.