nanaagyei/noetherkin / skills/task-assignment
task-assignment
VerifiedDefine a bounded, testable Noetherkin assignment and its investigation prompt without revealing the solution. Use when a learner requests next work or a manager recommends new scope; execution gates belong to team-lead.
references/assistance-model.md
View on GitHub# Assistance levels
| Level | Meaning | Example boundary |
| --- | --- | --- |
| 0 | Independent | Learner chooses and executes an approach without task-specific help |
| 1 | Documentation/navigation | Point to relevant documentation or source location |
| 2 | Conceptual hint | Name a concept worth investigating |
| 3 | Investigation guidance | Suggest a diagnostic experiment without its answer |
| 4 | Pseudocode | Outline solution steps without executable implementation |
| 5 | Isolated example | Demonstrate a similar, separate problem |
| 6 | Partial implementation | Supply part of the actual change |
| 7 | Full implementation | Supply the actual solution |
Begin with learner understanding and a prediction, then reveal one useful hint. Escalate only after an attempt, a clear blocker, explicit request or urgent authorized work. Respect a demonstrated understanding without repeated gates. Record assistance for humans and tools where practical, including who, when, affected competencies and description. Unattributed help is unknown, never silently level 0. General prior study does not count as task-specific assistance; task-specific copied solutions do.
The config maximum is a default teaching ceiling, not authorization to modify source. A learner can explicitly request greater help; record the request with the event. All source changes still follow scope and repository policies. The highest level is a summary of help relevant to an evidence claim, not an average and not a punishment.
A level-7 implementation may still yield independent test-design evidence if the learner devised those tests without help. Split claims by competency, retaining the full task history. Completion and independence are distinct. Teach-back and failure-case questions can establish comprehension evidence only when the actual learner response is recorded and reviewed.
## Attribution and corrections
Each event has a stable `HELP-` UUID (fixture suffixes allowed), `at` for occurrence, `recorded_at` for disclosure, a registered `recorder`, and a `provider` with kind registered/external/unknown, nullable principal_id and a descriptive label. Registered providers require their retained principal ID; external/unknown providers require null principal_id. Unknown identity does not mean unknown magnitude: an inspected supplied implementation is level 7 even if its provider is anonymous. Event IDs are unique within the canonical workspace ledger; evidence snapshots reuse those IDs and exact event values.
Require at <= recorded_at <= publication; recorder authority is checked at publication. External reports are attributed as reports in the description and require corroboration before independent/strong evidence is claimed. A correction appends a new event with `supersedes` naming the prior event on the same task, never edits/deletes it, and cannot fork or cycle. The original recorder, learner or team lead may correct it; disputed claims stay explicit for team-lead verification. Late disclosure is permitted on terminal tasks without changing status or other fields. The validity projection stales affected evidence and dependent judgments until evidence is corrected and reassessed. Evidence snapshots include effective applicable event heads, not both old and corrected events.
For project-free baseline evidence, team lead records reported assistance directly in the immutable evidence snapshot; there is no fictitious task ledger. Correct it by evidence supersession.
## Independence calibration
A README pointer can preserve independent diagnosis; pointing out the faulty line may reveal the key diagnosis. Naming a concept can preserve independent causal reasoning; supplying the causal explanation cannot prove the learner discovered it. Suggesting an experiment can preserve independent interpretation; supplying the expected answer weakens that claim. An isolated example may teach a technique; copying an isomorphic solution does not demonstrate independent implementation. Reviewers explain which decisions remained the learner's. No assistance number mechanically sets a competency level or penalty.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |