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/evidence-model.md
View on GitHub# Evidence and quality
An evidence record makes exactly one competency claim in one context, optionally project-free for a baseline. `task_id` may be null for a baseline observation; task-related records must reference their task. `fact` records what was observed, when, and supporting artifacts. `interpretation` separately records independence, signal direction, target level, strength and rationale. A successful test does not automatically prove the interpretation. Author and verification reviewer are attributable registered principals.
Verification statuses are `unverified` (not corroborated), `verified` (artifact inspected and observation corroborated), and `refuted` (artifact contradicts observation). A failed test can be verified evidence: verification concerns truth of the observation, not task success. Corrections supersede the record; never erase a contrary signal. Reviewer cannot be the learner whose behavior is evaluated. Team lead verifies formal evidence; peers may publish provisional unverified evidence but cannot certify it for formal evaluation.
Strong evidence requires inspected reproducible artifacts, a direct observation, meaningful stated scope, known assistance and an explicit connection to the competency. Moderate evidence has corroboration but narrower context; weak evidence is limited, indirect or uncertain. `strong` cannot accompany unverified/refuted verification or unknown assistance. Missing artifacts must be recorded as a gap, not substituted with a fabricated URI. Unknown assistance uses `known: false`, null highest level and no inferred independence; known assistance has integer highest level equal to the maximum event level, or 0 if independently observed work has no assistance events.
Task evidence includes the assistance events relevant to that competency and observation interval from the canonical task history. This is an immutable observation snapshot, not a second editable assistance ledger. Later-discovered assistance requires correction and reassessment of affected judgments. Each event names who provided help and what behavior it affected. Separate evidence can describe learner comprehension after someone else implemented a change.
A formal assessment/review may cite unverified or refuted records to explain uncertainty or contrary signals, but affirmative capability claims must use verified supporting evidence. A promotion cannot be supported solely by weak, unknown-assistance or self-reported material. Evaluate repetition, recency relative to technology/context change, diversity of independent situations, scope, independence and contrary evidence qualitatively. Repeated citations to the same artifact are not repeated demonstrations. Simulated incidents never establish real production operations experience. Fixture evidence is excluded from live judgments.
Published evidence cannot be deleted to improve a learner's history. Missing, inaccessible or changed artifact revisions make dependent judgments stale and require explicit review. This protocol provides provenance and review discipline, not cryptographic proof of learner authorship or an external credential.
## Baseline and gaps
Project-free baseline observations use project_id and task_id both null. They may contain real inspected learner response artifacts and receive normal verification; they do not select a repository. A project-specific prior-work baseline may instead use a curator-imported immutable project snapshot before selection, with explicit learner authorization; import alone grants no task or source execution authority. Task evidence must match its task's non-null project.
Artifact-free statements belong in profile self-report or learner notes until actual supporting material exists. A resolver report or disclosure transcript proves only the report/disclosure, not the unavailable work. Provisional `unverified` metadata means the named reviewer checked the report but could not corroborate it at checked_at; it is not a fabricated verification event. If no check occurred, return a proposal rather than a published evidence record. Availability and downstream staleness follow [judgment validity](judgment-validity.md).
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 |