Skip to content

nanaagyei/noetherkin / skills/task-assignment

task-assignment

Verified

Define 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.

FileKindSize
SKILL.mdskill md3.7 KB
assets/proposal.mdmarkdown2.0 KB
references/FOUNDATION_V1.mdmarkdown5.4 KB
references/OPEN_QUESTIONS.mdmarkdown3.5 KB
references/PROJECT_CHARTER.mdmarkdown6.4 KB
references/apprenticeship-config.schema.jsonconfig3.0 KB
references/architectural-invariants.mdmarkdown2.3 KB
references/assessment.schema.jsonconfig5.0 KB
references/assistance-model.mdmarkdown4.0 KB
references/attention-advisory.schema.jsonconfig3.8 KB
references/bundle.jsonconfig12.0 KB
references/cli.mdmarkdown18.6 KB
references/competencies.yamlconfig37.7 KB
references/competency-model.mdmarkdown5.0 KB
references/competency-state.schema.jsonconfig3.4 KB
references/conformance.mdmarkdown22.0 KB
references/contract-README.mdmarkdown6.9 KB
references/contract-manager.mdmarkdown2.4 KB
references/contract-task-assignment.mdmarkdown3.2 KB
references/contract-team-lead.mdmarkdown2.8 KB
references/current-project.schema.jsonconfig2.3 KB
references/current-track.schema.jsonconfig1.5 KB
references/evidence-model.mdmarkdown4.1 KB
references/evidence.schema.jsonconfig10.3 KB
references/forge.schema.jsonconfig9.5 KB
references/index.mdmarkdown3.6 KB
references/judgment-validity.mdmarkdown6.6 KB
references/learner-profile.schema.jsonconfig1.5 KB
references/leveling-framework.mdmarkdown3.4 KB
references/levels.yamlconfig8.0 KB
references/overview.mdmarkdown2.9 KB
references/permissions-model.mdmarkdown6.7 KB
references/project.schema.jsonconfig9.4 KB
references/proposals.mdmarkdown4.2 KB
references/review-model.mdmarkdown7.2 KB
references/review.schema.jsonconfig13.0 KB
references/runtime.mdmarkdown7.8 KB
references/selection-model.mdmarkdown6.1 KB
references/simulation-lifecycle.mdmarkdown8.0 KB
references/state-model.mdmarkdown13.7 KB
references/task.schema.jsonconfig13.8 KB
references/track.schema.jsonconfig1.3 KB

references/judgment-validity.md

View on GitHub
# Judgment validity and reconciliation

Adopted ACP-01. Distinguish publication validity, present support and permission to advance. Historical records are checked against the complete committed snapshot and actor grants at publication. Subsequent retirement, artifact loss or changed evidence never rewrites that historical check. Missing canonical records, forged publication or broken receipt history are integrity errors requiring recovery, not ordinary staleness.

## Deterministic projection

1. Build immutable dependencies: evidence depends on its artifacts and assistance snapshot; assessments depend on cited evidence and, for design checkpoints, task/design revision; reviews depend on evidence, assessments and reviewed change; promotions additionally depend on `recommendation_review_id`, `readiness_assessment_id`, scope-agreement artifact and `previous_promotion_id`. All dependencies must precede or occur in the same ordered transaction before the dependent record. Cycles are invalid.
2. A superseded record is historical and ineligible for new support. Seed staleness in records citing it. Seed staleness in evidence/judgments whose cited immutable artifact cannot be resolved or whose bytes/revision mismatch. Task revision changes also invalidate applicability of old design/code/task approvals to current gates, without erasing their historical validity.
3. A newly disclosed assistance event or correction conservatively stales task evidence for each named competency with observation time at or after help occurrence, unless that snapshot already contains the effective event. Unknown occurrence must be recorded conservatively at task creation with uncertainty in the description. Help correction heads replace prior help only in the projection, never the ledger.
4. A new verified contrary evidence record conservatively stales older longitudinal findings for the same learner and competency until an explicit new synthesis cites and addresses it. No model decides silently that a contradiction is immaterial. The team lead's new longitudinal rationale must explain materiality, cite relevant older supporting and contrary observations and specify the resulting finding. Unverified allegations require investigation and cannot be converted into verified contrary facts by a reader.
5. Propagate staleness transitively through dependency edges until no new records are reached. An assessment with any stale finding is ineligible as a readiness packet. Completed task status stays historical; display stale acceptance/support alongside it rather than rewriting completion. Supersession links are correction relationships, not support edges; a correction does not inherit staleness solely by naming its predecessor.
6. Determine `last_awarded_level` as the highest historically valid PROMOTE target ever published, or E0 if none. Select unsuperseded promotion heads and find the unique supported adjacent chain from E0. Two supported successors for the same effective predecessor are invalid. Any unsupported historical award not replaced by a complete supported chain through that awarded level makes `standing: unresolved` and `effective_level: null`. Otherwise standing is current and effective level equals the chain tip. Non-promoting reviews never change the historical award. Stale unrelated evidence can coexist with current standing but must remain in `stale_record_ids` and cannot support a new decision.

Display both last awarded level and standing together. Never label the historical award as currently demonstrated when unresolved. Cache entries for stale aggregate findings are contested/null. Readers return a valid unresolved projection, not a guessed lower level or an unreadable workspace. Missing artifacts are reported with the exact unresolved reference; record a gap in learner notes or a truthful resolver-report artifact, never fabricate the missing original.

## Authorized repair path

Schema/reference/authorization validation permits honest new evidence, assistance disclosures, corrections, reassessments and reconciliation even when dependent historical decisions are stale. It rejects *new affirmative claims* using stale support. Reads, learner work and task repair remain possible; new upward promotion beyond the last historical award is blocked while standing is unresolved. Paused/archived mode restrictions still apply.

If an exact missing artifact is restored and there was no supersession or new contrary/help information, recomputation can clear that availability seed. A replacement artifact with different content cannot do so. Evidence corrections require a new same-observation, same-learner, same-project, same-task, same-competency evidence record; team lead inspects and publishes it. Never rewrite old dependencies to point to the correction.

For invalidated awards, reconcile from the earliest unsupported edge upward. Obtain a fresh longitudinal readiness assessment, manager recommendation and learner authorization under the normal pinned-input promotion protocol. The promotion reviewer publishes a replacement review with `supersedes` naming the current head for the same learner/from_level/target_level decision and `previous_promotion_id` naming the supported predecessor (null at E0). Use that predecessor's level as `from_level` during reconciliation even while global effective level is null. This is the sole exception to ordinary advancement requiring current standing. All independence, coverage and evidence requirements still apply.

PROMOTE restores that one edge; each old downstream decision must be explicitly superseded using fresh pinned inputs and the new predecessor ID, in ascending order. Old stale successors do not conflict with a supported replacement; two supported unsuperseded successors do. Partial restoration leaves standing unresolved until the chain reaches the last awarded level. NOT YET or INSUFFICIENT EVIDENCE keeps standing unresolved and records the needed next observation; a later retry supersedes that same correction head. No demotion, revocation or contested-award override is available in V1. These optional appeal policies require a future version, but ordinary correction and restoration are fully specified here.

Same-subject correction keys: evidence uses the observation identity described above; checkpoint assessment uses learner, task and competency set; longitudinal assessment uses learner and exactly the same competency set; code/task review uses kind and task; performance review uses learner, period and evaluation scope; promotion uses learner and level edge. Supersession must be acyclic and have one direct successor. Later observations and changed assessment scopes are new records, not corrections of an unrelated subject.
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: 471kUpdated Claude CodeCodexVerifiedInstall
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: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
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 availableStars: 322kUpdated Claude CodeCodexVerifiedInstall