JovaniPink
UserPortable agent skills for Codex and Claude, with clean-room workflows, provenance, security reviews, validation, and cross-client packaging.
Categories
Indexed Skills (47)
acceptance-evidence-ledger
Create or update an acceptance ledger for substantial work that needs explicit outcomes, current evidence, re-verification, and visible blockers or abandonments. Invoke explicitly when a user requests an inline or repository-owned completion contract; use focused testing, execution, or claim-verification skills for one-time requests.
plan-execution
Execute a named, approved implementation plan with checkpoints, validation, deviation tracking, and strict stopping boundaries. Invoke explicitly when the user authorizes implementation against an identified plan.
finish-development-branch
Reconcile a development branch and present evidence-based options for pull request, merge, retention, or cleanup. Invoke explicitly when implementation is complete enough to decide the branch's next state without assuming push, merge, deletion, or abandonment authority.
merge-conflict-reconciliation
Reconcile an active Git merge, rebase, or cherry-pick conflict by inspecting exact repository state, preserving both sides' intent, validating the result, and keeping abort available. Invoke explicitly for an identified conflicted worktree; staging, continuing, committing, and pushing require separate authority.
multi-agent-orchestration
Coordinate authorized parallel agents on independent bounded tasks with explicit ownership, evidence contracts, cost awareness, and final reconciliation. Invoke explicitly when parallel work can reduce latency without creating edit collisions or delegating irreversible decisions.
prototype-spike
Run a bounded prototype or technical spike to answer a named uncertainty with disposable or isolated work and explicit evidence. Invoke explicitly when the user authorizes implementation for learning but has not authorized production adoption, merge, or release.
request-code-review
Prepare an evidence-bounded code review packet and, only when separately authorized, send or publish the request. Invoke explicitly when a change is ready for reviewer attention and its scope, checks, risks, and unresolved decisions must be stated precisely.
agent-context-state-memory-design
Design or review how an agent separates prompt context, session events, scoped state, long-term memory, and artifacts. Use when identity, authority, freshness, retention, provenance, tenant isolation, poisoning, or deletion behavior could change what an agent knows or does; use context-reliability-review for a supplied context packet rather than the hosting architecture.
agent-protocol-interoperability-review
Review interoperability across agent, tool, and service protocols using exact versions, discovery records, schemas, identity, authorization, delegation, streaming, errors, retries, idempotency, cancellation, and trust zones. Use for A2A, MCP, or another agent-facing protocol boundary; use api-contract-compatibility-review for an ordinary API change without agent delegation semantics.
agent-tool-action-boundary-review
Review the boundary between an agent and its registered tools or actions, including identity, permissions, schemas, argument controls, network reach, confirmation, idempotency, replay protection, rollback, and audit evidence. Use before an agent can read sensitive data or cause an external effect; use application-security-review for broader application controls.
agentic-system-security-review
Review an agentic system across agent identity, models, instructions, tools, memory, retrieval, delegation, protocols, guardrails, outputs, supply chain, cost, revocation, and emergency stopping. Use for a read-only threat and control assessment of model-mediated actions; use application-security-review for conventional application controls and skill-security-review for a skill package.
google-adk-engineering-profile
Review or design a Google Agent Development Kit system using exact framework, language, model, session, tool, evaluation, and deployment evidence. Use for human-facing ADK architecture and version discovery; combine it with the focused agent security, evaluation, context, retrieval, and operational skills instead of treating framework defaults as proof of safety or readiness.
retrieval-grounding-quality-review
Review whether an agent retrieval and grounding system selects authorized, current, relevant evidence and produces claims that the evidence actually supports. Use for corpus authority, tenant filtering, retrieval metrics, citation support, abstention, poisoning, and injection; use source-grounded-research for a human research task and source-output-conformance-audit for deterministic source-to-output transformation.
agent-evaluation-design
Design an evaluation contract for an AI agent, model-assisted workflow, or prompt-driven system using representative cases, baselines, configurations, graders, error classes, thresholds, cost, latency, human review, and limitations. Use when the decision depends on measured AI behavior across cases; use test-strategy for ordinary software coverage and test-quality-review for an existing test suite.
context-reliability-review
Review the reliability of context assertions supplied to an AI system by tracing provenance, authority, freshness, effective and recorded time, permissions, conflicts, supersession, and revocation. Use when agent behavior depends on whether contextual claims are current and allowed; use authority-boundary-review for system-of-record architecture rather than assertion-level context.
source-output-conformance-audit
Audit whether exact source identity and expected values survive parsing or extraction, validation, persistence, and readback using source-cited oracles and mutation-sensitive tests. Use for end-to-end source-to-output fidelity claims; use test-quality-review for test-suite quality and authority-boundary-review for ownership architecture alone.
functional-motion-review
Review whether interface motion clarifies evidence, sequence, comparison, or reader actions. Use for animated editorial and data experiences; not for routine styling, implementing animation, or claiming engagement from visual polish.
alignment-interview
Establish a shared, decision-ready understanding before implementation by inspecting available evidence and asking one material question at a time. Use when goals, constraints, terminology, or success criteria are unclear; do not use to delay an already approved and sufficiently specified change.
accessibility-review
Review web or application accessibility using WCAG 2.2 success criteria, semantic structure, keyboard behavior, assistive-technology needs, visual presentation, and user impact. Use for evidence-based findings without claiming conformance from incomplete testing.
api-contract-compatibility-review
Review API, event, schema, and client changes for backward, forward, and rollout compatibility. Use when producers and consumers may upgrade independently or when a change can alter behavior, shape, timing, errors, or semantics.
application-security-review
Review application or service security against current threat evidence and verifiable control requirements. Use for read-only assessment of authentication, authorization, input handling, data protection, session, error, configuration, and abuse risks; do not use this skill to audit agent packages.
code-change-review
Review an exact code change for correctness, regressions, security, compatibility, test quality, and maintainability using the diff and repository evidence. Use before commit, pull request, merge, or release when the user wants findings rather than implementation.
data-migration-readiness
Review a proposed data or schema migration for authority, compatibility, sequencing, reconciliation, rollback, and operational evidence. Use before a migration or cutover; do not execute or ratify the migration.
dependency-supply-chain-review
Review software dependencies and build supply-chain evidence for provenance, integrity, maintenance, vulnerability, and compromise risk. Use for a read-only assessment of manifests, lockfiles, registries, build inputs, CI actions, artifacts, and update policy.
implementation-planning
Produce a decision-complete implementation plan grounded in repository and platform evidence. Use when a change needs concrete interfaces, ordered steps, tests, rollout controls, checkpoints, and stopping conditions before work begins.
module-interface-design
Design or review a module boundary, vocabulary, interface, and hidden implementation so callers face a smaller and more stable contract. Use when code is coupled, concepts leak across layers, responsibilities are unclear, or a redesign needs concrete options before implementation.
observability-design
Review or design observability for services and applications using logs, metrics, traces, alerts, and user-impact signals. Use when operators need to detect, explain, and respond to failures without confusing telemetry volume with operational understanding.
operational-readiness-review
Review whether a service or application is operationally ready for release and sustained ownership. Use for read-only assessment of objectives, dependencies, monitoring, alerts, capacity, failure handling, incidents, runbooks, recovery, and support boundaries.
performance-scalability-diagnosis
Diagnose latency, throughput, resource, contention, and scalability problems using measured evidence and competing hypotheses. Use when a system is slow, unstable under load, or approaching a capacity boundary; do not silently optimize code.
problem-framing
Frame a proposed feature, defect, workflow, or system change before implementation by identifying users, outcomes, evidence, constraints, unknowns, options, and measurable success. Use when a request is important but the problem or desired result is not yet decision-ready.
respond-to-code-review
Evaluate code review feedback against current code, tests, contracts, and authority before accepting, rejecting, or deferring it. Use when review comments contain claims or requested changes that require technical verification.
test-driven-change
Guide feature and defect work through red, green, and refactor evidence while preserving a clear exception path when test-first work is unsuitable. Use for behavior changes where an executable test can prove the defect or desired outcome before implementation.
test-quality-review
Review the quality of a test suite or change-specific tests for meaningful assertions, defect sensitivity, isolation, determinism, risk coverage, and maintenance cost. Use when existing gates pass but the strength of the tests themselves is uncertain.
test-strategy
Design risk-proportionate test coverage across unit, integration, contract, end-to-end, property, migration, and manual layers. Use when a feature, architecture change, or defect needs a justified coverage plan rather than only running existing checks.
worktree-isolation
Assess whether dirty or concurrent repository work needs Git worktree isolation and design a safe isolation approach. Use when work may overlap existing changes; require explicit authority before creating, moving, locking, repairing, or removing a worktree.
change-adoption-planning
Plan organizational or product change around affected people, behavior, capability, incentives, support, measurement, and feedback. Use when technical delivery alone will not produce the intended operating outcome.
cross-capability-dependency-mapping
Map dependencies across products, teams, data, platforms, decisions, and operating capabilities. Use when delivery or outcomes depend on multiple owners and hidden sequencing, authority, or failure paths must be made explicit.
data-authority-migration-ratification
Prepare a proposed data-authority and migration decision record covering owners, writers, readers, cutover, reconciliation, rollback, and acceptance. Use when decision owners need evidence to ratify or reject a migration contract; the skill cannot ratify it.
decision-governance-records
Create or review decision records that preserve context, options, evidence, ownership, status, consequences, and review triggers. Use when a proposal must be distinguished from a ratified, rejected, superseded, or unresolved decision.
iteration-postlaunch-learning
Plan and synthesize postlaunch learning from observed behavior, incidents, feedback, experiments, and outcome measures. Use after a release when the team must decide what to retain, change, investigate, or stop.
kpi-outcome-measurement
Design or review outcome measurement with precise definitions, units, populations, baselines, targets, data authority, and decision use. Use when a program or product needs measures that distinguish activity, output, outcome, and impact.
requirements-synthesis
Synthesize interviews, documents, issues, observations, and constraints into traceable requirements. Use when multiple sources must become a decision-ready scope without turning proposals or preferences into ratified facts.
stakeholder-technical-communication
Translate technical evidence for a named stakeholder decision without overstating certainty, hiding risk, or exposing inappropriate detail. Use for executive summaries, review packets, status updates, decision briefs, and cross-functional explanations.
value-attribution-evidence-maturity
Assess claimed value, benefit, savings, risk reduction, or outcome contribution against an explicit evidence-maturity model. Use when stakeholders need to distinguish observed results, estimates, attribution, contribution, and causal claims.
workshop-planning-synthesis
Plan a workshop around a concrete decision or learning outcome, then synthesize its evidence without manufacturing consensus. Use for agendas, facilitation structure, prework, participation, decisions, actions, and follow-up.
change-impact-analysis
Analyze the blast radius of a proposed or completed change across callers, contracts, storage, jobs, clients, security, rollout, and operations. Use before review, merge, or release when downstream effects may be missed; do not substitute speculation for traced dependencies.
example-authority-policy
Apply a fictional repository's local authority policy when reviewing example records. Use only inside this sanitized example to demonstrate private overlay placement.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.