Personal Workflow Skills

Plugin personal-workflow-skills in adamamer20/personal-workflow-skills. Commands read from the repo's manifest — 1 month ago.

Skills (16)

• abstraction-opportunity-audit

  Find and prioritize safe abstraction opportunities in codebases, including abstract base classes, protocols, strategy objects, and reusable mini-framework seams. Use when asked to reduce duplication, standardize behavior across concrete implementations, introduce extension points, or design a shared

• codex-thread-handoff

  Start or reconcile a fresh peer Codex task, recover an interrupted existing peer, or send exactly one bounded peer message. Use native task tools without retry loops, routine polling, recurring routing, or silent fallback to child agents.

• collect-evidence

  Perform a bounded read-only investigation over a repository, documents, logs, artifacts, or primary sources and return a compact evidence pack. Use when planning, review, or recovery would otherwise load a large amount of material into the main worker context. Do not use to implement, decide product

• dead-code-elimination-audit

  Identify and safely remove legacy dead code that has no real call path, no imports, and no runtime use evidence. Use when asked to clean up unused functions/classes/modules, shrink maintenance surface, reduce stale code risk, or perform aggressive dead-code reduction with safety checks in Python cod

• dedup-naming-audit

  Detect repeated code paths and confusingly similar names, then propose safe extraction and renaming plans with concrete file/line evidence. Use when asked to deduplicate code, reduce copy-paste logic, identify reusable components, or clarify naming in Python services and pipelines.

• define-visual-contract

  Turn product context, real content, existing renders, brand constraints, and visual references into one explicit visual direction, sentinel set, and observable rubric before visual implementation. Use for frontend/UI, slides, document generation, visual systems, or other work whose success depends o

• execute-milestone

  Implement exactly one decision-ready typed capsule, verify its observable gates, and return one durable typed result. Not for planning or broad refactors.

• fallback-upstream-audit

  Audit fallback logic across a repository, especially guessed defaults, inferred identifiers, silent coercions, downstream data cleanup, and consumer-side repair code. Use when a system is compensating for missing or dirty upstream data, when fallback-heavy code obscures the canonical source of truth

• indirect-attribute-access-audit

  Find and prioritize Python interface-drift risks caused by indirect attribute or method access. Use when auditing code for getattr, setattr, hasattr, dynamic method calls, fallback-heavy interface checks, weak duck typing, or when a user asks to enforce direct typed access through concrete classes,

• overabstraction-audit

  Detect and prioritize harmful implementation machinery in service and pipeline codebases. Use when architecture reviews, PR feedback, or refactors need evidence on pass-through layers, speculative protocols, reinvented capabilities, dependency overreach, unclear ownership, debug-hostile call chains,

• plan-work

  Turn substantial or architecturally uncertain work into one decision-ready typed capsule with explicit acceptance and ownership. Not for small localized changes.

• recover-milestone

  Diagnose and recover a milestone after repeated failure, context degradation, interruption, or a guarantee/architecture mismatch. Use the assigned workspace and preserve accepted scope. Default to finishing or replanning automatically; require the user only for genuinely material underdetermination

• review-work

  Independently review a fixed change or artifact through one explicitly named lens: outcome/spec, correctness/integration, standards/maintainability, visual quality, or security boundary. Use only when the plan names a review or promotion risk. Classify findings by evidence and promotion impact, and

• run-discovery-spike

  Answer one material feasibility, integration, performance, or architecture question with a bounded throwaway experiment. Use when inspection alone cannot distinguish viable approaches and choosing incorrectly would cause substantial rework. Do not use for production implementation or an open-ended p

• strong-typing-audit

  Audit and strengthen weak Python typing across a repository. Use when Codex needs to replace `dict[str, Any]`, broad `Any`/`object`, ad-hoc raw dictionaries, or unstable container types with precise dataclasses, validation models, generics, protocols, or narrowly scoped boundary types; or when tight

• workflow-control

  Invoke the packaged codex-flow controller for one typed capsule and report its durable status.

Hooks (4)

• PreToolUse

  python3 "$PLUGIN_ROOT/hooks/lifecycle.py"

• PostToolUse

  python3 "$PLUGIN_ROOT/hooks/lifecycle.py"

• Stop

  python3 "$PLUGIN_ROOT/hooks/lifecycle.py"

• SessionStart

  python3 "$PLUGIN_ROOT/hooks/lifecycle.py"

Similar plugins