judgement-yields-navigation

Plugin judgement-yields-navigation from marketplace jyn-plugins in jyn514/dotfiles. Commands read from the repo's manifest — today.

Skills (47)

• sandbox-performance

  Diagnose and reduce startup, interactive-readiness, or cleanup latency in this repository's Codex sandbox and Pi launcher. Use for sandbox timing breakdowns, repeated startup benchmarks, proxy or TicGit latency, post-preparation delay, and proposals to remove or consolidate sandbox runtime work. Do

• architecture-design

  Design, critique, or choose a bounded software architecture: subsystem ownership, dependency direction, data/effect and lifecycle contracts, API shape, maintainability gain, and churn. Use for architecture documents, extraction candidates, unresolved architecture choices, or whether a split/refactor

• audit-spec-against-impl

  Compare what the current normative specification promises with implementation and tests, in both directions. Check for missing or partial implementation, undocumented intentional behavior, spec lag, and obsolete requirements. Use for bounded or whole-spec conformance audits and bidirectional drift.

• autonomous-work

  Organize an explicitly requested autonomous or unattended root task, or an active user-authorized root goal, through verified completion as an executive. Use for implementation, research, writing, planning, or review runs, including resumption. Delegated assignments do not inherit the parent's auton

• behavioral-testing

  Create and minimize behavioral tests of AI agents. Use when reproducing agent failures, building evaluations of agent decisions or actions, or reducing a behavioral reproducer while preserving its measured outcome. Covers repeatability and nearby controls. Not for ordinary application unit tests, so

• cleanup-triage

  Find, prioritize, and break cleanup work into tasks. Use after cleanup has been selected as the target, for a messy subsystem, cleanup roadmap, or opening and refining cleanup issues. For broad project opportunity discovery, use opportunity-scan first.

• commit-quality

  Write or review commit messages, split work into atomic commits, and prepare a finished change for commit. Use whenever creating commits, deciding commit boundaries, proposing messages, or judging commit quality.

• completion-audit

  Check whether a completed substantial owned change fulfills the original request, governing requirements, and acceptance conditions, using final-state evidence. Use after multi-component implementation, refactoring, or migration, or for an explicit completion-audit request. Map requirements to concr

• contradiction-debugging

  Investigate failures that seem impossible under the current model of a system. Use when observed behavior contradicts an invariant, source inspection says a reported failure cannot happen, or repeated successful probes leave a historical failure unexplained. Covers software, distributed systems, and

• creative-inquiry

  Explore a task's purpose, assumptions, and unexpected connections to discover possible approaches. Use for open-ended design, writing, research, or brainstorming; when purpose, assumptions, or framing need exploration; or when the user questions whether we are solving the right problem. Use work-sel

• delegation

  Participate in coordinated work with other agents. Use when taking an assigned part of a joint activity, actually splitting or coordinating work, handing off evidence or artifacts, or resolving ownership conflicts. Covers all subagents, including technical workers who do not spawn agents. Not for or

• dependency-review

  Review third-party dependencies and externally sourced tooling when a task proposes adding, updating, replacing, pinning, or vendoring a library, package, framework, plugin, action, CLI, toolchain, binary distribution, container image, or package manager. Do not use merely to inspect or use an uncha

• dockerfiles

  Create, edit, or review Dockerfiles and their Docker Bake build definitions. Use for container image packaging, multi-stage layouts, build performance, cache invalidation, runtime contents, and build or runtime file ownership. Not for generic application profiling, orchestration, or host Docker setu

• goal-to-issue-graph

  Turn a brief or broad goal into an actionable issue graph with observable root acceptance, bounded tickets, justified dependencies, and integration checks. Use when asked to break a goal into issues, create a dependency graph, or repair a graph whose tasks do not establish the promised outcome. Also

• grounded-futurism

  Analyze possible futures, map what would have to happen, or critique forecasts and scenarios through causal dependencies, alternative pathways, feedback, and evidence. Use for social, technological, economic, geopolitical, and institutional futures, especially disputed dates or probabilities. Not fo

• holistic-writing

  Write, substantially revise, or editorially review standalone prose, including documentation usability and correctness. Covers explanatory development, reading experience, voice, and factual fidelity. When a dedicated skill owns the requested writing or review, start there and follow its editorial-p

• human-review-packets

  Frame consequential human approvals and explanations that need substantial context, including review handoffs, unfamiliar trade-offs, and follow-ups asking for concrete consequences. Not for simple preferences, routine confirmations, requests for missing facts, routine factual definitions, technical

• implementation-plan

  Work out or review how to implement a given ticket in the current repository. Use when explicitly asked to plan a ticket's implementation or review its implementation plan. Identify the existing code path, required changes, verification, and material unresolved choices. A concrete requested change c

• issue-reporting

  Draft or revise upstream bug reports, support questions, and design discussions for maintainers. Use when asked to write an upstream issue or discussion, fill an issue template, or prepare a maintainer follow-up about a concrete workflow, limitation, or workaround. Not for internal design documents

• jj-conflict-resolution

  Resolve Jujutsu conflicts semantically by composing snapshots and diffs, never with ours/theirs or union heuristics. Use when interpreting or resolving actual or expected file conflicts, including conflicts from rebase, merge, or restack. Use jj-workflow for history operations and their mandatory pr

• jj-workflow

  Use when a task involves inspecting file or revision history, establishing change provenance or ownership, comparing revisions, or shaping history through splits, rebases, restacks, conflicts, or abandonment. Use also before changing an unexplained constraint, compatibility behavior, regression, or

• mutation-testing

  Set up, scope, run, and analyze mutation-testing campaigns using project-compatible tools or bounded hand-selected mutations. Use for mutation testing, surviving or equivalent mutants, questions about whether assertions are useful or tests catch plausible implementation mistakes, or diagnosing test

• new-tool-development

  Orchestrate the lifecycle and stage checks for a newly introduced, independently invoked tool, CLI, service, or reusable executable subsystem. Use when initiating or governing product brief, requirements, implementation-spec, or implementation handoff work; do not use for substantive requirements, a

• opportunity-scan

  Find the highest-leverage project improvement without anchoring on recent changes, current work, churn, or already-completed corrective campaigns. Use for broad codebase improvement, project opportunity scans, and deciding what to work on next; not for designing a solution within an already-selected

• pain-axis

  Analyze code history to learn from failures, repair sequences, and successful designs. Use before redesigning a legacy or failure-prone subsystem to identify recurring architectural pain and candidate invariants, or for open-ended research into historical defects, regressions, reverts, and preserved

• performance-investigation

  Diagnose unexpected execution delay, hangs, observed contention, and quota-constrained operations, or perform explicitly requested profiling and benchmarking. Use for commands, tests, builds, CI, startup, configuration reload, APIs, cloud queries, or remote acquisition when one of those conditions r

• pi-export-privacy-review

  Review Pi session exports for credentials, personal information, private project material, cross-session disclosure, and internal agent data. Use for sensitivity, privacy, secret, or publication-safety audits of Pi HTML or JSONL exports. Not for ordinary session summaries, arbitrary repository secre

• process-debugging

  Diagnose Linux native-process crashes, hangs, spins, and missing debugger symbols using GDB, core dumps, and bounded syscall traces. Use when a live process or crash artifact needs stack-level investigation, ptrace fails, or debuginfod does not load matching symbols. Not for ordinary application bug

• property-based-testing

  Write, review, and debug property-based tests — Hypothesis, fast-check, proptest, jqwik, rapid, and Echidna or Medusa for Solidity invariants. Use whenever tests should cover a whole input domain rather than a hand-picked list of examples: encode/decode and serialize/deserialize pairs, parsers, cano

• ratchet

  Define enforcement policy for an established invariant or check. Use when deciding how to block new violations while reducing an existing baseline, preserve violation identity or counts, bound exceptions and their expiry, or change an existing rule's enforced scope or guarantees. Do not load merely

• readme-writing

  Write, substantially revise, or review a repository, package, or tool-entrypoint README that introduces a project and helps readers decide whether or how to use it. Use for introductory READMEs, including those for libraries, integrations, books, and non-executable collections. Not for API-reference

• reorganize-docs

  Restructure existing project documentation or navigation, or review its organization and routes without editing. Use for information architecture, consolidation, splits, renames, archives, and linked-document navigation assessment. Not for choosing where a new paragraph or section belongs, prose-onl

• requirements-definition

  Author solution-neutral requirements from an approved product brief when explicitly asked to draft requirements. Use for requirements definition and traceability; do not use for product discovery, implementation specifications, architecture, or requirements review.

• rustc-diagnostic-gap-tests

  Survey Rust compiler and rustdoc diagnostics for weak or missing beginner-facing guidance, investigate helpers and implementation difficulty, or add non-duplicated compiletest UI cases and snapshots. Use in a rust-lang/rust checkout when asked to find diagnostic gaps, expand tests/ui or tests/rustdo

• session-closeout

  Close out a session. Use when asked to wrap up or end it; for forgotten-work or docs/skills/tooling reviews only when explicitly framed as ending the session. Not for ordinary commit requests, audits, or mid-session process reviews.

• session-title-curation

  Review, propose, and apply descriptive titles to past Pi sessions. Use when the user asks to name, rename, organize, or curate historic sessions, including sampling session history before changing titles.

• skill-authoring

  Create or evaluate portable agent skills, including existing or proposed discovery triggers, scope, workflows, and supporting package structure. Use for substantive skill changes, skill critique, replacing instructions with a skill route, changing when a caller invokes a skill, or diagnosing ambiguo

• spec-review

  Review what a supplied specification promises, against its own statements and governing requirements. Audit internal consistency, traceability, unsupported restrictions, and completeness appropriate to its declared stage. For an unqualified request to review a specification, start with its contract;

• ticgit

  How to use TicGit (`ti`) to track Git-native tickets in a local repository. Use when you need to interface with the tracker and the repository documents TicGit or an active workflow explicitly selects it as the fallback.

• tighten-prose

  Polish wording, diction, sentence rhythm, repetition, and local clarity in existing prose. Use for explicit local edits to sentences or short passages within an established explanation, including docs, reports, specifications, prompts, instructions, and skills. Do not use for whole-artifact editoria

• typst-diagrams

  Create or revise Typst diagrams, especially Flower spec `.typ` files. Add or clarify a figure, convert prose/dataflow, use Fletcher, or validate rendered layout.

• typst

  Write Typst (.typ) markup, expressions, and layout. Use when creating Typst documents, changing markup or scripting, or working with unfamiliar Typst syntax. Do not use for prose-only edits that preserve existing markup, such as wording changes or renaming a command inside an existing raw span. Does

• work-estimation

  Estimate or revise how long agent work will take, including task ETAs, remaining-time forecasts, and effort comparisons between approaches. Use when asked “how long,” when volunteering a work-duration estimate, or when checking past estimates against session history. Distinguish agent execution from

• work-selection

  Reassess the next action before adding another substantive prerequisite, workaround, or diagnostic round—even when earlier steps succeeded—or when a supposedly settled approach gains dependencies, results challenge its assumptions, or slow feedback lets supporting activity displace the requested out

• workstation-hardening

  Assess and improve a developer workstation or local host that stores or invokes sensitive credentials or capabilities. Use for broad workstation-hardening requests, same-uid process isolation, launcher secret handling, hardware-token workflows, PAM or sudo trade-offs, SSH key posture, Secure Boot, e

• zulip-api

  Read and analyze Rust Zulip conversations, using the local read-only proxy for private channels and the unauthenticated REST API for web-public channels. Use when a task cites a Rust Zulip channel, topic, message, or narrow URL; asks to save, find, follow, summarize, or quote Rust project discussion

• pi-link-tools

  Coordinate Pi sessions with link_set_identity, link_send, link_list, and link_compact. Use for link identity or group changes, direct messages, group clarifications, delivered-author replies, callbacks, status, and remote compaction. Covers manager-delivered worker replies and configured worker tool

Similar plugins