esoul-skills

Plugin esoul-skills from marketplace esoul in eSoul-cz/skills. Commands read from the repo's manifest — today.

Skills (15)

• esoul-data-task-guard

  Use on eSoul data tasks (label data-migration, legacy imports, data fixes, reports, analyses) whenever data work reveals a needed change to application code, migrations, schema, seeders or factories. Do not change code or the data model beyond what the issue specifies (approved in-scope changes are

• esoul-freelo-task-authoring

  Prepare, preview and create well-specified eSoul tasks in one confirmed Freelo project, with duplicate search and approval before every write. Use to add, create, specify, split or revise an eSoul Freelo task, or turn a feature, bug, spike, data task or LTT into Freelo work.

• esoul-maintain-application-documentation

  Create, refresh, verify, review, and render application documentation as audience-specific Markdown and reproducible PDFs. Use for application user guides, developer or operator guides, documentation specifications, screenshot plans and redaction, documentation correctness or completeness reviews, b

• esoul-mockup-authoring

  Author, adapt, validate, locally preview, and publish ready-built interactive mockups for the eSoul mockups module. Use for desktop/mobile frames, addressable interaction states, image-only designs, and iterations from revision-bound feedback. Core workflow uses GitHub and the staff UI, not MCP.

• esoul-model-routing

  Choose the model, effort and delegation for an eSoul task so the weekly usage limit buys the most good output. Use when starting a task under esoul-workflow, when delegating to a subagent or a Zeron child chat, when the user asks to make work cheaper, faster or more thorough, or when usage runs out

• esoul-project-data

  Use whenever you save, back up, unpack or prepare project data (client files, photos, media, SQL dumps, backups, import files, exports) or a guide for colleagues. Data goes to assets/ via Git LFS, guides to docs/; never to gitignored storage/, *.nosync, tmp/ or Downloads.

• esoul-project-lifecycle

  Run an eSoul client project from first client materials to agent-ready work - bootstrap or retrofit the repository, ingest podklady and legacy evidence, keep questions, glossary, ADRs, guidelines, design brief and a sliced plan, batch slices into issue + Freelo pairs, and audit conformance. Use when

• esoul-tasking

  Prepare, preview and create eSoul work as a Freelo task or a linked GitHub issue + Freelo task, per the repository's split/no-split convention in CONTEXT.md. Use when invoked, when pairing GitHub issues with Freelo tasks, or under a tasking convention. Requires esoul-freelo-task-authoring; needs app

• esoul-uat

  Plan, verify, run and report UAT of eSoul apps, e-shops and websites as an end user in an isolated Chrome (functionality and design). Use for UAT, test scenarios, acceptance runs, "click through the app", responsive or design QA, or turning findings into bugs via esoul-tasking.

• esoul-workflow

  Default operating workflow for eSoul work on web applications (PHP/Laravel/Symfony, frontend), WordPress sites, and data analyses. Use at the start of any non-trivial eSoul task (build, change, fix, debug, review, analyse, or plan work) to size the task, load repository context, route each step to t

• final-pr-review

  Finish work by creating or updating a GitHub PR (single or native stacked), then looping on CI and CodeRabbit (or a pre-pr-review substitute when CodeRabbit is limited) until the current revision is approved. Use for final PR review, finish the PR, make CI green, or address CodeRabbit feedback. Does

• freelo

  Use when interacting with Freelo.io project management through the `freelo` CLI — managing projects, tasklists, tasks, subtasks, comments, time tracking, work reports, labels, notes, files, custom fields, templates, pinned items, workers, notifications, events, invoices, or searching in Freelo. Trig

• gitmoji-commit

  Split, stage, verify, and create focused Git commits using one Gitmoji shortcode per intention. Use whenever the user asks to commit, stage and commit, commit the current work, split work into commits, or write a commit message, and whenever an agent is about to create a normal agent-authored commit

• pre-pr-review

  Local review of the exact diff against the real base, CodeRabbit-style, in four verified passes with a credential scan. Use as the CodeRabbit stand-in (missing, rate-limited, out of credits), for a local or pre-push review without a PR, or when final-pr-review cannot use CodeRabbit. Not for finishin

• test-queue

  Shared queue (`testq`) that stops parallel agent sessions on one machine from running test suites, E2E, mutation tests and builds all at once and driving it into swap. Installs the CLI, a Claude Code hook that queues test commands, and a macOS menu bar widget. Use when installing or tuning the queue

Subagents (7)

• critical-reviewer

  Read-only correctness and security review on Opus at high effort. Use only when a change touches authorization, money, migrations or destructive data scripts, personal data, or concurrency, or when a reviewer finding needs a second, stronger opinion. Everything else goes to reviewer.

• implementer

  Implements a well-specified change (code and its tests) from a self-contained brief on Sonnet. Use for the build step of workflow size M and L slices after the orchestrator has settled the decisions, for fixing failing tests or review findings with a clear cause, and for mechanical multi-file edits.

• ops

  Mechanical repository and tracker chores on Haiku - commits through gitmoji-commit, gh and git queries, reading Freelo tasks, comments and work reports, and executing Freelo or GitHub writes that the user has already approved word for word, and task lifecycle moves (assignee, worker, tasklist, finis

• reviewer

  Read-only review on Sonnet with a lens and checklist given in the brief. Use for pre-pr-review maintainability and tests-and-contract lenses, correctness and security lenses on low-risk changes, UAT catalogue audits, and documentation checks. Returns verifiable findings only.

• scout

  Read-only code and log search on a cheap model. Use instead of Explore for "where is X", "how does X work", "what uses X", file inventories, and extracting facts from logs, transcripts or exports. Returns a short answer with file:line evidence, never file dumps.

• test-runner

  Runs tests, builds, linters, static analysis or a UAT sweep script on a cheap model and returns only the verdict and the failures, so long logs never enter the orchestrator's context. Use for every verification run that the orchestrator does not need to watch line by line.

• uat-runner

  Executes one esoul-uat chapter (or another scripted browser scenario set) in the UAT Chrome on Sonnet and records results with evidence. Use for the run step of esoul-uat so screenshots and page snapshots stay out of the orchestrator's context; failure analysis and bug triage stay with the orchestra

Hooks (2)

• SessionStart

  cat "${CLAUDE_PLUGIN_ROOT}/hooks/session-rules.md"

• PreToolUse

  command -v node >/dev/null 2>&1 && node "${CLAUDE_PLUGIN_ROOT}/hooks/agent-guard.mjs" || true

Similar plugins