agent-os

Plugin agent-os from marketplace agent-os-local in Elouderb/Claude-Delegation-Plugin. Commands read from the repo's manifest — 1 month ago.

Skills (24)

• architecture-research

  Preload for the research-planner agent; use to investigate unfamiliar architecture, libraries, alternatives, or unresolved requirements before coding.

• browser-verification

  Preload for the frontend-engineer and verification-engineer agents; drive the UI in a real browser via playwright to confirm a change actually works.

• card-workflow

  Use whenever creating, selecting, updating, logging, reviewing, or completing project-local Jira-style cards.

• code-graph-usage

  Use for repository discovery, symbol lookup, dependency analysis, caller analysis, bounded architecture context, and code-change impact analysis.

• database-graph-usage

  Use for live SQL Server schema discovery, table and column inspection, foreign-key paths, routine dependencies, and database impact analysis.

• doc-review

  Preload for the codebase-consultant agent; audit the repository's documentation against the actual code and structure, and report what is stale.

• execute-card

  Use proactively when a project card has clear acceptance criteria and is ready for implementation by the appropriate specialist subagent.

• graph-file-discovery

  Preload for the codebase-consultant agent; use the code graph to find the files relevant to a question, then read only those files.

• graph-query-discipline

  Use whenever querying either graph to keep results bounded, current, unambiguous, and suitable for downstream agents.

• implementation-handoff

  Preload for the implementer and complex-implementer agents; use when returning completed implementation work to the lead and reviewer.

• independent-code-review

  Preload for the code-reviewer agent; use to review a diff independently against its card, graph context, and project standards.

• library-docs

  Preload for code-editing and research agents; before writing or debugging code against an unfamiliar library, framework, SDK, CLI, or cloud API, fetch current docs via context7.

• lsp-diagnostics

  Preload for code-editing agents; after editing source, use the LSP tool to pull real diagnostics for the language before declaring the change done.

• migration-safety

  Preload for the database-engineer agent; use for schema changes and migrations requiring live dependency inspection and rollback-aware planning.

• plan-feature

  Use proactively when a non-trivial feature, refactor, bug fix, or project change should be decomposed into project-local cards before implementation.

• requirements-to-cards

  Preload for the research-planner agent; use to convert ambiguous or broad requirements into executable, dependency-aware cards.

• review-and-close

  Use after implementation to coordinate testing, independent review, graph refresh, card updates, and final completion.

• review-risk-triage

  Preload for the code-reviewer agent; use to prioritize review findings by severity, confidence, and affected graph surface.

• runtime-verification

  Preload for the verification-engineer agent; run the actual application and observe its behavior to confirm a change works, beyond what the test suite covers.

• scoped-implementation

  Preload for the implementer and complex-implementer agents; use to implement a clear card with minimal, complete, in-scope changes.

• search-to-graph

  Preload for the codebase-consultant agent; start from grep / tree / glob when you lack good symbol names, then pivot into the code graph for structure and reach.

• sql-routine-analysis

  Preload for the database-engineer agent; use for stored procedures, functions, queries, indexes, and routine-to-table or column dependencies.

• targeted-test-planning

  Preload for the test-engineer agent; use to select the smallest adequate test suite from the card, diff, callers, and dependency graph.

• test-execution-reporting

  Preload for the test-engineer agent; use to execute tests, isolate failures, and report reproducible results without hiding uncertainty.

Subagents (10)

• code-reviewer

  Use automatically after an implementation agent finishes. Reviews the diff against the card, architecture, repository graph, database graph, and acceptance criteria.

• codebase-consultant

  Read-only repository investigator. Delegate questions about the codebase to it — "how does X work?", "where is Y handled?", "what depends on Z?", "what would break if I change this?", "are these docs accurate?" — and it does the searching and file reading in ITS context, returning a tight, sourced a

• complex-implementer

  Use for difficult, complex, or architecturally significant implementation — changes with repo-wide or cross-module implications, large or risky refactors, new subsystems, or anything where getting the design and blast radius right matters more than speed. Prefer this over `implementer` whenever a ch

• database-engineer

  Use for schema changes, migrations, SQL queries, stored procedures, functions, indexes, foreign keys, and any task whose impact crosses database objects.

• frontend-engineer

  Use for user-interface work — building or refining web UI, components, styling, layout, and client-side behavior. It implements the change, pulls LSP diagnostics, and verifies the result in a real browser. Prefer this over `implementer` whenever the task is primarily about the UI / frontend.

• implementer

  Use proactively to implement a clearly scoped card when architecture and acceptance criteria are already defined. Appropriate for features, bug fixes, refactors, integrations, and documentation changes. Do not use for unresolved architecture decisions.

• research-planner

  Use when requirements, architecture, external APIs, unfamiliar libraries, or implementation choices need investigation before coding begins.

• security-reviewer

  Use for an independent security review of a diff that touches authentication, authorization, secrets, input handling, output encoding/escaping, deserialization, file/network access, or third-party dependencies. Read-only; complements the general code-reviewer with a security-specific pass.

• test-engineer

  Use proactively to create, improve, or execute tests after implementation, especially for behavior changes, bug fixes, database work, APIs, and complex integrations.

• verification-engineer

  Use to confirm a change actually works by running the real application or driving it in a browser — beyond what the unit/integration test suite covers. Distinct from test-engineer (which writes and runs the test suite). Read-only.

Similar plugins