Bellman

LLM Wiki — persistent markdown knowledge base that compounds across sessions (Karpathy model)

Structured visual QA verdict for screenshot-to-reference comparisons

Verify that a change really works before you claim completion

Durable multi-goal workflow that persists plan/ledger artifacts under .omc/ultragoal and prints Claude /goal handoff text for the active session

Evidence-driven tracing lane that orchestrates competing tracer hypotheses in Claude built-in team mode

N coordinated agents on shared task list using Claude Code implicit agent teams

Turn a repeatable workflow from the current session into a reusable OMC skill draft

Manage local skills - list, add, remove, search, edit, setup wizard

Autonomous evolutionary code improvement engine with tournament selection

Investigate an open question and return grounded, sourced findings

Review reusable project knowledge and decide what belongs in project memory, notepad, or durable docs

Generic release assistant — analyzes repo release rules, caches them in .omc/RELEASE_RULE.md, then guides the release

Consensus planning entrypoint that auto-gates vague ralph/autopilot/team requests before execution

Self-referential loop until task completion with configurable verification reviewer

Worktree-first dev environment manager for issues, PRs, and features with optional tmux sessions

Install or refresh oh-my-claudecode for plugin, npm, and local-dev setups from the canonical setup flow

Evaluate finished work for defects, risk, and simplification before it ships

Strategic planning with optional interview workflow

Diagnose and fix oh-my-claudecode installation issues

Writing-time discipline for the prose an agent speaks to a human (replies, reports, pointers): a protected core never rewritten (code, commands, paths, errors, negators, numbers), filler dies, no narration, auto-clarity where terse misleads

YAGNI-ladder coding discipline for writing changes — existence-first, reuse before writing, dependency ladder, shortest correct diff, with non-negotiables that must never be minimized away

Loft the shape before cutting steel — answer a design question that prose cannot settle by building a throwaway artifact: a pure logic module in a clickable shell, or structurally different UI variants behind one route. The captain reacts to the artifact; the answer folds into the decision; the artifact never docks. Use when a design question stalls in words, when a navigator map carries a loft ticket, or when a spec discussion reaches "we would have to see it".

Shipyard's governed delivery pipeline — converge the mission, synthesize a durable spec, decompose vertical-slice tickets with blocking edges, run the frontier in parallel via team, close with verification, and report with a full decision log. Two entry gates — the yard gate (drydock audit) and the fog gate (an effort whose destination is unclear is routed to /ask-navigator before this pipeline starts). Humans own the checkpoints where there is no unique answer or the error cost is severe; agents continuously run everything repeatable and acceptable-by-evidence.

Shipyard's internal requirements intake for non-engineer contributors (support/ops) — turn a pasted chat log or verbal problem report into a five-section intent.md through numbered batch questioning with a completion gate, walk it through tracker review with harbor's four records, and hand the accepted intent to /launch as its mission brief. Produces decisions, not code.

Configure HUD display options (layout, presets, display elements)

Harbor intake for external work — the captain only handles unresolved decisions. Sweeps incoming issues and PRs, verifies every claim before disposition, reuses every decision already made, and hands the maintainer a docket whose pending items each carry one question with options, recommendation, impact and evidence. Agent-autonomous for facts and for actions covered by standing authorization; signed for every new judgment. Never merges.

Deterministic orchestration graph runtime - declarative DAG pipelines with journal-based crash recovery

Invoke parallel document-specialist agents for external web searches and documentation lookup

Carry an approved task through to working, verified code

Lay the keel of the shipyard harness in any repo — the 4-pillar shared environment (Context, Rules, Tools, Standards) across 5 surfaces (CLAUDE.md, skills, design-system, mcp/cli, shared context) so that every human and agent inherits the same design language and anyone can ship. Run once per repo; re-run with --check to audit drift.

Use the smallest visual when prose must carry structure—control flow, call depth, module ownership, or change shape: pseudocode, call tree, component/file tree, Mermaid diagram, or diff. Skip it when prose already answers the question.

Deep codebase initialization with hierarchical AGENTS.md documentation

Socratic deep interview with mathematical ambiguity gating before explicit execution approval

Diagnose the current OMC session or repo state using logs, traces, state, and focused reproduction

Configure notification integrations (Telegram, Discord, Slack) via natural language

Cancel any active OMC mode (autopilot, ralph, ultragoal, swarm, ultrapilot, pipeline, team) and clean up retired legacy state

Stateful single-mission improvement loop with strict evaluator contract, markdown decision logs, and max-runtime stop behavior

Full autonomous execution from idea to working code

Shipyard's navigator — chart a foggy effort (destination unclear, questions not yet stateable) into a map of decision tickets on the repo's issue tracker, then work the frontier one ticket per session until the way is clear, and hand the collapsed decisions to /launch as a mission brief. Wayfinding, not building: it produces decisions, never deliverables.

Process-first advisor routing for Claude, Codex, Gemini, Antigravity, Grok, or Cursor via `omc ask`, with artifact capture and no raw CLI assembly

Periodic architecture survey — walks the module graph and reports ranked deepening candidates (shallow modules, hypothetical seams, logic behind the wrong seam). Survey, not rescue: it finds candidates and hands them to the captain; it never refactors on its own.

Clean AI-generated code slop with a regression-safe, deletion-first workflow and optional reviewer-only mode

Writing-time discipline for documents agents consume (the five surfaces, specs, tickets, .omc/skills/) — every rule checkable and carrying a why, steps before reference, one meaning in one home, no restating what the environment already says. Mandatory at drydock seed generation and the launch C5 sediment pass; opt-in for any other agent-facing doc edit. The companion of minimal-code-discipline: that one disciplines code, this one disciplines papers.