By cwensel
Kata Flight — skill-driven kata issue shipping with roborev review handling and optional RDR process binding. Ships the /kata-* and /roborev-* wrapper skills plus RDR binding skills for projects that use an external RDR process.
Use to verify a Kata Flight setup. Read-only health check for the .kata-flight seam, marker vars, engine layout, context root, optional RDR binding, and skill links. Trigger for diagnose Kata Flight, check Kata Flight setup, $kata-flight-doctor, or /kata-flight-doctor.
Bind the current consumer repository to Kata Flight.
Use to ship a batch of katas selected by ids, priority, parent, or label. Orders the wave, runs scope review by default, then invokes kata-ship per kata. Trigger for ship all/label/priority katas or $kata-flight.
Use for a read-only kata flow dashboard or stuck-state report. Shows lifecycle depth, inbox, batch depth, and illegal/stale states without mutating git or kata. Trigger for backlog status, stuck katas, or $kata-flow-ops.
Use to drain human-owned kata inbox states (`inbox:hold` and `inbox:needs-review`). Grounds each item, recommends a disposition, asks the human to choose/approve, then applies tracker-only state transitions. Trigger for review/unblock/clear/triage held or needs-review katas.
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Kata Flight is a Claude Code and Codex plugin that wraps kata issue
resolution, roborev review handling, and optional RDR-backed planning into a
repeatable shipping workflow.
It is intentionally only a skill repository. The runtime dependencies are the
kata and roborev CLIs, plus any agent harness that can install local skills.
RDR support is a binding to an RDR process, not an embedded copy of RDR.
kata-flight drives a batch: it orders a wave of katas, runs them through the
scope-review gate, and ships each one with kata-ship (resolve → roborev refine
→ fast-forward merge → close). Design-fork katas peel off to kind:rdr-seed and
enter the RDR track; RDR implementations land and flight their bug children back
through the same loop.
In the diagrams: a leading / marks an invokable skill (slash command);
purple rectangles are skills, amber parallelograms are kata label/issue
states, the hexagon is the external RDR engine's multi-stage flow, gray
dashed boxes are reference libraries (read by §anchor, not invokable), and the
blue stadium is the external roborev tool.
flowchart TD
init["/kata-flight-init<br/>(bind repo)"] -.setup.-> flight
doctor["/kata-flight-doctor<br/>(health check)"]
ops["/kata-flow-ops<br/>(dashboard / reaper)"]
kinbox["/kata-inbox<br/>(human inbox drain)"]
flight["/kata-flight<br/>batch orchestrator"]
review["/kata-scope-review<br/>review gate"]
kship["/kata-ship<br/>single-kata ship"]
resolve["/kata-resolve<br/>fix in worktree"]
roborev(["roborev refine / respond"])
flight -->|"per wave"| review
flight -->|"per kata"| kship
kship -->|"resolve"| resolve
kship -->|"refine"| roborev
review -->|"design fork"| seed[/"kind:rdr-seed<br/>(kata label)"/]
review -->|"held / needs human"| inbox[/"inbox:*<br/>(kata label)"/]
inbox --> kinbox
kinbox -->|"READY"| flight
kinbox -->|"TO-SEED"| seed
seed --> seedtri["/rdr-seed-triage"]
seedtri -->|"/rdr-seed (Stage 1)"| rdrflow{{"RDR design flow<br/>Stages 1-7 (external engine)"}}
rdrflow --> impltri["/rdr-implement-triage<br/>(Stage 8: build)"]
impltri --> roretri["/roborev-triage"]
impltri --> implland["/rdr-implement-land<br/>(land)"]
implland -->|"flight bug children"| flight
pship["/prompt-ship"] -->|"refine"| roborev
wsp["worktree-ship-pipeline"]:::lib -. read by .-> kship
wsp -. read by .-> pship
lland["lib-land-rdr"]:::lib -. read by .-> impltri
lland -. read by .-> implland
classDef skill fill:#EDE9FE,stroke:#7C3AED,color:#1E1B4B;
classDef lib fill:#F5F5F4,stroke:#A8A29E,color:#1C1917,stroke-dasharray:4 3;
classDef label fill:#FEF3C7,stroke:#D97706,color:#713F12;
classDef ext fill:#DBEAFE,stroke:#2563EB,color:#1E3A8A;
class init,doctor,ops,kinbox,flight,review,kship,resolve,seedtri,impltri,roretri,implland,pship skill;
class seed,inbox label;
class roborev,rdrflow ext;
See ARCHITECTURE.md for the skill-role table, the flight/review ship loop, and the rdr-seed peel-off in detail.
kata — local-first issue tracking. Install and docs:
https://katatracker.com/. Go install:
go install go.kenn.io/kata/cmd/kata@latest.roborev — local continuous code review for agents. Install and docs:
https://roborev.io/. Quick install:
curl -fsSL https://roborev.io/install.sh | bash, or
go install go.kenn.io/roborev/cmd/roborev@latest.rdr — Recommendation Decisioning Records, the external RDR process the
RDR-aware skills bind to. Optional: kata-only skills work without it. Source:
https://github.com/cwensel/rdr. Bind a checkout with --rdr-home or the
auto-detection described under Bind a Project.arc (Arcaneum) — optional corpus search for RDR/reference grounding, quotes,
and prior art. Not a runtime dependency; init records whether arc is present
but selects no corpora. Source: https://github.com/cwensel/arcaneum./plugin marketplace add cwensel/kata-flight
/plugin install kata-flight@kata-flight
codex plugin marketplace add /path/to/kata-flight
codex plugin add kata-flight@kata-flight
Plugin versions use SemVer in .claude-plugin/plugin.json and
.codex-plugin/plugin.json. Keep them in sync with ./bump.sh <patch|minor|major|X.Y.Z>, which also creates the release commit and vX.Y.Z
tag.
To confirm the install path works end-to-end in a clean container (no host
state touched), run the smoke test in test/install:
docker build -t kata-flight-installtest test/install
docker run --rm kata-flight-installtest
The canonical kata JSON helpers in scripts/kata-q.sh have
an offline test (captured fixtures, no kata daemon needed):
bash test/kata-json/run.sh
npx claudepluginhub cwensel/kata-flight --plugin kata-flightSemantic (vector) and full-text (keyword) search across indexed corpora via the arc CLI. Dual-indexes PDFs, markdown, and source code into Qdrant + MeiliSearch with AST-aware chunking, frontmatter, and git metadata. Use full-text for exact symbols/phrases and semantic for conceptual queries — both are first-class. Requires Bash(arc:*) permission.
Recommendation Decisioning Records — a repeatable, skill-driven flow from idea to spec-verified implementation. Ships the /rdr-* skills (init, doctor, seed, propose, refine, resolve, prelock, reconcile, finalize, implement, status, cluster-reconcile) plus the engine (stages/, prompts/, TEMPLATE.md) they read in place.
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).
Harness-native ECC plugin for engineering teams - 67 agents, 279 skills, 94 legacy command shims, reusable hooks, rules, MCP conventions, and operator workflows for Claude Code plus adjacent agent harnesses
Evidence-gated AI coding workflow: scan → analyze → plan → TDD → execute → fix → verify → review, powered by Codebase Memory MCP >= 0.9.0 with optional Serena LSP intelligence. Includes blast-radius planning, test/cycle gates, independent review, and Windows Git Bash hook auto-resolution.
Claude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Plugin-safe Claude Code distribution of Agentic Awesome Skills with 1,933 supported skills.