Agent Skills by Blackeye
Showing 28 vetted skills indexed across 1 GitHub repositories.
cancel
Cancel the active OMK workflow and clean up its state.
swarm
Route swarm invocations to the Team skill's staged pipeline.
autoresearch
Orchestrate a persistent, validator-gated research loop with state storage.
ralph-init
Converts project descriptions into a structured requirements document with progress ledger.
trace
Visualize agent flow traces and summaries to diagnose execution paths.
deepsearch
Search codebases for concepts, functions, or patterns across multiple languages.
skill
Manage local oh-my-kimi skills across user and project scopes.
deep-interview
Clarify user intent and non-goals through a structured Socratic interview process.
web-clone
Clone live websites by extracting layout, visuals, and interactions with Playwright MCP.
ultraqa
Orchestrate tests, builds, linting, and type checks across iterative QA cycles.
ultrawork
Coordinate parallel execution of independent tasks across agent tiers.
doctor
Diagnose and fix oh-my-kimi installation issues across local environments.
ralph
Execute iterative tasks with mandatory verification before completion.
wiki
Store and retrieve project knowledge in a Markdown wiki under .omk/wiki with cross-reference links and searchable pages.
git-master
Coordinate Git workflows for atomic commits, safe rebases, and clean history.
ai-slop-cleaner
Identify and remove AI-generated slop in changed files using a regression-tests-first cleanup workflow.
team
Coordinate multiple Kimi agents to execute tasks in parallel.
review
Route plan reviews through Critic and return APPROVED, REVISE, or REJECT verdicts.
visual-verdict
Compare UI screenshots against reference images and output a JSON verdict.
ralplan
Design and validate software implementation plans with explicit approval workflows.
worker
Coordinate worker lifecycle and task flow in OMK teams.
ecomode
Route tasks to the lowest viable agent tier and prune context when budgets tighten.
tdd
Enforce Test-Driven Development by requiring failing tests before production code.
autopilot
Automate end-to-end software development from idea to working code.