Witwave AI
Official@witwave-ai · United States of America
A platform for building and operating teams of autonomous AI agents with multi-LLM support.
Agent Skills by Witwave AI
Showing 36 vetted skills indexed across 1 GitHub repositories.
witwave-airelease
Cuts tagged stable or beta releases after verifying CI status and updating the changelog.
witwave-aicall-peer
Send prompts to peer A2A agents via JSON-RPC message/send and return their replies.
witwave-aiself-tidy
Grooms an agent's own memory namespace, peer awareness memos, and agent-card for daily drift.
witwave-aidiscover-peers
Discovers A2A agents in a Kubernetes namespace via service env vars and agent-card probes.
witwave-aigit-identity
Configures local git commit author identity on a fresh repository checkout.
witwave-aigit-sync-source
Clone or fast-forward a primary Git repository checkout to match its default branch.
witwave-aigit-push
Push existing local commits to the remote default branch with rebase-and-retry handling.
witwave-aifeature-work
Implements new features end-to-end with tiered risk gating, atomic commits, and delegated CI monitoring.
witwave-aigap-work
Find and fill missing functionality in the witwave repository across eleven gap-source categories.
witwave-aibug-work
Find and fix correctness bugs across repo sections using static analyzers at configurable depth.
witwave-airisk-work
Find and fix security, reliability, performance, observability, and maintainability risks in the witwave repository.
witwave-aicode-format
Runs language-specific formatters and linters across Python, Go, YAML, shell, and Helm source files.
witwave-aicode-cleanup
Orchestrates code formatting and comment verification, then delegates git pushes to a peer agent.
witwave-aicode-verify
Validates source-code comments against actual code across Python, Go, Helm, Dockerfiles, shell, and GitHub Actions.
witwave-aicode-document
Adds grounded comments and docstrings to undocumented Go, Python, Helm, Dockerfile, shell, and workflow code.
witwave-aiplatform-health
Inspects Kubernetes pods, operator status, and agent connectivity to report platform health.
witwave-aiask-peer-clarification
Asks a peer agent one read-only question to enrich a detected platform anomaly before escalation.
witwave-aidiscuss-questions
Scans GitHub Discussions General category and posts grounded replies to community questions.
witwave-aiteam-pulse
Scores team activity events and posts updates to GitHub Discussions.
witwave-aidiscuss-bugs
Investigate GitHub Discussions bug reports and route confirmed bugs to the team.
witwave-aidiscuss-comments
Scan GitHub Discussions for unanswered comments and post threaded replies via GraphQL.
witwave-aipost-discussion
Publishes a drafted post to a GitHub Discussion category via the gh api graphql mutation.
witwave-aigenerate-blog-entry
Generate source-backed Markdown blog posts from git history, agent memory, and CI state.
witwave-aipost-twitter
Publishes single posts and threads to X via the X API v2 with OAuth 1.0a authentication.
Frequently Asked Questions About Witwave AI
FAQPage SchemaWhat tasks can Witwave AI's skills perform?▼
The 36 skills cover autonomous bug and risk fixing, feature implementation with a 1-10 risk-tier ladder, gap analysis, code formatting via ruff/gofmt/prettier, docstring and Helm values documentation, markdown link and consistency validation, semver releases, git sync/push, platform health checks, staged team upgrades, and GitHub Discussions plus X/Twitter publishing.
Who are these skills designed for?▼
They target platform engineers and teams running autonomous agent fleets on Kubernetes. Named agent roles (zora, iris, piper, milo, kira, nova) handle dispatch, CI pushes, community engagement, roster audits, docs, and code hygiene, coordinated through A2A peer calls within a shared namespace.
How do the agents discover and call each other at runtime?▼
The discover-peers skill introspects Kubernetes-injected service env vars, probes each candidate's /.well-known/agent.json, and caches confirmed peers as memory entries. call-peer then sends prompts via JSON-RPC message/send with a fresh messageId per call, making delegation idempotent on the wire.
How are releases and code changes kept safe?▼
The release skill refuses to ship on red CI and never auto-bumps major versions. Bug, risk, and gap skills commit per finding, delegate push and CI watch to iris, and fix-forward on failure with revert as the bounded fallback. team-upgrade canaries on Milo first, then cascades one peer at a time with rollback and quarantine on failure.
What are the prerequisites and external dependencies?▼
Skills assume a Kubernetes namespace with peer agents, a primary git repo on a workspace source volume, GitHub access via gh api graphql for Discussions, and the piper-x-credentials Secret holding four OAuth 1.0a credentials for X API v2 posting. Formatting requires ruff, gofmt, goimports, prettier, yamllint, and helm lint.