spacebot

Specialized Spacebot development skill with deep architectural knowledge. Use when working on any Spacebot feature, bug fix, refactor, or investigation. Covers the five-process agent model (Channel, Branch, Worker, Compactor, Cortex), memory system (SQLite + LanceDB + redb), task lifecycle, 60+ tool definitions, Rig framework integration, prompt templates, identity system, messaging adapters, config hot-reload, security/sandbox, and API server. Activate for any Rust code changes in src/, prompt template edits in prompts/, migration work, tool additions, or architectural questions.

Use this skill when preparing a release bump or updating release notes. It writes a launch-style release story from the actual change set, then runs `cargo bump` so the generated GitHub notes and the marketing copy land together in `CHANGELOG.md`.

Use when bumping a bun-managed dependency in interface/, docs/, packages/api-client/, or spaceui/. Avoids the "lockfile updated, manifest spec stale" drift class that opened PRs

Use when auditing project documentation for drift, stale claims, missing changelog entries, completed-but-still-listed TODOs, or when the user asks to review/refresh/cleanup docs across interface/, spaceui/, docs/, or repo-root markdown files. Complements session-sync (which handles CLAUDE.md + memories only).

Pre-stage the fastembed BGESmallENV15 ONNX model cache so the 4 memory::search integration tests can run reliably. Call when the user reports `EmbeddingFailed("Failed to retrieve onnx/model.onnx")`, when memory::search tests fail with `native-tls` cert errors, or before a `cargo test --lib` run on a fresh checkout / clean target/.

Use this skill when adding a new REST handler under `src/api/`. Scaffolds the full utoipa + typegen dance: handler function with correct extractor ordering (State → Path → Query → Json), request/response structs with `#[derive(utoipa::ToSchema)]`, `#[utoipa::path(...)]` annotation, router registration in `src/api/server.rs`, and the `utoipa::OpenApi` registration block. Reminds to run `just typegen` + `just check-typegen` before commit. Canonical rule: `.claude/rules/api-handler.md`.

Reference knowledge about the Talos Kubernetes cluster that Spacebot deploys into. Use PROACTIVELY when working on containerization, Dockerfile changes, deployment configuration, health checks, environment variables, networking, storage, secrets, database connections, or any work that touches how Spacebot runs in production. Also use when the user mentions the cluster, Kubernetes, k8s, Talos, Flux, namespaces, pods, or asks about Spacebot's production environment.

End-of-task sync — rebuild code graph, check CLAUDE.md for drift, sync Serena memories against codebase ground truth, and commit doc fixes. Run after completing any significant work (feature, migration, refactor, bug fix) to keep all documentation surfaces current. Use proactively when the user says "sync", "update docs", "session done", "wrap up", or finishes a multi-step task.

Session bootstrap primer for Spacebot development. This skill should be used when starting a new Claude Code session, when the user says "prime", "bootstrap", "get up to speed", "load context", "new session", "start fresh", "primer", "where did we leave off", "catch up", or "resume". Activates all memory systems, retrieves cross-session context, loads project knowledge, and prepares the assistant to continue work without re-learning. Typically the first skill invoked in any session.

Use this skill after completing a bulk mechanical sweep (em-dash replacements, severity escalations, module-doc resyncs, grep-driven multi-file edits) that touched files across multiple gate boundaries. Maps modified file patterns to the specific verification gates that apply, then runs only the relevant gates. Encodes the CI-typegen lesson from PR

Use this skill when adding a new REST handler under `src/api/` that needs the Phase-4 authz gate, OR when rolling the gate onto an existing ungated handler family. Scaffolds the ~45-line inline gate block (per N1 decision), the `AuthContext` extractor insertion (per api-handler.md ordering), the `.await set_ownership` on create paths (A-12), the pool-None fallback with file-family metric label, and the five canonical tests in `tests/api_<resource>_authz.rs`. Canonical references: `.claude/rules/api-handler.md` + `.scratchpad/plans/entraid-auth/phase-4-authz-helpers.md` (decision N1). Trigger: `/handler-authz-rollout` or when a PR adds a new `src/api/*.rs` handler that returns ownable data.

Use this skill when a PR has received review findings (typically from `/pr-review-toolkit:review-pr`, manual code review, or `/spacebot-dev`) that need to be addressed before merge. Takes an aggregated findings document (e.g., `.scratchpad/YYYY-MM-DD-prN-review-aggregated.md`) and drives a multi-commit remediation loop with narrow verification gates between commits. Enforces commit-message discipline (review-item IDs cited in every commit message), commit grouping (fix-class + test-class + doc-class commits stay separate), and prevents the common failure mode of lumping unrelated fixes into one "address review" commit. Canonical reference for the batch pattern: Phase 4 PR 2's 27-commit remediation at `feat/entra-phase-4-pr-2-handler-rollout`.

Audit open Dependabot PRs, replace them with direct-to-main SHA-pinned commits carrying audited intent, and add `ignore:` rules that prevent policy-violating majors from being re-proposed. Codifies the pattern established 2026-04-21 responding to PRs

Query the local graphify knowledge graph to answer semantic cross-document questions. Wraps `just graphify-query` with synthesis and citation. Use when a question spans design docs, RFCs, or screenshots that a single-file read cannot answer, or when the user explicitly asks about "what connects X to Y", "why did we decide Z", or "which doc explains W". Requires a pre-built graph at `graphify-out/graph.json` — if missing, stops and tells the user to run `just graphify-rebuild docs/design-docs/` first.

Check the size of `target/` and run `just sweep-target` if it exceeds a threshold. Use when disk pressure is a concern, at mid-phase checkpoints, or after a TDD cycle that compiled many test binaries. The canonical storage-budget reference is `.scratchpad/plans/entraid-auth/INDEX.md` § "Cargo discipline — storage + time budget." Reports bytes reclaimed.

Use this skill to scan staged or working-tree files for writing-guide violations before commit. Catches em-dashes in prose, semicolons in prose, phase-number drift, and common AI-generated phrasing patterns ("Not X. Not Y.", "This isn't X, it's Y") across markdown AND code comments (`//!`, `///`, `--`). The inline pre-commit grep patterns in .claude/rules/writing-guide.md only cover markdown; this skill extends to Rust and SQL comment styles where violations repeatedly landed during Phase 2.

Invoke when writing or reviewing Rust unit/integration tests that exercise code which internally calls `tokio::spawn`, constructs `runtime::Tokio`-flavored opentelemetry_sdk types, or otherwise touches the Tokio runtime at construction time. Catches the "bare `#[tokio::test]` + `BatchSpanProcessor`" deadlock class before a 20+ minute rebuild discovers it.

Guide for deploying Spacebot into the Talos Kubernetes cluster via Flux GitOps. Use when the user wants to create or update Spacebot's Kubernetes manifests, add Spacebot to the cluster, scaffold the Flux app structure, configure secrets for the cluster, set up ingress/networking, or prepare Spacebot for Kubernetes deployment. Also triggers on "deploy to cluster", "add to k8s", "create helm release", "flux app", or "cluster manifest".

Use when creating or editing memories. Covers when to write vs skip, the eight memory types, importance scoring, association discipline, content craft, and anti-patterns that bloat the memory graph.

Use when creating, updating, or shepherding tasks through the lifecycle. Covers when to elevate vs decompose, priority signals, status discipline, the owner-vs-assigned split, and the global-task vs per-agent distinction.

This skill should be used when the user asks to "review the prompt", "audit the system prompt", "check prompt quality", "inspect what the LLM sees", "debug prompt issues", or "find prompt engineering problems". Pulls the live rendered prompt via the API, explains how it's composed, and reviews it for issues.

Specialized Spacedrive development skill for AI coding assistants working on Spacedrive integration. Covers the VDFS architecture, CQRS operations, library system, hybrid indexing, P2P sync (HLC/CRDT), Iroh networking, extension system (WASM), virtual sidecars, CLI commands, React hooks, semantic color system, and the Spacebot-Spacedrive integration contract (connection modes, remote execution, file system intelligence, ContextNode). Use when working on Spacedrive integration, the vendored `spacedrive/` subtree at the repo root, P2P networking, file indexing, sync systems, or the Spacebot-Spacedrive contract.

Specialized SpaceUI component library skill for AI coding assistants. Covers the 6-package in-tree monorepo (@spacedrive/tokens, primitives, forms, ai, explorer, icons), 42 UI primitives, 13 AI agent components, design token system, Tailwind v4 configuration, form integration (React Hook Form + Zod), semantic color system, component variants (CVA), and consumer integration patterns. Use when working on UI components, styling with SpaceUI tokens, building AI chat interfaces, task management UI, or integrating SpaceUI into consuming apps.

This skill should be used when the user asks to "open a PR", "prepare for review", "address review comments", "run gates", or "verify before pushing" in this repository. Enforces preflight/gate workflow, migration safety, and review-evidence closure.

This skill should be used when the user asks to "address review feedback", "fix PR comments", "close findings", "respond to reviewer notes", or "reduce review churn". Enforces a finding-to-evidence closure loop for every review round.

Debug, clean, redeploy, and validate Spacebot's running deployment in the Talos Kubernetes cluster. Use when the user wants to check pod status, view logs, clean up a failed deployment, force a redeployment, troubleshoot a crash loop, inspect cluster resources, or validate Spacebot is running correctly. Also triggers on "pod logs", "restart spacebot", "why is spacebot down", "flux reconcile", "clean deploy", "redeploy", or "cluster status".

Archive a completed change in the experimental workflow. Use when the user wants to finalize and archive a change after implementation is complete.

Implement tasks from an OpenSpec change. Use when the user wants to start implementing, continue implementation, or work through tasks.

Use when: User wants to run Archon workflows, CREATE workflows or commands, set up Archon, or manage Archon configuration. Triggers (run): "use archon to", "run archon", "archon workflow", "use archon for", "have archon", "let archon", "ask archon to". Triggers (create): "create a workflow", "write a workflow", "make a command", "author a workflow", "new workflow", "new command", "archon workflow yaml". Triggers (setup): "set up archon", "install archon", "how to use archon", "configure archon", "archon setup", "get started with archon". Triggers (config): "change my archon config", "modify archon config", "archon config", "change archon settings", "update my config", "help me change my config", "edit archon config", "archon configuration". Triggers (init): "initialize archon", "set up .archon", "archon init", "add archon to repo". Capability: Runs AI workflows in isolated git worktrees for parallel development. Also: Creates and manages workflow YAML files, c

Update and audit project dependencies across Rust, frontend, and docs. Use when the user mentions dependency updates, package upgrades, security vulnerabilities in deps, Dependabot alerts, cargo update, bun update, outdated packages, or wants to patch known CVEs. Also use proactively when cargo audit or bun audit reveals fixable vulnerabilities.

Use this skill when the user asks to "commit all", "commit everything", or wants all outstanding changes committed. Groups unrelated changes into separate, well-described commits instead of one catch-all commit.