Menu
Writing hooks for the 8 hook points, execution modes, and decision semantics
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Internal architecture guide for the Meerkat agent platform. This skill should be used when understanding crate ownership, trait contracts, the agent construction pipeline, session service lifecycle, runtime control plane, mob orchestration internals, machine authority, comms wiring, or making cross-cutting architectural changes. Oriented toward AI agents and developers working on meerkat internals, not end users.
Build on the Meerkat agent platform. Covers every shipping surface (CLI, REST, JSON-RPC stdio/TCP, MCP, Python/TypeScript/Web/Rust SDKs), realm-scoped sessions, streaming, skills, hooks, memory, multimodal content, mob orchestration (spawn/fork/helpers/flows/profiles), durable scheduling, and provider auth via `auth_binding` + AuthMachine (env keys, OAuth, cloud IAM). Use when integrating with Meerkat, picking a surface, wiring auth, building agents, scheduling automated runs, deploying mobpacks, or asking how a feature exposes through a particular SDK.
Exact rkat CLI command contract for Meerkat help answers. Use this as the authority for command names, flags, aliases, and negative CLI facts.
Developer guide for the Meerkat WASM embedded runtime (meerkat-web-runtime) and the `@rkat/web` browser SDK. Covers wasm32 compilation, building/deploying WASM bundles, writing/auditing wasm_bindgen exports (sessions, mobs, comms, JS tool callbacks, external auth resolver, subscription handles), the auth_binding + registerExternalAuthResolver auth model, tokio_with_wasm aliasing, cfg-gating patterns, mobpack bootstrap, and common wasm32 gotchas. Use when touching meerkat-web-runtime, sdks/web, browser auth flows, or making a meerkat crate wasm32-compatible.
Ruthless Meerkat dogma inquisition for architecture, code, PRs, plans, tickets, ledgers, and designs. Use when Codex must find semantic-authority violations, shadow truth, provider leakage, surface-owned meaning, projection/mirror abuse, typed-truth failures, terminality lies, generation theater, or meta-level repeated dogma failure patterns; use for brutally honest reviews that cite concrete evidence and group symptoms by root dogma failure.
How to communicate with peers in a collaborative mob
| name | Hook Authoring |
| description | Writing hooks for the 8 hook points, execution modes, and decision semantics |
| requires_capabilities | ["hooks"] |
Use hooks for runtime observation and policy decisions at typed lifecycle points. Hooks can observe, allow, or deny; they should not become hidden owners of runtime truth.
Meerkat provides 8 hook points in the agent lifecycle:
Hooks return one of:
Semantic hook patches are retired. Hooks can observe typed projections and deny through the typed decision shape; provider parameters, assistant text, tool arguments/results, and final run text remain owned by the runtime/tool/LLM authority that produced them.
failure_policy is retained for compatibility. Current hook runtime failures
fail closed through typed engine errors; they are not converted into
warning-only success or hook-local denials by fail_open / fail_closed.
Use WorkGraph for durable work state, Schedule for time, and memory for knowledge retrieval. Hooks may observe those surfaces, but they do not own their semantics.
