Skip to content
github-actions[bot] edited this page Jun 3, 2026 · 49 revisions

Crickets — Inspired by the Noisy Cricket

Inspired by the Noisy Cricket — agent primitives that punch far above their weight.

CI Latest release License: MIT

Works with Claude Code + Antigravity — see compatibility

Crickets is the noisy cricket — small, focused agent customizations that punch above their weight. Skills, hooks, sub-agents, bundles, MCP servers, slash commands, status lines, output styles, workflows, rules, snippets, settings-fragments. The primitives you carry into any project to make Agent M (agentm) effective.

This wiki is the contributor-facing documentation for Crickets itself. Every page is written for a single reader intent (learn / do / look up / understand) per the Diátaxis convention.

Note

Repo: github.com/alexherrero/crickets. Public, MIT-licensed. PII guardrails (script + skill + pre-push hook) are foundational — see Purpose and scope for why.

📚 New here? Learn by doing.

🔧 Trying to do something specific?

  • Install crickets plugins — three install modes (one-liner / marketplace / manual) across Claude Code + Antigravity.
  • Develop a crickets plugin locally — the source → generate → dogfood → commit loop.
  • Add a skill — add a new standalone skill.
  • Quality-Gates-Recipe — operator-facing recipe for the 5-primitive quality-gates set (replaces the v1.x quality-gates bundle; kind: bundle reserved-future in v2.0.0).
  • Use the evaluator — dispatch the evaluator sub-agent for PASS / NEEDS_WORK grading against a precise rubric.
  • Use the base hooks — kill-switch, steer, commit-on-stop. Operator-precision control for long-running Claude Code sessions.

The evidence-tracker hook and the design, memory, and diataxis-author skills, along with Antigravity plugin authoring, moved to Agent M in v2.0.0 (V4 #36 reorg). See the Agent M wiki for their operational docs.

📖 Looking up a detail?

  • Customization Types — what each of the 13 kinds means and where to put them (kind: bundle + kind: plugin reserved-future in v2.0.0).
  • Manifest Schema — YAML frontmatter contract.
  • Per-Host Paths — destination paths per kind per host.
  • Installer CLI — flags, prereqs, exit codes.
  • Compatibility — supported hosts (Claude Code, Antigravity) + OS matrix + per-manifest supported_hosts contract.

💡 Want to know why?

  • Purpose and scope — what this repo is for, what it's not.
  • Cross-Repo Memory Protocol — how agentm reads from + writes to the toolkit-side /memory skill at phase boundaries.
  • V3 Retrospective — what shipped, what we learned, what's next (closes the V3 arc; ships with harness v3.0.0 + toolkit v1.0.0).
  • Agent Memory Evolution: From ContextVault to V7 — V1→V7 HLD of the AgentMemory architecture (V3 ships with this release; V4+ is roadmap-deferred).
  • Device-Wide Architecture — V4 architectural shift from per-repo harness to device-wide agentic OS; rationale for the agentm/crickets split + which primitives live where.
  • Memory-OS ArchitectureV5, the unbundling: agentm becomes a storage-agnostic memory OS + plugin host; the workflow/docs/PM/storage capabilities unbundle into crickets plugins. The memory↔process + memory↔storage seams, the device-local default, and the vault-conserving cutover.

Architecture decisions

Designs

The canonical "Why we built X" entry points (published via the /design skill, surfaced here when the parent design's last queued part hits Status: done and harness /release transitions the design to launched).

Conventions

Page templates, filename rules, and the Diátaxis four-mode split mirror the sibling repo's documentation convention — see agentm/harness/documentation.md. The mode-purity lint (scripts/check-wiki.py --strict) is the same.

Clone this wiki locally