Automation

Jev-powered verbatim compaction: once context reaches 60%, prunes stale tool calls with zero generation. Skips (never falls back to the LLM summarizer) when nothing is prunable; records evidence for learning.db.

Recalls the few memories related to each prompt: one line for the human, one context entry for the model. Local files only, no network on the hot path.

Pass-through evaluation instrumentation; never changes tool results.

Your own local Claude Tag, in the Claude app, Discord, Telegram, or a custom Claude Code channel.

Requests a session compaction when the live context passes a token threshold, so verbatim compaction plugins run near the working ceiling instead of at the model window.

Prints which instruction files the engine loaded (tier, path, @ imports, size) once per context load, and hints when a project CLAUDE.md keeps its AGENTS.md from being read.

Scores each unattended turn against the task the person asked for with TypeSafe's Jev model, and after consecutive off-task turns redirects or pauses the next machine-submitted prompt.

Lossless context management — DAG-based summarization that preserves every message

Pins every subagent to the fable model: rewrites the model on agent.spawn unless the spawn is a fork, with /fable-pin on, off and status. Touches nothing but its own store. Needs function hooks (early access).

Host-bound cognitive lifecycle events for The Homie.

Uses TypeSafe's Jev to allow or deny routine tool calls in auto mode, and hands uncertain ones to the built-in classifier

Shared Claude runtime adapter: normalises function-hook events into ClaudeRuntimeEvent, exposes $.runtime (emit bus, supports, judge, snapshot) to other plugins, and writes a JSONL event log

REALITY BENDER: intercepts a harmless tool result and alters what Claude receives; prints REAL vs RECEIVED

BLACK BOX RECORDER: structured per-session timeline of prompts, model steps, tool calls, subagents and denials; /blackbox prints it

Interception probe: deny, rewrite, replace result, delay, throw, register tool, classify

Permission override, subagent visibility, agent.spawn model rewrite, turn.step routing, prompt context injection

Spike: does $.session.compact() work from a plugin, and what does the session.compact event carry?

Spike: does a tool.describe rewrite reach the model, does agent.offer hide a type, does prompt.suggest fire?

pi agent for claude: a pi subagent type whose model steps run the pi CLI, so pi subagents and agent-team teammates live in Claude Code's own agent UI, streamed live.

Records each tool call's permission verdict and wall time so auto-mode classifier cost is measurable from the session index

Drive the herdr terminal workspace manager: panes, tabs, workspaces, and sibling coding agents

Compact once after 50 idle minutes, while the 1h prompt cache is still warm.

Lets Bash commands written for bash run under zsh: unmatched globs, leading =, and a missing timeout