Skip to content
Octopus Octopus Octopus

consigliere-lens

The generic knowledge engines read like flat reports on any base. Run against a manager’s private workspace, their findings should read like the consigliere — the political risk others miss, the heuristics you already hold, your own register. consigliere-lens is the wrapper that turns a generic audit into that.

What it solves

A manager’s workspace is not just another knowledge base. It carries the political nuance of who owns what and what is safe to push, plus the manager’s own operating heuristics. A generic hygiene or briefing report flattens all of that away — it tells you a node is stale, not that the stale node is the one the finance VP is quietly waiting on. The lens delivers the manager-multiplier value by reusing the engines rather than forking a separate set for the manager.

How it works

It is a wrapper, not a fork — the three engines stay generic and untouched.

  • A deterministic helper, octopus lens, surfaces the grounded material for a node: its playbook heuristics, the lines under its Political risk section, and its blockers.
  • The lens runs an engine (hygiene, synthesize, or briefing) read-only against the workspace, then reframes each finding through that material — in the consigliere’s voice.
  • The voice is the consigliere role, an Opus persona, because political read and “thinks like you” judgment warrant the stronger model — unlike the engines’ cheap-tier narration.
  • It is read-only: it composes engine runs without --fix and never writes the workspace. Managerial data is never mutated by the lens.

Usage

octopus lens profile <root>     # the lens profile a root declares
octopus lens context <node>     # playbook + political-risk + blockers for a node

The skill drives those helpers, runs the chosen engine against the consigliere root, and narrates the result. Every line is grounded — it cites the node and the playbook/state line it drew from, and surfaces no political read the workspace does not support.