Module task_matrix 

Agent task-to-command matrix rows shared by fallow schema, the init --agents template, the agent hook block, root --help, and the fallow://task-matrix MCP resource. Data only; the CLI owns rendering. Single source of truth for the agent-discoverability task-to-command matrix (R2/R3). One const slice drives every render surface: the fallow schema manifest (task_matrix), the init --agents AGENTS.md template, the hooks install --target agent managed block, the root --help cheat sheet, and the fallow://task-matrix MCP resource. The scripts/generate-agent-docs.mjs generator renders the same table into SKILL.md from the schema-serialized form, so the Markdown surfaces stay consistent without duplicating the rows.

This module carries data only. The Markdown renderer and the clap probe drift test live in crates/cli, which owns the command tree; the MCP server projects the rows without probe.

Read-only-evidence principle (R1): the matrix carries NO mutating commands (fix, init, hooks, migrate, setup-hooks, watch). Unit tests in this crate and in crates/cli pin that contract, mirroring the next_steps[] builder in the CLI report layer.

Structs

TaskRow

One task-to-command row for the agent-discoverability cheat sheet (R2/R3).

Constants

MUTATING_COMMANDS

Mutating command tokens the matrix must never reference (R1 read-only principle). Shared with the CLI schema exclusion test.

TASK_MATRIX

The canonical task-to-command matrix. Verified against the live clap command tree; the CLI schema drift test re-checks every non-empty probe.

Functions

leading_command_token

The first command token after the fallow prefix, or the empty string for a bare flag-fragment row.