refactor(aidd-telemetry): the consent skill is named after its first minute

Localiza la skill que actualmente se llama 00-init y, después, busca referencias en el plugin README, CATALOG.md, el marketplace manifest y las skills que hacen routing hacia ella. Lee su router y las acciones 01-check, 02-enable, 03-verify, 04-identify y 05-forget para conservar el alcance existente. Se considera terminado cuando el nombre y la descripción de la skill cubren permitir, identificar, detener y borrar la medición, todas las referencias están actualizadas y la numeración de las demás skills no ha cambiado.

Problem

aidd-telemetry's first skill is called 00-init, but initialisation is only one of the three things it owns. Its own router states them as independent asks: allow measurement on this project, choose whether a person is named on their own records, and remove what was measured. Its actions say the same: 01-check, 02-enable, 03-verify, 04-identify, 05-forget.

So the skill that holds consent — granting it, proving it took, attaching a person to it, withdrawing it, and erasing what it produced — is named after the first minute of its life. Two consequences, both user-visible:

• Someone looking for how to stop being measured, or to erase what was recorded, has no reason to open a skill called "init". The way out is less findable than the way in, which is exactly backwards for a measurement feature.
• The plugin README compressed the row to "turns measurement on", which is how the false claim it carried until #871 got there in the first place.

This also sits against the repository's own naming rule: name by intention, not by mechanism or moment (CLAUDE.md, "Name by intention").

Scope

• Includes: renaming the skill to name its responsibility — consent over measurement — and updating every reference: the plugin README's table, CATALOG.md, the marketplace manifest if it names skills, and any skill that routes to it.
• Includes: the description field, which is what a tool matches on. It should mention stopping and erasing, not only starting.
• Excludes: splitting the skill. Keeping the exit in the same place as the entrance is the property worth protecting; a separate "forget" skill would make erasure harder to find, not easier.
• Excludes: any change to the actions themselves, to the CLI, or to what is recorded.

Acceptance

• The skill's name states what it owns, not when it runs.
• Its description names the three asks: allow, attach a person, stop and erase.
• A person who asks their tool "how do I erase what was measured about me" reaches it without knowing the skill's name.
• Every reference is updated in the same change, and the numbering of the other skills stays coherent.

Prior art

• #871 fixed a README row that claimed this skill ran check; the compression that produced that claim came from describing the skill as "init".
• The router and its five actions, which already state the three asks.

Out of scope

The other two skills, 01-cost and 02-check, whose names already state what they answer.

🤖 Generated with Claude Code

https://claude.ai/code/session_01VWNxk63AGKkqE8HRqHLjGp