# `Agentic.Tools.Activation`

Tracks which external tools are "activated" (promoted to first-class
tool schemas) in the current agent session.

Activated tools appear as top-level tool definitions in the LLM request,
giving the model direct access without going through `use_tool`.

A budget system limits how many tools can be active at once to prevent
context window bloat. When the budget is exceeded, the least-recently-used
tool is deactivated automatically.

State is stored in the loop context -- not a separate process --
because activation is per-session.

## Callbacks

Uses `ctx.callbacks[:get_tool_schema]` to fetch tool schemas on activation.

# `activate`

Activate a tool -- add its full schema to the session tool list.

Returns `{:ok, schema, updated_ctx}` or `{:error, reason}`.
If budget is exceeded, the LRU tool is auto-deactivated.

# `active?`

Check if a tool is activated.

# `active_tool_definitions`

Get the list of activated tool definitions (for LLM request).

# `active_tool_names`

List active tool names.

# `budget_status`

Current activation count vs budget.

# `deactivate`

Deactivate a tool -- remove it from the session tool list.

# `init`

Initialize activation state in ctx.

# `touch`

Touch a tool's last_used timestamp (called on each use).

---

*Consult [api-reference.md](api-reference.md) for complete listing*