> ## Documentation Index
> Fetch the complete documentation index at: https://docs.app.useharmonia.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Ryan — content strategist

> Current Strands registration, model, inputs, tools and authority.

## Role

Proposes startup positioning, content pillars, channels, objectives and grounded briefs from the operator context and validated source analysis.

## Contract

| Property                 | Value                                                  |
| ------------------------ | ------------------------------------------------------ |
| Registered specialist    | `ryan_strategist; ryan_research_strategist`            |
| Default Bedrock model ID | `us.anthropic.claude-sonnet-4-6`                       |
| Typed boundary           | `StrategistInput → StrategistResult → ContentStrategy` |
| Returned state key       | `strategist_result`                                    |

## Skills

The host preloads `ryan-strategy-skills` method guidance and approved references into the system prompt. No `load_skill` or `load_skill_resource` tool is registered. Host bootstrap records activation provenance. Skill guidance is method, never evidence.

## Tools and research

`ryan_strategist` has no tools. The host may select `ryan_research_strategist` with the single `ryan_gateway_search` tool for an exact supplied researchRequest. The before-tool hook consumes one-shot research authority before dispatch.

## Prompt method

Separate supported facts from assumptions. Strategy choices cite supplied operator, source or returned research evidence. Research is bounded by the host request and cannot silently widen the task.

## Evidence and provenance

Only the supplied typed input and authorized returned tool evidence can support factual claims. Preloaded skills, model memory and generated prose do not establish facts. The host retains request hashes and validated references without logging private prompts or chain-of-thought.

## Authority boundary

The specialist cannot approve, grant permissions, publish, schedule an external effect, access credentials, mutate workflow state or create verification receipts. Those operations belong to deterministic host code and persisted authorization.

## Handoff

The host validates and persists the proposal. Human strategy approval binds its digest before Temi can operationalize it.

## Runtime enforcement

`LocalStrandsTeamRuntime` creates a fresh Strands `Agent` per invocation with `structured_output_model`, `AuthorityHooks` and automatic model retries disabled. The host selects the registration, projects bounded input and validates output before a stage mutation. Provider repair follows the bounded host policy. Authenticated AWS execution remains pending live verification.

## Implementation references

* Registration: `agent/harmonia_agent/agents.py`
* Prompt: `agent/harmonia_agent/ryan_prompt.py`
* Method/tool implementation: `agent/harmonia_agent/ryan_skills.py`
* Runtime: `agent/harmonia_agent/team_runtime.py`
* Model assignment: `agent/harmonia_agent/role_models.py`