> ## Documentation Index
> Fetch the complete documentation index at: https://ctpf.q-uestionable.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent-Operable Lifecycle

> Governed control and human govern surfaces for an untrusted AI caller

Source `main` exposes a machine lifecycle for an autonomous AI caller and a separate
TTY-only human governance surface. Neither is part of public PyPI `v0.14.0`. Do not treat this
page as a release commitment.

The harness is intended primarily for an AI agent and secondarily for a human operator. Humans
retain target and policy approval, spend and risk authority, scientific adjudication, and
publication. Mechanical results are not research conclusions.

## Surfaces

| Surface                   | Role                                                                           | Who                         |
| ------------------------- | ------------------------------------------------------------------------------ | --------------------------- |
| `ctpf experiment control` | Query-only discovery, validate, start, execute, status, cancel, result, verify | Autonomous caller           |
| `ctpf experiment govern`  | Signing key, signed policy, and per-run approval                               | Human on an interactive TTY |
| `ctpf experiment run`     | Direct human-operated packaged experiments                                     | Human                       |

Adapters (MCP, Skills, plugins) are optional later convenience layers. They receive no independent
policy, credentials, execution engine, evidence model, or scientific authority.

## Safe agent entry

From a source checkout on `main`:

```bash theme={null}
uv sync --group dev
uv run ctpf experiment control capabilities
uv run ctpf experiment control validate < runspec.json
```

`capabilities` and `validate` are query-only. They must not create research output or mutate
authority. A human creates signed policy (and, for higher tiers, approval) through `govern` before
the agent may `start` or `execute`.

```bash theme={null}
uv run ctpf experiment control start < runspec.json
uv run ctpf experiment control execute RUN_ID
uv run ctpf experiment control status RUN_ID
uv run ctpf experiment control result RUN_ID
uv run ctpf experiment control verify RUN_ID
```

Cancel is durable. Interrupted leases do not auto-resume. Exact budgets and deadlines are reserved
before effect boundaries.

## Contracts the agent must respect

* **RunSpec** — canonical JSON request bound by digest into grants and runs. Unknown fields fail
  closed. There is no arbitrary URL, shell, proxy, or replay field on this surface.
* **Policy** — human-signed standing or per-run authority: scenarios, target fingerprints, output
  roots, effect allowlists, resource ceilings, and authorization tiers.
* **Approval** — for higher tiers, a human-signed grant bound to one exact RunSpec digest.
* **Target identity** — full target ID plus validated-profile fingerprint; endpoint or model
  substitution denies before execution.
* **Scenario fingerprint** — installed scenario pin; fixture or prompt drift invalidates standing
  policy.

API keys stay in the OS keyring. Control envelopes carry credential aliases only.

## Authorization tiers

| Tier                | Meaning                                                                                    |
| ------------------- | ------------------------------------------------------------------------------------------ |
| Local synthetic     | Standing policy may allow packaged synthetic local/loopback work                           |
| Bounded remote      | Requires per-run human approval and explicit remote/retention acknowledgements             |
| Higher / human-only | Proxy replay, arbitrary targets, and destructive authority stay off the autonomous surface |

Exact tier names and policy fields are defined by the installed contracts. If validation denies,
fix the policy or approval; do not retry by mutating the rejected RunSpec.

## Evidence and verification

`control result` returns a mechanical record with stable promotion result and reason codes. It does
not contain a human conclusion or publication claim.

`control verify` checks the declared evidence bundle for internal consistency (manifest, hashes,
schema). It does not prove independent authenticity, scientific validity, or generality.

## Deployment residual

CTPF does not claim OS sandbox containment of a full-shell caller. Deploy the agent inside an
external OS/runtime sandbox and tool allowlist. Local HTTP listeners bind to `127.0.0.1` only.

## Related pages

* [Experiment Quickstart](/quickstart) — released `v0.14.0` human path
* [Pattern 2](/experiments/pattern2) — another unreleased source-only experiment
* [Responsible Use](/responsible-use) — authorization and sandbox expectations
* [Evidence and Integrity](/method/evidence) — bundles and verifier limits