# Spectrum card: Claude Code, on your own machine

> GAMMA-Ray v0.0.26 (beta, for review; every rating is experimental). The GAMMA-Ray Spectrum beta card of Claude Code, on your own machine: A4 · R2 · C4 · O2, Scaling; 75 provisions lit across 8 instruments.
> Page: https://gamma-ray.sgit.ai/spectrum/cards/claude-code-local/ · Source: data/scans/claude-code-local.json


Spectrum 0.1 beta · Public tier · read 2026-11-10 · not reviewed by its vendor yet

The command-line coding agent, running as the user’s own account on their machine, with the settings it ships with.

Claude Code, on your own machine a product, as documented SPECTRUM 0.1 BETA · Public tier · 2026-11-10 A4 Autonomy Orchestrates (available) R2 Reach Integrated (available) C4 Control Enforced (available) O2 Openness Documented Stage: Scaling (12 of 16) Standout. Autonomy 4 (Orchestrates): "approve the action so Claude may retry" Next unlock. Reach 4 (Ecosystem): public evidence that agents operate across organisational boundaries or disparate stacks would get there. Standards constellation. 75 provisions lit across 8 of 8 instruments. Salesforce EAAMM OWASP MCP AAGMM Microsoft NIST AI Act Evidence: 16 sources, captured and hashed · the card's page has the vault's read key · Not reviewed by its vendor yet · correct or claim this card

*Figure. The walk from the card down to each indicator, its public source and the instruments it touches; the sections below list the same in text.*

This card rates a product as its vendor documents it, not anyone's deployment of it. A feature the operator has to turn on counts, and is marked as available. Levels are computed from the evidence below by [published rules](../../evidence-map/index.md). A dimension with no evidence reads "not yet evidenced".

## Autonomy: level 4, Orchestrates

*What does the agent do on its own?* Runs multi-step workflows across systems or coordinates other agents; a person sets goals and reviews outcomes.

- `I10` It works across several domains or systems. The codebase, the shell, git and the development tools it integrates with. [Source](https://code.claude.com/docs/en/overview) (read, documented, captured 2026-11-10T06:10Z, sha256 `d52a8e91409de402`)
- `I12` Several agents coordinate inside the organisation. Agent teams coordinate several instances, each in its own context, that message each other. The vendor says they "are experimental and disabled by default". [Source](https://code.claude.com/docs/en/agent-teams) (read, documented, available, captured 2026-11-10T06:10Z, sha256 `edd1fdbbbccb70a3`)
- `I18` The agent works independently; the person approves critical decisions only. In auto mode a classifier reviews actions instead of prompting; an action it blocks is listed, and the person can approve it so Claude may retry. "approve the action so Claude may retry" [Source](https://code.claude.com/docs/en/permission-modes) (read, documented, available, captured 2026-11-10T06:10Z, sha256 `0eeeb38694c6d38a`)

This evidence touches 10 provisions: Salesforce, Microsoft, OWASP, AAGMM, EAAMM.

## Reach: level 2, Integrated

*How connected is it?* Documented connectors to common tools.

- `I10` It works across several domains or systems. The codebase, the shell, git and the development tools it integrates with. [Source](https://code.claude.com/docs/en/overview) (read, documented, captured 2026-11-10T06:10Z, sha256 `d52a8e91409de402`)
- `I11` It connects to external tools or services across a trust boundary. It can connect to Model Context Protocol servers; a project’s servers have their own approval prompt. [Source](https://code.claude.com/docs/en/security) (read, documented, available, captured 2026-11-10T06:10Z, sha256 `f05e21c4c65ff0eb`)

This evidence touches 8 provisions: Salesforce, Microsoft, OWASP, NIST AI RMF.

**Next unlock: level 4, Ecosystem.** Public evidence that agents operate across organisational boundaries or disparate stacks.

## Control: level 4, Enforced

*Can people stay in command?* Limits enforced outside the agent, described in public.

- `I45` A limit on it is enforced by something outside its own reach. Sandboxing restricts shell commands at the operating system level (filesystem and network); the operator turns it on. [Source](https://code.claude.com/docs/en/sandboxing) (read, documented, available)
- `I48` Policy is enforced as machine-readable rules. An organisation can deliver managed settings that a user’s own files do not change; hooks run on every tool call and can block it. This is an enforcement mechanism, not a policy. [Source](https://code.claude.com/docs/en/settings) (read, documented, available, captured 2026-11-10T06:10Z, sha256 `b6dfc139db87f8e1`)

This evidence touches 7 provisions: OWASP, AAGMM, Microsoft.

## Openness: level 2, Documented

*Does it show its work?* Public documents on how it works, its data, its models and its limits.

- `I22` Its limits and known failure cases are documented for those who use it. The security page states limits, for example that an approved shell command can write anywhere the account can. [Source](https://code.claude.com/docs/en/security) (read, documented, captured 2026-11-10T06:10Z, sha256 `f05e21c4c65ff0eb`)
- `I58` The data it uses is described. A page describes what data is sent, kept and used. [Source](https://code.claude.com/docs/en/data-usage) (read, documented, captured 2026-11-10T06:10Z, sha256 `0f5fa5f2a836052b`)
- `I61` The models and third-party components it uses are listed. The model configuration page lists the models it uses by alias (Opus, Sonnet, Haiku, Fable) and the version each alias resolves to for each provider. "Uses the latest Opus model for complex reasoning tasks" [Source](https://code.claude.com/docs/en/model-config) (read, documented, captured 2026-11-10T06:10Z, sha256 `1d0950edada35be5`)

This evidence touches 9 provisions: NIST AI RMF, EU AI Act, AAGMM, EAAMM, Salesforce, OWASP.

Also evidenced: the rule for level 4. It will count when the level before it is.

**Next unlock: level 3, Measured.** Public evidence that test or evaluation results are published.

## The standards constellation

One piece of public evidence can speak to several frameworks at once. "Touched" never means "complies". The share is of the provisions that can be read from outside.

| Instrument | Provisions touched | Of those publicly readable |
|---|---|---|
| Salesforce | 15 | 24 |
| EAAMM | 3 | 8 |
| OWASP | 12 | 17 |
| MCP authorization | 4 | 8 |
| AAGMM | 7 | 7 |
| Microsoft | 12 | 23 |
| NIST AI RMF | 17 | 20 |
| EU AI Act | 5 | 6 |

## Upper layers

The public card is the front door. The same evidence, read with the governance rules, is this agent's [GAMMA-Ray rating](../../../scans/claude-code-local/index.md): freedom, governance, the level held, the grant and a work order. That layer is written for an operator.

## The vault

Every source behind this card was captured and hashed into a public sgit vault (`08k79drk`): `captures.json` lists each URL with the SHA-256 of what was fetched and when. Read key:

`sgit_public_read_04133526e7f502cb0c0a0e18705b0b34195413427a40eb785512bd74df17b880:08k79drk`

With [sgit](https://sgit.ai/): `sgit clone sgit_public_read_04133526e7f502cb0c0a0e18705b0b34195413427a40eb785512bd74df17b880:08k79drk`.