Agent Relay Claude Code module

[github-actions[bot]]

A discussion dedicated to the Agent Relay Claude Code module. Share your thoughts, questions, and feedback here.

Module Scorecard

Presentation & Onboarding	Agent Integration	Credential Hygiene	Restricted-Environment Readiness	Engineering Quality	Overall
12 / 17	0 / 25	20 / 20	12 / 20	10 / 10	59 / 100

Drilldown

Presentation & Onboarding — 12 / 17

Criterion	Max	Score	Notes
Configuration-mode examples	12	12	The module has a single operational mode (run the self-hosted runner) with a minor install_cli toggle. The README's main example shows install_cli = true with a comment explaining the bake-into-image alternative. All 7 input variables are documented in the Parameters section with sensible defaults and descriptions. No separate code example is needed for a module with this configuration surface.
Visual preview	5	0	No image, GIF, or video embedded in the README. The frontmatter references an icon file (claude.svg) but that is not a visual preview of the module in action.

Agent Integration — 0 / 25

Criterion	Max	Score	Notes
AI governance	10	0	No mention of Coder AI Gateway or Agent Firewall anywhere in the README or source. The module relies on Agent Relay's dispatch mechanism for credential delivery, but does not document how Coder governs auth, routing, or policy enforcement at the gateway/firewall level.
Dashboard entry point	5	0	No coder_app resource declared or documented. The module is a headless runner with no UI to open from the dashboard.
Session continuity	5	0	No documentation of resuming a Claude Code session across reconnects or relaunches. No mention of tmux, screen, boo, or native session-ID resume. The agent_relay_session_id parameter supports relay-level deduplication, not agent-session continuity.
Managed configuration	5	0	No documentation of managed MCP servers, settings files, policies, or workdir configuration. base_dir sets the session checkout directory but does not constitute managed agent configuration.

Credential Hygiene — 20 / 20

Criterion	Max	Score	Notes
Secrets marked sensitive	16	16	agent_relay_credential is styled with mask_input = true and is ephemeral. The README states "the credential is masked." No inline secrets appear in any README example; the only example shows agent_id and install_cli.
Non-hardcoded auth path	4	4	The README documents that the credential is a "Single-use work order JWT" stamped by Agent Relay on dispatch. The user never pastes a raw key; auth flows through the relay's dispatch mechanism into SELF_HOSTED_RUNNER_ENVIRONMENT_SECRET.

Restricted-Environment Readiness — 12 / 20

Criterion	Max	Score	Notes
Mirrorable artifact source	5	0	The install URL https://claude.ai/install.sh is hardcoded in install.sh.tftpl. No module input variable overrides this download URL. cli_binary overrides the binary path, not the source URL. install_cli is a skip toggle, not a URL override.
Bring-your-own binary	10	10	install_cli = false (documented in README: "Bake the CLI into the image and set this to false for faster workspaces") disables the download entirely. The install script checks command -v and skips when the binary is present. The start script adds ~/.local/bin to PATH unconditionally so a pre-baked CLI is found.
Egress transparency	3	0	No dedicated README section enumerates external endpoints. The install contacts claude.ai/install.sh (→ downloads.claude.ai) and the runner contacts Anthropic's API, but these are only visible in source code, not in a dedicated network/offline/air-gapped section.
Runs without sudo	2	2	All three scripts (install.sh.tftpl, start.sh.tftpl, status.sh.tftpl) never invoke sudo. The installer writes to ~/.local/bin; the start script creates directories under $HOME. No root required for core functionality.

Engineering Quality — 10 / 10

Criterion	Max	Score	Notes
Input quality	6	6	All 7 variables have clear descriptions and sensible defaults. cli_binary has a regex validation rejecting shell metacharacters; exit_if_unused_min validates whole-number ≥ 0. Descriptions explain the "why" (e.g., why base_dir defaults to $HOME/workspace rather than the CLI's /workspace).
Test coverage	4	4	main.tftest.hcl covers the parameter contract (names, ephemeral flags, styling), runner wiring (env var names, script ordering, state-file consistency), install behavior, and injection safety. main.test.ts runs the actual install/start scripts in a container with a stub claude binary, verifying lifecycle states (idle, working, done, failed), wrapper behavior, exit codes, path overrides, and serving_log_pattern as fixed-string data.

Overall — 59 / 100

Raw 54 / 92 → round(54 / 92 × 100) = 59

---

Scored against SCORECARD.md on 2026-09-28 with solstice-1.