[chore] Add Claude harness on Daytona sandbox

[junaway]

Context

Claude was the one harness x sandbox cell missing from the matrix on Daytona (it already worked on local and, in a sibling branch, on E2B). Both Codex-style credential modes needed a Daytona-side provisioning path analogous to the existing Pi one.

Changes

Adds prepareDaytonaClaudeAssets / uploadClaudeAuthToSandbox in daytona.ts. Own-login credential upload uses an explicit allow-list of exactly [".credentials.json"] (never a directory scan) sourced from process.env.CLAUDE_CONFIG_DIR || join(homedir(), ".claude"), written to DAYTONA_CLAUDE_DIR (defaults to /home/sandbox/.claude, overridable via AGENTA_AGENT_SANDBOX_CLAUDE_DIR). Managed-credential mode is also supported.

Tests / notes

• New coverage in sandbox-agent-daytona.test.ts covering both credential modes and the allow-list behavior.
• Note: chore/add-remote-tools-gate in this same batch means Claude-on-Daytona runs with tools will hit the new loud-refuse gate rather than silently dropping tools, until the relay-client shim lands.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
agenta-documentation	Ready	Preview, Comment	Jul 2, 2026 9:42pm

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 5624a208-9378-427d-87ca-dcdd40001f45

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch chore/add-claude-daytona

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[Copilot]

Pull request overview

Adds the missing Claude × Daytona cell to the runner’s sandbox matrix by provisioning Claude “own-login” credentials into Daytona sandboxes (with a strict allow-list), wiring the provisioning into the Daytona run flow, and extending unit coverage to cover both managed and runtime-provided credential modes.

Changes:

• Add Daytona Claude asset preparation (prepareDaytonaClaudeAssets) and host → sandbox credential upload (uploadClaudeAuthToSandbox) with an explicit allow-list for .credentials.json.
• Wire Claude Daytona preparation into runSandboxAgent alongside existing Pi Daytona preparation.
• Extend unit tests to cover Claude Daytona credential behavior (including env overrides and best-effort error handling) and update the alive-heartbeat test to reflect the current heartbeat contract.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
services/runner/tests/unit/session-alive.test.ts	Updates the release-heartbeat assertion to match the heartbeat payload contract (no status field).
services/runner/tests/unit/sandbox-agent-daytona.test.ts	Adds unit coverage for Claude-on-Daytona credential upload behavior and env override handling.
services/runner/src/engines/sandbox_agent/daytona.ts	Implements Claude credential upload + Daytona Claude asset preparation, including destination-dir override support.
services/runner/src/engines/sandbox_agent.ts	Invokes the new Daytona Claude preparation during Daytona runs.
docs/design/agent-workflows/projects/add-claude-daytona/tasks.md	Adds project task checklist for Claude-on-Daytona work.
docs/design/agent-workflows/projects/add-claude-daytona/specs.md	Adds detailed specs/acceptance criteria for Claude-on-Daytona credential provisioning behavior.
docs/design/agent-workflows/projects/add-claude-daytona/research.md	Documents investigation findings and rationale for the chosen implementation shape.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.