Download and extract any file from a DevRev artifact ZIP directly in your workflow. Built-in support for parsing Computer session traces into structured Agent Studio format.

Why this snap-in exists

When users submit bug reports from DevRev Computer, a ZIP file containing session traces is automatically attached to the ticket. Extracting that file, parsing it, and routing it to the right analysis pipeline previously required custom code in every workflow. This snap-in eliminates that by providing two reusable workflow operations that any team can drop into their automation. Beyond Computer traces, the operations work on any artifact ZIP — log bundles, exports, debug archives — making this a general-purpose extraction primitive for DevRev workflows.

Operations

extract_artifact_zip_entry

Low-level primitive. Provide an artifact DON and a path inside the ZIP — the operation downloads the artifact, extracts the file, and returns its content. Set computer_reported_bugs: yes to parse the extracted file as raw MELTS/OTel session traces and receive a fully structured Agent Studio output with per-turn skill calls, inputs, outputs, token usage, and timing. Inputs:

• artifact_id — DevRev artifact DON
• entry_path — path to the file inside the ZIP (e.g. traces.json)
• computer_reported_bugs — yes / no (default no)

extract_artifact_from_ticket_body

Higher-level dual-mode operation. Use when you have a ticket and want to find the right artifact automatically. Direct mode — same as above, provide artifact DON directly. Ticket mode — provide the ticket body JSON (from an HTTP node), the artifact filename to match, and the entry path. The operation resolves the artifact DON by filename, downloads the ZIP, and extracts the file. Handles the double-encoded { "body": "..." } format that HTTP nodes produce. Inputs:

• Direct mode: artifact_id, entry_path, computer_reported_bugs
• Ticket mode: ticket_body (JSON string), artifact_name (exact filename), entry_path, computer_reported_bugs

Security

• Zip-slip protection on all archive paths
• Configurable max entry size (default 50 MiB) and max entry count (default 10,000)
• Token resolved from service_account_token or workflow-injected access_token — no PAT required