cowork-harness

Test or debug a Claude Code skill/plugin under Claude Cowork's runtime — sandboxed agent, default-deny egress, the can_use_tool permission/question protocol — using the cowork-harness CLI. Use when validating or regression-testing a skill, authoring or debugging a scenario YAML (prompt + scripted answers + assert:), choosing a fidelity tier, scripting AskUserQuestion / tool-permission answers, or asserting artifacts, egress, or sub-agent dispatch. Especially when a harness run no-ops an assertion, fails on an unanswered gate, false-greens, a steered answer never reaches the model, or a web_fetch is unexpectedly denied or gated. Also when iterating or hardening a skill across fixes, or grounding a skill's self-critique against its own run evidence — including a document-analysis skill (cap table, deck, financial model, transcript) that needs an uploaded file attached to be critiqued at all. NOT for generic unit testing (pytest/vitest of your own scripts) or non-Cowork CI. Covers the skill / run / chat / record

Live-lane probe — dispatches one sub-agent that must research with the WebSearch tool.

Live-lane probe — reads an attached uploaded file with the Read tool and quotes its marker line.

Internal harness probe for resume-continuity. Use when asked to run the resume-continuity probe, to load the probe skill, or to retrieve the vault passphrase.

fixture skill A — has a candidate source under scripts/

fixture skill B — sibling of skill-a, must be excluded when the target is skill-a directly

fixture skill for collectArtifactSources — proves scripts/ is reached

Normalize a CSV's EUR amounts to USD using live FX rates, with a graceful offline fallback. Use to convert or normalize a currency column in tabular data.

Compute summary statistics for a CSV and write a JSON + markdown report. Use when the user asks to analyze, summarize, profile, or compute metrics for a CSV / tabular dataset.

Example skill under test. Summarize a PDF and write action items.