brigade-work

# brigade-work Brigade only earns its keep if real work flows through it. Installed and never used, it is dead weight: the daily brief is empty, the outcome ledger never fills, and `outcome rank` says "ranking: none". This skill is the loop that feeds it. Check what is pending, run your verification *through* Brigade so it produces a signed receipt, record the outcome against the skill or card that did the work, export receipts so the next run gets an evidence brief, and hand off what you learned. **Core principle:** the model never grades its own work. A real exit code captured as a Brigade verify or run receipt is the only signal that counts. Route verification through Brigade, capture the outcome, feed receipts forward via MiseLedger when installed, and the loop is honest and measured (`context_eval.brief_hit_rate`). ## When this applies The repo or workspace is Brigade-wired: a `.brigade/` directory exists, or `brigade status --target .` succeeds. If neither is true, this skill does not apply (run `brigade init` first only if the user wants Brigade here). ## The loop ### 1. Start: see what is pending At the start of a work session, before deciding what to do: ```bash brigade work brief --target . # pending work, imports, warnings, suggested next command ``` In a memory/operator workspace, `brigade daily status --target .` is the lighter equivalent. Read it first. ### 2. Verify THROUGH Brigade, not raw When you would run tests or a check to confirm a change, ...