Search BrigadeStack

Search BrigadeStack pages and docs.

BrigadeStack
Menu
Docs menu

Docs / Reference

Extras

Commands for chat exports, a self-check run, and ingesting handoffs from several repos.

Run brigadestack extras on once to enable these commands, and brigadestack extras status to see whether they are enabled.

brigadestack extras on
brigadestack extras status

brigadestack extras off hides them again. The sections below cover three of them: chat, dogfood and repos ingest.

Chat exports

Use brigadestack chat when you keep chat exports on disk and want the actionable items in them to reach your task inbox. It is local only. BrigadeStack reads files you point it at, validates safe finding summaries, and routes actionable items into the existing scanner inbox. It does not call live chat APIs, perform OAuth, send webhooks, run a daemon, edit memory or promote imports automatically.

Configure the exports

brigadestack chat surfaces manages the local config in the gitignored file .brigade/chat-surfaces.toml.

  • brigadestack chat surfaces init writes the file with example entries.
  • brigadestack chat surfaces list, show <surface-id>, and doctor inspect local export paths, providers, privacy mode, evidence policy, confidence thresholds, and stale output health.
brigadestack chat surfaces init
brigadestack chat surfaces list
brigadestack chat surfaces show discord-export
brigadestack chat surfaces doctor

Each entry in the file includes id, provider, workspace_label, channel_label, export_path, sweep_output_path, enabled, privacy_mode, evidence_policy and confidence_threshold. Supported provider families are discord-export, slack-export, telegram-export, clickclack-export, and generic-jsonl.

Validate and import findings

  • brigadestack chat sweep validate <path> checks a local export finding file without writing.
  • brigadestack chat sweep ingest <surface-id> normalizes a configured export into .brigade/chat-memory-sweeps/<surface-id>-latest.json.
  • brigadestack chat sweep import-issues <surface-id> imports normalized actionable findings into the work inbox (.brigade/work/imports/inbox.jsonl), creating it if needed, with source chat-memory-sweep. Imports get stable fingerprints and stay dismissed until the finding changes.
brigadestack chat sweep validate .brigade/chat-surfaces/discord-export.json
brigadestack chat sweep ingest discord-export
brigadestack chat sweep import-issues discord-export

Export findings should contain provider, surface_id, channel_label, thread_label, message_range_label, issue_id, issue_type, priority, confidence, safe_summary, evidence_summary, suggested_task_text, acceptance_criteria and source_fingerprint.

Privacy boundary

Raw exports and normalized output stay under .brigade/ and should remain gitignored. Raw private chat fields are rejected by default, including raw_text, raw_messages, messages, message_text, quotes, and transcript. Use safe summaries, labels, message ranges, local evidence paths, and confidence values.

Self-check

brigadestack dogfood runs BrigadeStack on itself or on another trusted repo, so you can see the tool work on real code. It uses a built-in single-agent roster (Codex by default, selectable with --agent-cli), a prompt-level read-only policy, normal run artifacts, a default Memory Handoff, and an artifact summary.

Set it up once:

brigadestack dogfood init --target /path/to/repo

That writes local defaults to .brigade/dogfood.toml, which is gitignored because it stores machine-local paths and preferences. New configs default handoffs to .codex/memory-handoffs/ because the roster is Codex-driven. Pass --handoff-inbox to write them somewhere else.

Daily commands:

brigadestack dogfood
brigadestack dogfood status
brigadestack dogfood next
brigadestack dogfood --target /path/to/repo
  • brigadestack dogfood runs the configured daily path from the repo.
  • brigadestack dogfood "review today's changes" overrides only the task.
  • brigadestack dogfood status checks paths, sandbox mode, CLI availability, ignore rules, and the latest run.
  • brigadestack dogfood latest shows the latest configured run.
  • brigadestack dogfood next prints the latest extracted next step.

It writes summary.md beside each run’s JSON artifacts when a final answer or next step exists, and defaults to a 600 second per-agent timeout. By default runs use a prompt-level read-only policy with trusted-workspace execution. The native read-only sandbox applies only with --native-read-only-sandbox. Run artifacts are the same ones described on the run page.

Useful switches:

  • --agent-cli selects the agent CLI: codex, claude, opencode, antigravity, pi, cursor, or ollama:. Defaults to codex.
  • --no-handoff skips the handoff.
  • --no-inspect skips the artifact summary.
  • --native-read-only-sandbox uses Codex’s native read-only sandbox instead of the trusted-workspace default.

Ingest handoffs from several repos

If you keep handoffs in more than one repo, brigadestack repos ingest goes through every repo registered in the fleet config, routes each repo’s handoffs into the workspace you point --target at, and archives the processed handoffs back in the source repo. Run it from the workspace that holds the fleet config, or point --target at that workspace.

brigadestack repos ingest --target .
brigadestack repos ingest --target . --apply

It is a dry run by default and --apply writes the changes. --no-promote-cards skips automatic card promotion and --no-route-documents skips automatic document routing. Add --json for machine-readable output. For the handoff files themselves, see handoff.