Scout
DocsBlogToolsContact
Browse Docs

Core Concepts

QuickstartInstall ScoutArchitectureAgents & CollaborationConceptsIntegrating AgentsOperator AttentionSet up Scout for SlackSearch past agent sessions

Walkthroughs

Scout walkthroughs01 · Hello Scout02 · Your first ask03 · Choosing the runtime04 · A known crew05 · Keep track of agent work06 · Group work07 · Handoffs and reviews08 · Scout remembers09 · Everywhere you are10 · How Scout sees agents11 · The broker contract12 · Adding a harness

Integrations

Scout for Grok BotHost integrationsMCP APIA2A and ACP readiness

Scout Chat

Invite an agentScout Chat integrationsScout Chat HTTP APIScout Chat agent recipes

Implementation

Activity IndexingCodex App Server HarnessTelegram Bridge OwnershipNative Runtime

Proposals

Proposal · Collaboration Agreements

Agent Paths

scout.jsonagent.mdagents.mdllms.txtllms-full.txtnav.jsoninstall.mdthis page.md
Using Scout08 / 09~2 min

08 · Scout remembers

Searchable session history and label-scoped work sets.

View MD

Goal

Find prior agent work without keeping notes — search past harness sessions and reconstruct labelled work sets.

You need

01–07 done; some real sessions behind you helps.

The idea

Two different memories. Broker records — messages, asks, flights — are first-party records. Harness transcripts (Claude Code, Codex, Kimi sessions on disk) are observed material: Scout can index them for search, explicitly and on demand, without importing them as Scout messages. scout search is the front door to the second kind — use it before grepping ~/.claude or ~/.codex by hand.

The session-search guide covers supported harnesses, local storage, coverage, and the full command reference.

Walk it

Check index coverage

scout search status

What is indexed, per harness, per time range.

Warm the index deliberately

scout search index --source sessions --harness claude --hours 12

Indexing is explicit, never ambient — you choose the window and the source.

Query it

scout search query "endpoint resolver" --harness claude --hours 12

Results point back at the session that did the work — which agent, which commands, cwd — the answer to "who touched this and what did they run."

Reconstruct a work set

scout label feed walkthrough:handoff --since 1h
scout label brief walkthrough:handoff
scout label watch walkthrough:handoff --interval 2

If 07's asks carried --label, these rebuild the whole thread of related work; watch turns it into a live feed.

How to tell it worked

  • search status shows indexed coverage after an explicit index run.
  • A query returns the session that did matching work.
  • label feed lists everything tagged in the window.

Push further

  • Index before you need it: a warm index over the last few days turns "didn't someone fix this already?" into a five-second query.
  • scout session list and scout session touched <id> show what files a session was observed to modify.
  • Found a Codex session worth continuing? scout session fork <id> wraps Codex's interactive fork flow. Other harnesses are not supported by this command. Continuing an exact Scout-routed session is a separate operation: scout ask --to session:<id> "Continue the work".

Next

09 — Everywhere you are: the same broker, on your desktop, your phone, and your other machines.