Source: https://internal-agents.com/agents/slack-context-system # Slack — Multi-agent context system Slack's engineering team runs an agentic service that investigates security alerts. A Director agent orchestrates rounds, Expert agents gather evidence, and a Critic scores each finding for credibility. Instead of message history, three written channels carry the state between rounds. > This entry describes supporting infrastructure that other work builds on. The catalog classifies it as a supporting pattern. The record also reports a workflow. - Approach type: Supporting pattern - Deployment stage: Research - Autonomy: Human in loop - Evidence strength: Mixed - Status: Internal - First reported year: 2026 - Work: Research - Invocation: Unknown - Entry reviewed: 2026-09-15 ## Where people stay involved - **long-running investigation → synthesized report** — Unknown · Level unknown ## Overview ### Summary Slack's engineering team runs an agentic service that investigates security alerts. A Director agent orchestrates rounds, Expert agents gather evidence, and a Critic scores each finding for credibility. Instead of message history, three written channels carry the state between rounds. Fact · Reported · High confidence · `slack-context-system--summary` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ## How it works ### Supporting component Structured working memory: findings, decisions, questions, hypotheses Fact · Reported · High confidence · `slack-context-system--primitives-0` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ### Supporting component A truth filter with credibility scores over submitted findings Fact · Reported · High confidence · `slack-context-system--primitives-1` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ### Supporting component A chronological, deduped, conflict-resolved synthesis retained across steps Fact · Reported · High confidence · `slack-context-system--primitives-2` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ## Supervision evidence ### Operating model assessment Unclassified for long-running investigation → synthesized report; human attention boundary: unknown. Inference · Catalog judgment · Unverified confidence · `slack-context-system--operating-models-0` Confidence reason: The research pattern documents agent coordination and criticism but not the normal human attention boundary. Qualifications: - Observation date: 2026 Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) ## Implementation details ### Harness Coordinator/dispatcher: a central coordinator dispatches to expert agents and to critic agents Fact · Reported · High confidence · `slack-context-system--architecture-harness` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ### Model Not specified Fact · Reported · High confidence · `slack-context-system--architecture-model` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ### Tool access Expert agents produce reports; critic agents evaluate them using evidence-inspection tools Fact · Reported · High confidence · `slack-context-system--architecture-tool-access` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ### Context management Three channels; Director's Journal (working memory), Critic's Review (credibility-weighted findings), Critic's Timeline (deduped chronological synthesis) Fact · Reported · High confidence · `slack-context-system--architecture-context-mgmt` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ## Reported metrics ### Headline claim Context management for security investigations spanning hundreds of inference requests and megabytes of output Metric · Reported · High confidence · `slack-context-system--headline-metric` Confidence reason: A linked first-party source states the claim. Qualifications: - The source does not report the denominator of this figure. - Reported by: Slack - Scope: Complex security investigations requiring tailored multi-agent context; qualitative workload scale Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) · Preserved content.md, lines 34–38 - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) · Preserved content.md, lines 22 ### Key observation Handles multi-agent runs spanning hundreds of requests and megabytes of output Metric · Reported · High confidence · `slack-context-system--key-metrics-0` Confidence reason: A linked first-party source states the claim. Qualifications: - The source does not report the denominator of this figure. - Reported by: Slack - Scope: Complex security investigations requiring tailored multi-agent context; qualitative workload scale Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) · Preserved content.md, lines 34–38 - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) · Preserved content.md, lines 22 ## Lessons and interpretation ### Lesson Don't pass all information at every step; build structured summaries agents can reliably build on Inference · Catalog judgment · Medium confidence · `slack-context-system--lessons-learned-0` Confidence reason: The catalog derives this observation from the linked sources. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ### Lesson Separate expert agents (produce) from critic agents (evaluate); corroborated findings are prioritized Inference · Catalog judgment · Medium confidence · `slack-context-system--lessons-learned-1` Confidence reason: The catalog derives this observation from the linked sources. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ### Lesson Context management becomes its own subsystem once runs get long Inference · Catalog judgment · Medium confidence · `slack-context-system--lessons-learned-2` Confidence reason: The catalog derives this observation from the linked sources. Evidence: - Supports · [1] [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-1/content.md) - Supports · [2] [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/slack-context-system-source-2/content.md) ## Sources 1. [Managing context in long-running agentic applications](https://slack.engineering/managing-context-in-long-run-agentic-applications/) - Engineering blog · First party · Evidence - Original URL: - Accessed: 2026-08-12 · Last verified: 2026-08-31 - Preserved copy in the repository: 2. [How Slack manages context in long-running multi-agent systems](https://www.infoq.com/news/2026/04/slack-agent-context-management/) - News · Independent secondary · Evidence - Original URL: - Accessed: 2026-08-12 · Last verified: 2026-08-31 - Preserved copy in the repository: