Source: https://internal-agents.com/agents/coinbase-forge-mux # Coinbase — Forge / Mux Forge, started as Claudebot, turns a Slack/GitHub/Linear discussion into a Linear issue, fix, PR, and one-off build; Mux lets employees run many coding agents concurrently; as of February 2026, 5% of all PRs merged company-wide come from these background agents. - Approach type: Agent system - Deployment stage: Scaled - Autonomy: Drafts reviewed - Evidence strength: Detailed primary - Status: Internal - First reported year: 2026 - Work: Coding, Code review - Interfaces: Slack, Github, Linear - Invocation: Event driven, Interactive - Entry reviewed: 2026-09-15 ## Where people stay involved - **Slack, GitHub, or Linear request → reviewed pull request and build** — Work product review · Level 3 ## Overview ### Summary Forge, started as Claudebot, turns a Slack/GitHub/Linear discussion into a Linear issue, fix, PR, and one-off build; Mux lets employees run many coding agents concurrently; as of February 2026, 5% of all PRs merged company-wide come from these background agents. Fact · Reported · High confidence · `coinbase-forge-mux--summary` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) - Supports · [2] [Coinbase × Linear (Forge workflow)](https://linear.app/customers/coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-2/content.md) - Supports · [3] [Chintan Turakhia: 5% of all PRs merged now come from our in-house background agents](https://x.com/chintanturakhia/status/2019921786834411728) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-3/content.md) · Preserved content.md, line 11 - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, lines 114–126 ## How it works ### Supporting component Custom harness: Slack bug discussion → Linear issue → fix → PR → one-off build Fact · Reported · High confidence · `coinbase-forge-mux--primitives-0` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [2] [Coinbase × Linear (Forge workflow)](https://linear.app/customers/coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-2/content.md) - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, lines 104–118 ### Supporting component Concurrency layer; one human coordinates many isolated coding agents, each in its own worktree/branch/terminal Fact · Reported · High confidence · `coinbase-forge-mux--primitives-1` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) ## Supervision evidence ### Operating model assessment Level 3 for Slack, GitHub, or Linear request → reviewed pull request and build; human attention boundary: work-product-review. Inference · Catalog judgment · High confidence · `coinbase-forge-mux--operating-models-0` Confidence reason: The source describes delegated implementation that returns a pull request and build for human review. Qualifications: - Observation date: 2026 Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) - Supports · [3] [Chintan Turakhia: 5% of all PRs merged now come from our in-house background agents](https://x.com/chintanturakhia/status/2019921786834411728) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-3/content.md) · Preserved content.md, line 11 ## Implementation details ### Sandbox Mux gives each concurrent agent its own git worktree, branch, and terminal Fact · Reported · High confidence · `coinbase-forge-mux--architecture-sandbox` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) ### Harness Portfolio approach; Claude Code, OpenCode, Cursor, and Copilot rather than one harness; Forge is a custom Slack-native harness with the same tools and context as engineers, invokable from Slack, GitHub, and Linear; built in-house because security requirements blocked cloud background agents Fact · Reported · High confidence · `coinbase-forge-mux--architecture-harness` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) - Supports · [2] [Coinbase × Linear (Forge workflow)](https://linear.app/customers/coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-2/content.md) - Supports · [3] [Chintan Turakhia: 5% of all PRs merged now come from our in-house background agents](https://x.com/chintanturakhia/status/2019921786834411728) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-3/content.md) · Preserved content.md, line 11 - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, line 120 ### Model Portfolio across multiple providers; the Slack agent Claude bot can use multiple underlying models despite its name Fact · Reported · High confidence · `coinbase-forge-mux--architecture-model` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, line 114 ### Interfaces slack, github, linear Fact · Reported · High confidence · `coinbase-forge-mux--architecture-interfaces` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) ### Tool access Linear treated as the structured product context / source of truth Fact · Reported · High confidence · `coinbase-forge-mux--architecture-tool-access` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [2] [Coinbase × Linear (Forge workflow)](https://linear.app/customers/coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-2/content.md) - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, line 120 ### Knowledge Linear as the durable structured-context layer for product work Fact · Reported · High confidence · `coinbase-forge-mux--architecture-knowledge` Confidence reason: A linked first-party source states the claim. Evidence: - Supports · [2] [Coinbase × Linear (Forge workflow)](https://linear.app/customers/coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-2/content.md) ## Reported metrics ### Headline claim 5% of all PRs merged come from in-house background agents (Feb 2026) Metric · Reported · Medium confidence · `coinbase-forge-mux--headline-metric` Confidence reason: Self-reported share stated by the engineering leader who built the harness; measurement method and period not documented. A thread reply says the share covers the whole company. Qualifications: - Reported by: Coinbase - Scope: PRs merged by in-house background agents - Denominator: All PRs merged company-wide - Observation date: 2026-02 Evidence: - Supports · [3] [Chintan Turakhia: 5% of all PRs merged now come from our in-house background agents](https://x.com/chintanturakhia/status/2019921786834411728) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-3/content.md) · Preserved content.md, line 11 ### Key observation Mux: 600+ users including engineers, PMs, and designers (335 active, 197 power users) Metric · Reported · High confidence · `coinbase-forge-mux--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: Coinbase - Scope: Registered Mux users including engineers, PMs, and designers; 335 active and 197 power users Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) · Preserved content.md, lines 24–34 ### Key observation 5% of all PRs merged come from in-house background agents (Feb 2026) Metric · Reported · Medium confidence · `coinbase-forge-mux--key-metrics-1` Confidence reason: Self-reported share stated by the engineering leader who built the harness; measurement method and period not documented. A thread reply says the share covers the whole company. Qualifications: - Reported by: Coinbase - Scope: PRs merged by in-house background agents - Denominator: All PRs merged company-wide - Observation date: 2026-02 Evidence: - Supports · [3] [Chintan Turakhia: 5% of all PRs merged now come from our in-house background agents](https://x.com/chintanturakhia/status/2019921786834411728) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-3/content.md) · Preserved content.md, line 11 ### Key observation Wallet rewrite PR review times reduced from 150 hours to 15 hours (self-reported; period and method not documented) Metric · Reported · Low confidence · `coinbase-forge-mux--key-metrics-2` Confidence reason: The episode description states the reduction without a date, denominator, or measurement method. Qualifications: - The source does not report the denominator of this figure. - Reported by: Coinbase - Scope: PR review times for the wallet rewrite effort - Observation date: 2026-03 Evidence: - Supports · [4] [How Coinbase scaled AI to 1,000+ engineers (How I AI, with Chintan Turakhia)](https://www.lennysnewsletter.com/p/how-coinbase-scaled-ai-to-1000-engineers) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-4/content.md) · Preserved content.md, line 16 ### Key observation Speed run: about 100 participants opened about 70 PRs in 15 minutes (opened, not merged; single event) Metric · Reported · Medium confidence · `coinbase-forge-mux--key-metrics-3` Confidence reason: Described by the organizer as a single event; explicitly pull requests opened, not merged. Qualifications: - The source does not report the denominator of this figure. - Reported by: Coinbase - Scope: Pull requests opened during one speed-run session - Observation date: 2026-03 Evidence: - Supports · [4] [How Coinbase scaled AI to 1,000+ engineers (How I AI, with Chintan Turakhia)](https://www.lennysnewsletter.com/p/how-coinbase-scaled-ai-to-1000-engineers) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-4/content.md) · Preserved content.md, line 23 - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, line 170 ## Lessons and interpretation ### Lesson Support a portfolio of harnesses (Claude Code, OpenCode, Cursor, Copilot) rather than standardizing on one Inference · Catalog judgment · Medium confidence · `coinbase-forge-mux--lessons-learned-0` Confidence reason: The catalog derives this observation from the linked sources. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) ### Lesson Keep the system of record (Linear) as the agent's structured context; conversation can be the input, Linear stays durable Inference · Catalog judgment · Medium confidence · `coinbase-forge-mux--lessons-learned-1` Confidence reason: The catalog derives this observation from the linked sources. Evidence: - Supports · [2] [Coinbase × Linear (Forge workflow)](https://linear.app/customers/coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-2/content.md) ### Lesson Simple per-agent worktree/branch/terminal isolation is a pragmatic alternative to full sandboxing for concurrent coding Inference · Catalog judgment · Medium confidence · `coinbase-forge-mux--lessons-learned-2` Confidence reason: The catalog derives this observation from the linked sources. Evidence: - Supports · [1] [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-1/content.md) ### Lesson Security requirements that blocked cloud background agents motivated the in-house, Slack-native harness build Fact · Reported · High confidence · `coinbase-forge-mux--lessons-learned-3` Confidence reason: The episode writeup states the security requirement as the reason for the in-house build. Qualifications: - Observation date: 2026-03 Evidence: - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, line 120 ### Lesson Give background agents the same tools and context as engineers through a Slack-native interface Inference · Catalog judgment · Medium confidence · `coinbase-forge-mux--lessons-learned-4` Confidence reason: The catalog derives this design guidance from the builder's description of the harness. Qualifications: - Observation date: 2026 Evidence: - Supports · [3] [Chintan Turakhia: 5% of all PRs merged now come from our in-house background agents](https://x.com/chintanturakhia/status/2019921786834411728) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-3/content.md) · Preserved content.md, line 11 ### Lesson Spread adoption through visible team experiments, such as PR speed runs and shared Slack channels, rather than mandates Inference · Catalog judgment · Medium confidence · `coinbase-forge-mux--lessons-learned-5` Confidence reason: The catalog derives this adoption guidance from the episode accounts of speed runs and shared channels. Qualifications: - Observation date: 2026 Evidence: - Supports · [4] [How Coinbase scaled AI to 1,000+ engineers (How I AI, with Chintan Turakhia)](https://www.lennysnewsletter.com/p/how-coinbase-scaled-ai-to-1000-engineers) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-4/content.md) · Preserved content.md, line 25 - Supports · [5] [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/coinbase-forge-mux-source-5/content.md) · Preserved content.md, lines 170–172 ## Sources 1. [Coding had a concurrency problem: how Mux helped solve it](https://www.coinbase.com/de/blog/coding-had-a-concurrency-problem-how-mux-helped-solve-it) - Engineering blog · First party · Evidence - Original URL: - Accessed: 2026-08-12 · Last verified: 2026-08-31 - Preserved copy in the repository: 2. [Coinbase × Linear (Forge workflow)](https://linear.app/customers/coinbase) - Case study · First party · Evidence - Original URL: - Accessed: 2026-08-12 · Last verified: 2026-08-31 - Preserved copy in the repository: 3. [Chintan Turakhia: 5% of all PRs merged now come from our in-house background agents](https://x.com/chintanturakhia/status/2019921786834411728) - Social post · Direct participant · Evidence - Original URL: - Published: 2026-02-06 · Accessed: 2026-09-15 · Last verified: 2026-09-15 - Preserved copy in the repository: 4. [How Coinbase scaled AI to 1,000+ engineers (How I AI, with Chintan Turakhia)](https://www.lennysnewsletter.com/p/how-coinbase-scaled-ai-to-1000-engineers) - Podcast · Direct participant · Evidence - Original URL: - Publisher: Lenny's Newsletter · Published: 2026-03-02 · Accessed: 2026-09-15 · Last verified: 2026-09-15 - Preserved copy in the repository: 5. [How I AI: Chintan Turakhia's playbook for AI adoption at Coinbase](https://www.chatprd.ai/how-i-ai/playbook-for-ai-engineering-adoption-at-coinbase) - Podcast · Direct participant · Evidence - Original URL: - Publisher: ChatPRD · Published: 2026-03-02 · Accessed: 2026-09-15 · Last verified: 2026-09-15 - Preserved copy in the repository: