Perplexity Computer adds Claude Code and GitHub CLI to its coding workflow
Original: You can now use Claude Code and GitHub CLI directly inside Perplexity Computer. View original →
Perplexity’s Computer product account used X on March 9, 2026 to show a new developer workflow: Claude Code and GitHub CLI can now run directly inside Perplexity Computer. In the public demo, the system took an open issue from Openclaw, forked the repository, wrote a plan, opened Claude Code to implement the fix, and then submitted a pull request through GitHub CLI. The post did not include a longer technical explainer, but the workflow shown is substantial on its own.
The announcement matters because it moves Perplexity Computer closer to a coding workstation rather than a general-purpose browser agent. Instead of only summarizing pages or clicking through websites, the system is being positioned to operate on real software tasks with repository state, command-line tooling, and contribution flows. That narrows the gap between flashy computer-use demos and actual software delivery work.
Two parts of the demo stand out. First, the agent did not stop at analysis; it moved from issue intake to planning, implementation, and PR submission. Second, the tooling choice is pragmatic. Claude Code and GitHub CLI already fit into existing developer workflows, so Perplexity is not asking engineers to adopt an entirely new proprietary environment. It is showing that familiar tools can be orchestrated inside the Computer surface.
There are still limits to what can be concluded from one X post. Perplexity did not publish public technical details in the linked materials about sandboxing, repository permissions, review controls, or how much human approval is required before commands execute. Those unanswered questions matter if the company wants this feature to move from an impressive demo to a trusted engineering workflow. For now, the public evidence supports the narrower claim that Perplexity Computer can invoke these tools inside a real coding task.
Even with that caution, the release is notable. A growing share of agent competition is moving from model quality alone to tool orchestration, environment control, and clean handoff into developer ecosystems. By showing Claude Code and GitHub CLI inside Perplexity Computer, Perplexity is signaling that future agent products will compete on how well they can inhabit existing software workflows, not just how well they can chat about them.
Related Articles
Google has released Gemini 3.5 Flash, optimized for agentic workflows and complex tasks. It claims 4x faster output than competing frontier models at under half the cost, with top-tier scores on Terminal-Bench, MCP Atlas, and reasoning benchmarks.
Alibaba's Qwen team has released Qwen3.7-Max, an agent-focused frontier LLM. It ranks 5th on Artificial Analysis's Intelligence Index, nearly matching GPT 5.4, and is available as both an API and open weights.
Forge is a new open-source Python framework that applies structured guardrails to self-hosted LLMs. The best config — Ministral-3 8B Q8 — jumps from a 53% baseline to 86.5% on the 26-scenario eval suite, with 99% achievable on agentic tasks.