Pi 1.5.1 adds shell auth, cache warming

// 51m agoPRODUCT UPDATE

Pi 1.5.1 adds shell auth, cache warming

ANNOUNCEMENT PRODUCT GITHUB PRODUCT HUNT X

Pi 1.5.1 adds shell-level provider login/logout with API-key and no-browser flows, letting developers manage credentials without launching an agent session. Its idle cache-warming mode keeps eligible prompt caches warm for up to 30 minutes between runs, reducing cold-cache costs.

// ANALYSIS

Pi is focusing on the operational details that make terminal agents pleasant to use: fast authentication and fast re-entry matter as much as model quality.

–Shell-native auth improves onboarding and SSH workflows, though the interactive-terminal requirement limits fully automated CI use.
–`--api-key` and `--no-browser` make provider setup more practical on headless machines.
–Idle warming should reduce latency and cache rewrites for bursty work, but refreshes still consume usage and can cost money.
–The update reinforces Pi’s strength as a customizable, multi-provider coding harness rather than a closed AI editor.

// TAGS

pi-coding-agentclicoding-agentai-codingagentinferenceopen-source

DISCOVERED

51m ago

2026-11-10

PUBLISHED

2h ago

2026-11-09

RELEVANCE

9/ 10

AUTHOR

PiChangelog

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

BENCHMARK36m ago

AGIBOT G2 Logs 1,730 Store Tasks

Six AGIBOT G2 wheeled humanoid robots completed 1,730 customer-service, food-service, restocking, and cleaning tasks during a 24-hour Shanghai convenience-store trial, with four passing the final assessment. The company’s [official report](https://www.agibot.com/article/231/detail/134.html) showcases real-world endurance while leaving intervention and autonomy rates undisclosed.

NEWS36m ago

Figure 03 Fleet Fills Figure HQ

Brett Adcock’s November 6–7 videos show dozens of Figure 03 humanoids roaming Figure’s headquarters. The fleet supports internal testing, burn-in, Helix training data collection, and limited commercial deployments.

UPDATE43m ago

Entire adds Claude Fable 5.2 review agents

Entire now lets developers select Claude Fable 5.2 for custom review agents in Trails, its pull-request alternative built around confidence evaluations. Agents score code changes against criteria such as security, slop, drift, or migration risk on every push.