Telexed

AI news through a solo-operator lens — only what changes your day14 of 412

FILTER[All][Agents & tools][Models & API][Generative media][Infra & SaaS][ASO & growth][Indie business][Idea signals][Other][★6+ high-signal]

GitHub Changelog ✕clear filters

Yesterday2 dispatches

#0412
#0412Agents & tools GitHub Changelog23 hours ago
GitHub adds REST API auditing for `Copilot` cloud agent repo config
60radar
Repo-level agent settings can now be checked by API instead of manual UI review. Useful for keeping automation permissions visible before cloud agents touch production code.
- New endpoint: Get Copilot cloud agent configuration for a repository, currently in public preview.
- Best fit is policy drift checks across repos: scan whether agent access and configuration match your expected defaults.
- This is governance plumbing, not a coding-speed feature. Worth adopting if Copilot agents run on real repos.
Source: github.blog/changelog/2026-05-18-audit-repository-copiloRead original →
FIG-4121:1
60radar
FIG-4121:1
#0411
#0411Agents & tools GitHub Changelog24 hours ago
`Copilot Spaces API` is now generally available
70radar
Copilot SpacesGitHub Copilot feature — manages task-specific context spaces
Spaces can now be managed from your own apps via API. Useful for wiring repo context into internal tools or repeatable agent workflows.
- The API supports create, read, update, and delete for Spaces, so context setup no longer has to stay inside GitHub UI.
- Good fit for templates: bootstrap a project space per repo, customer, or feature branch and keep agent context consistent.
- This is more automation surface than end-user feature. Value depends on whether Copilot Spaces is already part of the coding workflow.
Source: github.blog/changelog/2026-05-18-copilot-spaces-api-now-Read original →
FIG-4111:1
70radar
FIG-4111:1

Mon, May 181 dispatches

#0410
#0410Agents & tools GitHub Changelogyesterday
`Copilot CLI` Remote Control Is GA on Mobile, Web, and VS Code
80radar
Terminal-started agent work can now keep moving from phone, browser, or editor. This turns CLI sessions into portable async coding jobs, useful when builds or fixes outlive one desk session.
- Remote control is now generally available across GitHub Mobile, github.com, and VS Code; less reason to keep one terminal window as the control plane.
- The workflow starts in the terminal, then continues elsewhere. Long-running refactors, CI fixes, and review follow-ups fit better than one-shot prompts.
- Copilot CLI is moving from local command tool toward persistent agent session UX. Expect more competition around cross-device coding control.
Source: github.blog/changelog/2026-05-18-remote-control-for-copiRead original →
FIG-4101:1
80radar
FIG-4101:1

Sat, May 162 dispatches

#0409
#0409Agents & tools GitHub Changelog4 days ago
`Grok Code Fast 1` deprecated across GitHub Copilot
50radar
Grok Code Fast 1Code LLM — offered as a selectable `GitHub Copilot` model
Anyone who picked Grok inside GitHub Copilot needs to switch now. The change is immediate across chat, edits, agent mode, and completions, so workflow stability matters more than model preference here.
- The deprecation took effect on May 15, 2026, not as a future notice, so affected setups break now.
- Scope covers all Copilot surfaces: chat, inline edits, ask, agent mode, and code completions.
- This is a reminder not to hard-wire one optional model into daily coding flows when the host tool controls availability.
Source: github.blog/changelog/2026-05-15-grok-code-fast-1-deprecRead original →
FIG-4091:1
50radar
FIG-4091:1
#0408
#0408Agents & tools GitHub Changelog4 days ago
`GitHub Copilot Memory` adds user preferences for `Pro` and `Pro+`
50radar
Memory now starts carrying personal coding preferences across sessions. It is early access and limited to paid tiers, but it points to a more opinionated default pair-programming workflow.
- The update adds user-level preferences, so behavior can persist beyond a single chat or task.
- Preferences can be both explicitly stated and inferred, which should reduce repeated style or workflow instructions.
- Access is limited to Copilot Pro and Copilot Pro+ in early access, so rollout is still narrow.
- This pushes GitHub Copilot closer to a personalized agent, not just a stateless autocomplete tool.
Source: github.blog/changelog/2026-05-15-copilot-memory-supportsRead original →
FIG-4081:1
50radar
FIG-4081:1

Fri, May 151 dispatches

#0407
#0407Agents & tools GitHub Changelog5 days ago
`GitHub Copilot app` enters technical preview as a desktop agent workspace
70radar
GitHub Copilot appDesktop coding agent app — starts from GitHub context
GitHub is turning Copilot into a GitHub-native desktop workspace for isolated agentic coding. If the handoff from issue to execution is smooth, this could become a serious alternative to browser-first agent flows.
- Positioning is clear: a native desktop surface tied to GitHub work, not just an editor add-on.
- The promise is isolated execution and steering, which matters when testing agent runs without polluting the main repo state.
- Technical preview means the direction is notable now, but reliability and workflow depth still need proof.
- If it reduces friction from issue, PR, or repo context into action, it can compete with standalone coding agents.
Source: github.blog/changelog/2026-05-14-github-copilot-app-is-nRead original →
FIG-4071:1
70radar
FIG-4071:1

Thu, May 142 dispatches

#0406
#0406Agents & tools GitHub Changelog5 days ago
`GitHub Copilot` cloud agent adds `Auto` model selection
50radar
Model choice can now be delegated to Auto, which picks from available models using health and capability signals. Useful for quicker starts, but manual selection still matters when cost or output style is critical.
- Auto in the model picker now works with the cloud agent, removing one setup choice before a run.
- Selection is based on system health and model fit, so routing may change without user intervention.
- This is a workflow speed win, not a capability jump; output quality still depends on whatever model gets picked.
- Teams that care about predictable cost, latency, or model-specific behavior will still want manual pinning.
Source: github.blog/changelog/2026-05-14-copilot-cloud-agent-supRead original →
FIG-4061:1
50radar
FIG-4061:1
#0405
#0405Agents & tools GitHub Changelog6 days ago
`GitHub Copilot` for JetBrains adds `CLI agent` and unified sessions view
60radar
Agent-style work now sits inside JetBrains, with live visibility into running and queued sessions. Useful if you already code in JetBrains and want less context-switching.
- Copilot CLI agent now runs inside JetBrains IDEs, bringing terminal-style agent flows into the editor.
- A unified sessions view shows live status for both running and queued sessions, so parallel work is easier to track.
- This is a workflow upgrade, not a model launch; value comes from tighter IDE control and less tool-hopping.
Source: github.blog/changelog/2026-05-13-introducing-copilot-cliRead original →
FIG-4051:1
60radar
FIG-4051:1

Wed, May 132 dispatches

#0404
#0404Agents & tools GitHub Changelog7 days ago
`GitHub Copilot` code review adds grouped suggestions and severity labels
50radar
GitHub CopilotAI coding tool — grouped PR review suggestions, severity
Review feedback is faster to triage instead of reading comments one by one. If you use GitHub PRs heavily, this trims review friction now; not a workflow reset, but a solid quality-of-life upgrade.
- The rollout is available to users already opted into the new pull requests experience, so adoption depends on that UI switch first.
- Suggestions are grouped and severity is surfaced together, which should cut the time spent sorting signal from noise in long review threads.
- This matters most on PRs where Copilot leaves multiple related comments; batching context lowers reviewer context-switch cost.
Source: github.blog/changelog/2026-05-12-copilot-code-review-comRead original →
FIG-4041:1
50radar
FIG-4041:1
#0403
#0403Agents & tools GitHub Changelog7 days ago
`GitHub Copilot` April usage reports now map activity to `AI credits`
50radar
April usage can now be translated into AI credits before billing changes land. Useful for estimating Copilot spend; actionable, but still incomplete without final rate details.
- You can download a usage report starting now and see how April activity converts to AI credits before billing flips.
- This is a pre-billing visibility feature, not a pricing announcement, so it helps with budgeting more than product evaluation.
- Solo seats using heavier chat or agent workflows get an early warning on whether Copilot usage will stay cheap or drift upward.
Source: github.blog/changelog/2026-05-12-april-reports-are-now-aRead original →
FIG-4031:1
50radar
FIG-4031:1

Fri, May 82 dispatches

#0402
#0402Agents & tools GitHub Changelog2 weeks ago
`GitHub Copilot` cloud agent gets more flexible secrets and variables
50radar
GitHub CopilotCoding AI — delegates tasks to an Actions agent
Background tasks can now receive secrets and variables inside their own GitHub Actions-backed dev environment. That makes real API calls and env-specific automation easier, though the update looks incremental rather than workflow-changing.
- Delegated work runs in a separate environment backed by GitHub Actions, not your local machine.
- Passing secrets and variables closes a common gap for tasks that need API keys, tokens, or config at runtime.
- This should reduce handoff friction for deploy, integration, and automation jobs, but the snippet includes no concrete limits or scope.
Source: github.blog/changelog/2026-05-08-more-flexible-secrets-aRead original →
FIG-4021:1
50radar
FIG-4021:1
#0401
#0401Agents & tools GitHub Changelog2 weeks ago
`GitHub Copilot` deprecates `Claude Sonnet 4` across all experiences
60radar
One model option disappeared across chat, inline edits, agent mode, and completions on May 6, 2026. If your Copilot output quality depended on it, prompt behavior and fallback model choice need a quick recheck.
- Coverage is broad: Copilot Chat, inline edits, ask, agent mode, and code completions all lose the same model path.
- The change landed on May 6, 2026, so teams using fixed Copilot habits may see response style shifts immediately.
- This is a tooling workflow issue, not just model news; saved prompts and comparison baselines can drift after removal.
Source: github.blog/changelog/2026-05-07-claude-sonnet-4-deprecaRead original →
FIG-4011:1
60radar
FIG-4011:1

Thu, May 71 dispatches

#0400
#0400Agents & tools GitHub Changelog2 weeks ago
`Rubber Duck` in `GitHub Copilot CLI` adds broader model support
50radar
Rubber DuckReview agent — cross-model critique inside Copilot CLI
Review no longer has to stay inside one model family. Rubber Duck can pair a GPT session with a Claude critic, which makes mixed-model checks easier and reduces tool lock-in.
- Rubber Duck is positioned as a cross-family review agent inside GitHub Copilot CLI, not just a single-model helper.
- A session running a GPT model can now invoke a Claude-powered critic agent. That opens a built-in second opinion path.
- The practical win is workflow simplicity: compare generation and critique across model families without leaving the CLI.
Source: github.blog/changelog/2026-05-07-rubber-duck-in-github-cRead original →
FIG-4001:1
50radar
FIG-4001:1

Wed, May 61 dispatches

#0399
#0399Agents & tools GitHub Changelog2 weeks ago
`GitHub MCP Server` secret scanning is now generally available
60radar
GitHub MCP ServerMCP server — GitHub workflows with built-in secret scanning
Secret scanning now lands directly inside MCP-based coding workflows. If you rely on agentic IDEs or CLIs, this is an immediate security upgrade with almost no adoption friction.
- GitHub MCP Server now supports secret scanning in GA, so the feature is no longer experimental.
- The target surface is MCP-compatible AI tools, including examples like GitHub Copilot CLI and agent-friendly IDE flows.
- This pushes security checks closer to where code is generated, which matters more as more edits originate from agents.
Source: github.blog/changelog/2026-05-05-secret-scanning-with-giRead original →
FIG-3991:1
60radar
FIG-3991:1

GitHub adds REST API auditing for `Copilot` cloud agent repo config

`Copilot Spaces API` is now generally available

`Copilot CLI` Remote Control Is GA on Mobile, Web, and VS Code

`Grok Code Fast 1` deprecated across GitHub Copilot

`GitHub Copilot Memory` adds user preferences for `Pro` and `Pro+`

`GitHub Copilot app` enters technical preview as a desktop agent workspace

`GitHub Copilot` cloud agent adds `Auto` model selection

`GitHub Copilot` for JetBrains adds `CLI agent` and unified sessions view

`GitHub Copilot` code review adds grouped suggestions and severity labels

`GitHub Copilot` April usage reports now map activity to `AI credits`

`GitHub Copilot` cloud agent gets more flexible secrets and variables

`GitHub Copilot` deprecates `Claude Sonnet 4` across all experiences

`Rubber Duck` in `GitHub Copilot CLI` adds broader model support

`GitHub MCP Server` secret scanning is now generally available