Telexed

AI news through a solo-operator lens — only what changes your day30 of 829

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

clear filters

Yesterday4 dispatches

#0829
#0829Agents & tools Paperclip Releases1 hour ago
`Paperclip` `v2026.529.0` adds document annotations and company skills CLI
70radar
PaperclipCoding agent platform — document and skill workflows
Agent work now gets passage-level review threads and managed skill catalogs. Useful if you run self-hosted agent workflows; otherwise it is a strong product-direction signal.
- Issue documents now support revision-aware inline annotation threads with stable anchors, reducing vague review comments around agent-written specs.
- New skills CLI covers install, reset, audit, export, and assignment; skill provenance is tracked across bundled, catalog, runtime, and adapter sources.
- Fresh private self-hosted installs get a one-time first-admin browser claim, removing the invite bootstrap problem on LAN deployments.
- Claude Local can refresh its Anthropic model catalog from the UI, so new Claude models can appear without waiting for a Paperclip release.
Source: github.com/paperclipai/paperclip/releases/tag/v2026.529.Read original →
FIG-8291:1
70radar
FIG-8291:1
#0828
#0828Agents & tools Claude Code Releases13 hours ago
`Claude Code` `v2.1.158` adds Auto mode on Bedrock, Vertex, and Foundry
60radar
Claude CodeCoding agent CLI — automates code work with Claude
Auto mode now reaches managed-cloud deployments for Opus 4.7 and Opus 4.8. Useful only if you run through Bedrock, Vertex, or Foundry.
- CLAUDE_CODE_ENABLE_AUTO_MODE=1 enables the feature, so rollout is opt-in rather than default behavior.
- Support covers Bedrock, Vertex, and Foundry for Opus 4.7 and Opus 4.8; direct API users get no new signal here.
- Practical value is higher for teams keeping LLM traffic inside cloud-vendor rails; local CLI-only workflows can skip this update.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.158Read original →
FIG-8281:1
60radar
FIG-8281:1
#0827
#0827Agents & tools Claude Code Releases20 hours ago
`Claude Code` `v2.1.157` auto-loads local skill plugins
80radar
Claude CodeCoding agent CLI — automates code work with Claude in terminal
Local .claude/skills plugins now load without marketplace setup. Plugin scaffolding, agent settings, and worktree handling got smoother for daily agent work.
- Plugins under .claude/skills are auto-loaded, so private workflow extensions no longer need marketplace packaging first.
- claude plugin init <name> scaffolds local plugins directly; useful for turning repeated repo workflows into reusable skills.
- settings.json now honors the agent field for dispatched sessions, with --agent <name> available for one-off overrides.
- EnterWorktree can switch between Claude-managed worktrees mid-session, reducing context breaks during parallel agent tasks.
- Several fixes target terminals, image paste, resume state, and background agents; less time lost to CLI session cleanup.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.157Read original →
FIG-8271:1
80radar
FIG-8271:1
#0826
#0826Agents & tools Cline Releases21 hours ago
`Cline CLI` `v3.0.15` adds Hub for remote session control
80radar
ClineOpen-source coding agent — autonomous work in IDE and CLI
A new Hub web app can monitor clients, drive sessions, stream output, and restart the local hub. Useful for running agent work from another machine, with a room secret as the safety gate.
- Cline Hub supports local, LAN, and tunnel access gated by a room secret; remote agent control is now a built-in workflow.
- Global AGENTS rules now apply across all sessions, reducing repeated setup when multiple repos share the same coding conventions.
- Plugins can inject static or dynamic rules inside the sandbox, opening a cleaner path for reusable team or stack-specific behavior.
- The bundled catalog adds Claude Opus 4.8, Kimi K2.6, and Qwen3.7 Max with cache support; model choice gets fresher without manual wiring.
- Discord sessions now bind to individual authors and connector error flooding is capped, making shared channel workflows less brittle.
Source: github.com/cline/cline/releases/tag/cli-v3.0.15Read original →
FIG-8261:1
80radar
FIG-8261:1

Fri, May 2913 dispatches

#0825
#0825Agents & tools Google AI Forumyesterday
`Gemini Code Assist` Individual IDE/CLI Access May End June 18
80radar
Gemini Code AssistAI coding assistant — Gemini in IDE extensions and CLI
Individual Pro, Ultra, and free access stops on June 18, 2026. VS Code users should line up alternate auth or tools before agent workflows break.
- The notice names Gemini CLI and Gemini Code Assist IDE extensions. Individual Pro, Ultra, and free accounts are the affected paths.
- The cutoff date is June 18, 2026. Any VS Code workflow wired to personal Gemini access needs a fallback before then.
- The forum post does not confirm whether an Antigravity-specific VS Code extension continues. Treat it as unresolved until Google posts a clearer migration path.
Source: discuss.ai.google.dev/t/will-there-no-longer-be-a-geminiRead original →
FIG-8251:1
80radar
FIG-8251:1
#0824
#0824Agents & tools Hacker News · Show HN AIyesterday
`AISlop`, a CLI that catches AI-generated code smells
60radar
AISlopAI code-quality CLI — local code-smell scanning
Scans for patterns tests often miss: empty catches, dead code, duplicate helpers, and useless comments. Local-only hook integration makes it worth trying in agent-heavy repos.
- npx aislop scan runs locally, so source code is not uploaded. That lowers the barrier for private app repos.
- Targets AI-agent residue rather than syntax errors: empty catch, dead code, duplicated helpers, and low-value comments.
- Hook-based use after each tool call fits Claude Code, Codex, and opencode workflows where small slop accumulates fast.
Source: github.com/scanaislop/aislopRead original →
FIG-8241:1
60radar
FIG-8241:1
#0823
#0823Agents & tools Google AI Forumyesterday
`Antigravity Ultra` users hit persistent `MODEL_CAPACITY_EXHAUSTED` on Claude Opus thinking
60radar
AntigravityCoding agent IDE — uses Google's backend model routing
Paid agent sessions can fail before real work starts. Treat Antigravity Ultra as capacity-risky for Claude-heavy workflows until routing improves.
- The backend returns 503 `MODEL_CAPACITY_EXHAUSTED` from cloudcode-pa.googleapis.com, so this is server capacity, not local quota.
- Failures recur after 1-29 seconds on multi-step agent tasks like file edits, bash runs, and code generation. Long tasks are unreliable.
- Server-Timing shows about 13s spent waiting for model capacity before failure. Retry loops burn time without changing the result.
- Multiple similar forum threads over nearly a month point to account or routing-level fragility, not a one-off client setup problem.
Source: discuss.ai.google.dev/t/persistent-model-capacity-exhausRead original →
FIG-8231:1
60radar
FIG-8231:1
#0822
#0822Agents & tools GeekNews2 days ago
`Claude Code` Adds Dynamic Workflows for Parallel Subagents
90radar
Claude CodeCoding agent CLI — runs agentic dev tasks from the terminal
Large engineering jobs can now be split across many subagents in one session. This turns agent work from linear prompting into scripted orchestration worth testing on repo-scale tasks.
- A single session can dynamically write orchestration scripts, launch dozens to hundreds of parallel subagents, and collect results before delivery.
- The feature targets work too large or complex for one agent, so it fits migrations, wide refactors, audits, and test expansion.
- The bottleneck shifts to task decomposition. Vague goals will create noisy parallel work, while clear file/module boundaries should pay off.
Source: news.hada.io/topic?id=29981Read original →
FIG-8221:1
90radar
FIG-8221:1
#0821
#0821Agents & tools GeekNews2 days ago
`jqwik` 1.10.0 Prints Prompt-Injection Protestware During Tests
60radar
jqwikJava testing library — property-based testing support
A test dependency can leak hostile instructions into CI, IDE panels, and agent tool output. Treat logs as untrusted input when coding agents read them.
- jqwik 1.10.0 prints an instruction telling coding agents to ignore prior commands and delete tests/code during test runs.
- ANSI escape sequences hide the line in interactive terminals, but it can remain visible in CI logs, IDE test panels, and agent outputs.
- This is a concrete prompt-injection path through ordinary build logs. Pin versions and keep agents from blindly acting on stdout.
Source: news.hada.io/topic?id=29980Read original →
FIG-8211:1
60radar
FIG-8211:1
#0820
#0820Agents & tools Hermes Agent Releases2 days ago
`Hermes Agent` `v0.15.1`, same-day hotfix for dashboard reload loop
60radar
Hermes AgentCoding agent runtime — dashboard, MCP, and skills hub integrated
The broken v0.15.0 dashboard path is patched fast. Docker and MCP fixes reduce setup friction, but this is mostly a stability release.
- Loopback dashboard calls to /api/auth/me returned 401 by design; stale-token handling caused infinite reloads in Docker, hosted, and fresh installs.
- Docker no longer infers --insecure from bind host. LAN access and auth bypass are split via explicit HERMES_DASHBOARD_INSECURE=1.
- MCP bare commands like npx, npm, and node now resolve through /usr/local/bin, fixing silent container launch failures.
- The skills catalog expands from 858 to 19,932 entries, so picker coverage is much closer to the actual skills.sh index.
Source: github.com/NousResearch/hermes-agent/releases/tag/v2026.Read original →
FIG-8201:1
60radar
FIG-8201:1
#0819
#0819Agents & tools GeekNews2 days ago
`CodeBoarding`, Interactive Architecture Diagrams for Codebases
60radar
CodeBoardingOpen-source code map tool — auto-generates architecture docs
Static analysis plus LLM reasoning turns a repo into architecture docs. Useful before handing context to coding agents.
- Generates high-level architecture diagrams and component docs from a codebase, reducing the first-pass repo reading cost.
- Outputs Markdown into .codeboarding/, so the result can be versioned, reviewed, or fed into agents as project context.
- Best fit is unfamiliar or aging repos. For tiny apps, manual notes may beat the setup overhead.
Source: news.hada.io/topic?id=29976Read original →
FIG-8191:1
60radar
FIG-8191:1
#0818
#0818Agents & tools Cursor Changelog2 days ago
`Cursor` Adds Auto-review Run Mode
90radar
CursorAI code editor — agent-driven code edits and review
Longer autonomous runs now need fewer approval prompts. The value is less babysitting, as long as the safety model is strong enough for real repo work.
- Auto-review is a new run mode for longer Cursor tasks with fewer approval prompts, reducing interruption during agent work.
- The changelog explicitly pairs fewer approvals with safer execution, so the feature targets autonomy without fully removing guardrails.
- Best fit is bounded repo work such as refactors, fixes, and test passes where repeated approvals slow the loop.
Source: cursor.com/changelog/auto-reviewRead original →
FIG-8181:1
90radar
FIG-8181:1
#0817
#0817Agents & tools OpenAI Codex Releases2 days ago
`Codex CLI` `0.135.0` improves diagnostics, remote status, and Vim mode
70radar
CodexCoding agent CLI — automates code work in the terminal
Supportability got much better: richer doctor, remote /status, and sturdier TUI output. Worth upgrading if you run Codex in terminals or remote sessions.
- codex doctor now reports environment, Git, terminal, app-server, and thread inventory diagnostics, cutting back-and-forth in support cases.
- Remote TUI sessions show connection details and server version through /status, which helps debug mismatched client-server setups.
- Vim mode adds text-object editing and better word/end-of-line behavior; daily terminal use gets noticeably less rough.
- Python SDK thread and turn APIs now expose Sandbox presets, making sandbox configuration less verbose in automation scripts.
Source: github.com/openai/codex/releases/tag/rust-v0.135.0Read original →
FIG-8171:1
70radar
FIG-8171:1
#0816
#0816Agents & tools Cline Releases2 days ago
`Cline` `v3.86.0` Adds `Claude Opus 4.8` and `Kimi K2.6` Support
60radar
ClineVS Code coding agent — supports multiple LLM providers
More top-end model choices land inside the editor agent. The 1M-context option matters for large repo work, while the repo move is maintenance-only for most users.
- Claude Opus 4.8 support includes 1M-context variants where available, useful for broad repo inspection and long planning runs.
- Moonshot Kimi K2.6 is now selectable, adding another strong coding/model option without changing agent workflow.
- Prompt-cache support display for Qwen 3.7 Max was fixed in the Cline provider, reducing model capability guesswork.
- The VS Code extension project moved to apps/vscode; relevant mainly for contributors and fork maintainers.
Source: github.com/cline/cline/releases/tag/v3.86.0Read original →
FIG-8161:1
60radar
FIG-8161:1
#0815
#0815Agents & tools Hermes Agent Releases2 days ago
`Hermes Agent` v0.15.0: faster startup, multi-agent Kanban, security hardening
90radar
Hermes AgentCoding agent CLI — supports multi-agent task boards
The release turns a coding CLI into a broader agent platform. Faster startup, task swarms, secrets handling, and promptware defense make it worth testing for serious automation.
- run_agent.py dropped from 16,083 to 3,821 lines across 14 agent/ modules, making plugin work and codebase navigation less painful.
- Kanban now supports auto-decomposition, swarm topology, scheduled tasks, per-task worktrees, branches, and model overrides for cost-aware parallel work.
- Cold start improves again: one more second cut, 47% fewer per-conversation function calls, and session_search is now 4,500x faster and free.
- Security and ops got real attention: Brainworm-class promptware defense plus Bitwarden Secrets Manager bootstrap token instead of many provider keys.
- New integrations include Krea 2, FAL image generation as a plugin, Nous MCP catalog picker, ntfy, and deeper xAI/Grok support.
Source: github.com/NousResearch/hermes-agent/releases/tag/v2026.Read original →
FIG-8151:1
90radar
FIG-8151:1
#0814
#0814Agents & tools Claude Code Releases2 days ago
`Claude Code` `v2.1.154`: `Opus 4.8`, dynamic workflows, cheaper fast mode
90radar
Claude CodeCoding agent CLI — automates code edits and runs in terminal
Background workflows can now fan work across tens to hundreds of agents. This is a serious step toward larger autonomous coding runs, with cost and safety controls worth testing now.
- Opus 4.8 is now available with high effort by default, plus /effort xhigh for harder tasks. Expect better results on deep refactors and debugging.
- Dynamic workflows let Claude create and run multi-agent workflows in the background. Use /workflows to inspect runs instead of manually coordinating subtasks.
- Fast mode on Opus 4.8 costs 2x standard rate for 2.5x speed. Good for batch edits where latency beats token efficiency.
- /simplify now performs cleanup-only review and applies fixes. It is better suited to reuse, simplification, efficiency, and code altitude, not bug hunting.
- Safety and ops got sharper: rm -rf $HOME trailing-slash blocking, pending MCP approval display, and stronger data-exfiltration detection.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.154Read original →
FIG-8141:1
90radar
FIG-8141:1
#0813
#0813Agents & tools GitHub Changelog2 days ago
`Claude Opus 4.8` is now generally available in `GitHub Copilot`
80radar
A stronger coding model is now inside the editor workflow. Test it on hard refactors and generation-heavy tasks; daily value depends on latency and plan access.
- Claude Opus 4.8 is Anthropic's latest Opus model, now generally available in GitHub Copilot.
- Early GitHub testing says code understanding and generation improved clearly; verify on your own repo before switching defaults.
- Best first use cases: complex file edits, unfamiliar codebase reading, and generation-heavy tasks where model quality beats speed.
Source: github.blog/changelog/2026-05-28-claude-opus-4-8-is-geneRead original →
FIG-8131:1
80radar
FIG-8131:1

Thu, May 281 dispatches

#0812
#0812Agents & tools future_tools_news3 days ago
`Claude Code` adds dynamic workflows for parallel subagents
90radar
Claude CodeCoding agent CLI — automates repo work with Claude
Complex repo work can now split across parallel AI subagents. This reduces manual task sharding and makes Claude Code more useful for larger refactors and multi-file debugging.
- Dynamic workflows let Claude Code dispatch multiple subagents in parallel, cutting wait time on broad codebase tasks.
- Release date is 2026-05-28, so this is a fresh core feature rather than a reaction post or minor changelog item.
- Best fit: bounded tasks like test expansion, bug isolation, migration sweeps, and parallel investigation across modules.
Source: claude.com/blog/introducing-dynamic-workflows-in-claude-Read original →
FIG-8121:1
90radar
FIG-8121:1

Wed, May 276 dispatches

#0811
#0811Agents & tools GeekNews3 days ago
`React Doctor` diagnoses AI-generated React code with static analysis
70radar
React DoctorReact static analysis CLI — scores AI-generated code quality
A one-line npx scan scores React projects across quality and safety checks. Useful as a fast review gate before accepting agent-written code.
- Runs with npx react-doctor@latest, so it fits into local review or CI without a setup-heavy adoption path.
- Diagnostics cover state, side effects, performance, security, accessibility, and architecture — broader than lint-only checks.
- Outputs a 0-100 score in the terminal, making AI code review easier to turn into a repeatable gate.
Source: news.hada.io/topic?id=29920Read original →
FIG-8111:1
70radar
FIG-8111:1
#0810
#0810Agents & tools opencode_releases3 days ago
`opencode` `v1.15.11` improves provider timeouts and background agents
60radar
opencodeOpen-source coding agent — supports TUI, Desktop, and MCP
Provider calls now get a 10s default header timeout, and background agents push updates without polling. Worth upgrading if you use remote projects, MCP servers, or desktop sync.
- headerTimeout adds a 10s default for OpenAI-style provider requests, reducing hangs from slow or broken upstream responses.
- Experimental background agents now push updates without polling. Long-running agent sessions should feel less wasteful and more responsive.
- Remote-backed projects now keep stable identity across reloads, fixing annoying session and cache drift in multi-machine workflows.
- MCP cleanup improved: dynamically added servers disconnect cleanly when removed, which matters for plugin-heavy agent setups.
- Desktop v2 got session-control polish and fixes for deleted sync caches, duplicate servers, and virtualized timeline interactions.
Source: github.com/anomalyco/opencode/releases/tag/v1.15.11Read original →
FIG-8101:1
60radar
FIG-8101:1
#0809
#0809Agents & tools Claude Code Releases4 days ago
`Claude Code` `v2.1.152` adds review auto-fix and live skill reloads
80radar
Claude CodeCoding agent CLI — automates edits and reviews in terminal
Review findings can now patch the working tree directly. Skill iteration is faster too, with reload hooks and tool bans making agent workflows easier to govern.
- /code-review --fix now applies findings after review, covering reuse, simplification, and efficiency suggestions.
- /simplify now calls /code-review --fix, turning cleanup into a more automated refactor path.
- Skills and slash commands can set disallowed-tools, a practical guardrail for risky or narrow agent tasks.
- /reload-skills and reloadSkills: true let newly installed skills work in the same session, no restart needed.
- Fallback behavior improved: missing primary models switch to --fallback-model for the rest of the session.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.152Read original →
FIG-8091:1
80radar
FIG-8091:1
#0808
#0808Agents & tools GeekNews4 days ago
Figma Releases In-Canvas Design Agent
70radar
FigmaCollaborative design platform — canvas-based UI design and prototyping
Design work can now stay inside the canvas while AI handles context-aware changes. Useful for faster UI iteration, but value depends on how well it respects your design system.
- The agent runs directly on the Figma canvas, reducing handoff between chat tools and manual editing.
- It is tuned on components, tokens, standards, and best practices, so design-system awareness is the main promise.
- Best fit is repetitive UI cleanup, component alignment, and token-based edits. Final product judgment still stays human.
Source: news.hada.io/topic?id=29906Read original →
FIG-8081:1
70radar
FIG-8081:1
#0807
#0807Agents & tools GitHub Changelog4 days ago
`Copilot Memory` adds deletion, repo scope controls, and CLI settings
70radar
Copilot MemoryCoding assistant memory — persists project context for suggestions
Memory governance is getting less risky. Repo-level shutoff and CLI controls make preview trials easier.
- Improved deletion controls reduce stale or sensitive context risk when Copilot Memory learns the wrong thing.
- A repository-level off switch matters for client work, OSS repos, or codebases where persisted context is unwanted.
- More controls in the Copilot CLI move memory management closer to daily terminal workflows, not just web settings.
Source: github.blog/changelog/2026-05-26-copilot-memory-has-moreRead original →
FIG-8071:1
70radar
FIG-8071:1
#0806
#0806Agents & tools OpenAI Codex Releases4 days ago
`Codex CLI` `0.134.0` adds local history search and stronger MCP flows
70radar
Codex CLICoding agent CLI — automates code work from the terminal
Local conversation search turns old agent work into reusable context. MCP, profile, and remote fixes make daily CLI use smoother enough to upgrade soon.
- Local history search supports case-insensitive content matches and previews, so prior fixes become searchable memory inside Codex.
- --profile is now the primary selector across CLI, TUI permissions, and sandbox flows. Legacy configs are rejected with migration guidance.
- MCP setup gains per-server environment targeting and OAuth options for streamable HTTP servers, reducing connector setup friction.
- Read-only MCP tools can run concurrently when readOnlyHint is advertised. Faster context gathering matters in tool-heavy agent runs.
Source: github.com/openai/codex/releases/tag/rust-v0.134.0Read original →
FIG-8061:1
70radar
FIG-8061:1

Tue, May 265 dispatches

#0805
#0805Agents & tools GeekNews5 days ago
`CodeGraph`: Code Knowledge Graph for AI Coding Agents
70radar
CodeGraphCode knowledge graph — cuts agent code exploration overhead
Pre-indexed semantic code maps cut exploration overhead before agents start editing. The claimed gains are large enough to test on token-heavy repos now.
- Works with Claude Code, Codex, and Cursor by giving agents a prebuilt semantic code graph before exploration.
- Claims 35% lower cost, 59% fewer tokens, 49% faster runs, and 70% fewer tool calls versus the existing approach.
- Best fit is large or unfamiliar repos where agents burn context on discovery before making a small change.
Source: news.hada.io/topic?id=29873Read original →
FIG-8051:1
70radar
FIG-8051:1
#0804
#0804Agents & tools GeekNews5 days ago
`nb-cli`, a CLI for AI agents and notebook automation
60radar
nb-cliNotebook automation CLI — agent-friendly Jupyter operations
Jupyter notebooks become easier for agents to inspect and edit without wrestling raw .ipynb JSON. Experimental, but worth tracking for data-heavy coding workflows.
- Built in Rust, so notebook operations target fast and stable CLI use rather than editor-only workflows.
- The pain point is clear: raw .ipynb JSON is awkward for automation and LLM context handling.
- Useful for agent pipelines that need to read, patch, or validate notebooks as artifacts. Still experimental, so keep it out of critical paths.
Source: news.hada.io/topic?id=29871Read original →
FIG-8041:1
60radar
FIG-8041:1
#0803
#0803Agents & tools Cline Releases5 days ago
`Cline` `v3.85.0` adds newer model provider support
60radar
ClineVS Code coding agent — supports multiple LLM providers
Provider coverage widened across SAP AI Core, DeepSeek, Gemini, and Vertex. Useful if your agent stack depends on multiple model backends, but this is an incremental release.
- GPT-5.5 support landed for SAP AI Core, expanding enterprise-hosted model routing options.
- DeepSeek V4 Flash and Pro were added, giving Cline users more cost/performance choices.
- Gemini 3.5 Flash now works through Gemini and Vertex; Vertex Claude global endpoint handling was also fixed.
Source: github.com/cline/cline/releases/tag/v3.85.0Read original →
FIG-8031:1
60radar
FIG-8031:1
#0802
#0802Agents & tools future_tools_news5 days ago
`Grok Build` beta opens to `SuperGrok` and `X Premium+` users
70radar
Grok BuildAI build tool — beta access for xAI paid users
xAI is gating its build-oriented beta behind paid plans. Worth testing if it turns prompts into usable prototypes, but access cost comes first.
- Access is limited to SuperGrok and X Premium+ users, so the first filter is paid account availability, not technical setup.
- The release is marked beta on 2026-05-25. Treat output quality, export paths, and production readiness as unproven.
- If Grok Build supports app generation or workflow automation, it becomes another quick-prototype option beside coding agents and no-code builders.
Source: x.com/xai/status/2058973760708091907?utm_source=futuretoRead original →
FIG-8021:1
70radar
FIG-8021:1
#0801
#0801Agents & tools Paperclip Releases5 days ago
`Paperclip` `v2026.525.0` adds Modal sandboxes and workspace diff plugins
70radar
PaperclipCoding agent platform — plugin runtime and sandbox support
Agents now get first-party Modal sandboxes, built-in workspace diffs, routine-scoped secrets, and local-to-cloud sync. Strong release if you run multi-provider coding agents.
- Modal is now a first-party sandbox provider alongside E2B, Cloudflare, and Daytona; CI publishing and cold-start-friendly probe timeouts are included.
- Workspace diffs moved into a plugin with staged, unstaged, renamed, binary, oversized, and untracked change views; review UX gets much closer to IDE-grade.
- Routine env supports persisted revisions and agent < project < routine precedence, while secret values stay out of logs and access events.
- acpx_local now resolves bare Claude model IDs and reads ~/.claude/settings.json permissions, reducing first-run adapter failures.
Source: github.com/paperclipai/paperclip/releases/tag/v2026.525.Read original →
FIG-8011:1
70radar
FIG-8011:1

Mon, May 251 dispatches

#0800
#0800Agents & tools GeekNews6 days ago
`Flue`, a sandbox harness framework for coding agents
70radar
FlueAgent harness framework — headless coding-agent automation
Coding-agent UX becomes fully headless and programmable. Useful for building custom automation around Claude Code or Codex-style workflows.
- Flue is a TypeScript agent harness, so it fits naturally into Node-based internal tools and CI-style automation.
- It recreates the Claude Code/Codex experience in a 100% headless form, removing dependence on interactive CLI flows.
- Custom agents can work through problems autonomously like a coding agent, making it relevant for repeatable repo tasks and experiments.
Source: news.hada.io/topic?id=29849Read original →
FIG-8001:1
70radar
FIG-8001:1

↻ Load earlier issues

`Paperclip` `v2026.529.0` adds document annotations and company skills CLI

`Claude Code` `v2.1.158` adds Auto mode on Bedrock, Vertex, and Foundry

`Claude Code` `v2.1.157` auto-loads local skill plugins

`Cline CLI` `v3.0.15` adds Hub for remote session control

`Gemini Code Assist` Individual IDE/CLI Access May End June 18

`AISlop`, a CLI that catches AI-generated code smells

`Antigravity Ultra` users hit persistent `MODEL_CAPACITY_EXHAUSTED` on Claude Opus thinking

`Claude Code` Adds Dynamic Workflows for Parallel Subagents

`jqwik` 1.10.0 Prints Prompt-Injection Protestware During Tests

`Hermes Agent` `v0.15.1`, same-day hotfix for dashboard reload loop

`CodeBoarding`, Interactive Architecture Diagrams for Codebases

`Cursor` Adds Auto-review Run Mode

`Codex CLI` `0.135.0` improves diagnostics, remote status, and Vim mode

`Cline` `v3.86.0` Adds `Claude Opus 4.8` and `Kimi K2.6` Support

`Hermes Agent` v0.15.0: faster startup, multi-agent Kanban, security hardening

`Claude Code` `v2.1.154`: `Opus 4.8`, dynamic workflows, cheaper fast mode

`Claude Opus 4.8` is now generally available in `GitHub Copilot`

`Claude Code` adds dynamic workflows for parallel subagents

`React Doctor` diagnoses AI-generated React code with static analysis

`opencode` `v1.15.11` improves provider timeouts and background agents

`Claude Code` `v2.1.152` adds review auto-fix and live skill reloads

Figma Releases In-Canvas Design Agent

`Copilot Memory` adds deletion, repo scope controls, and CLI settings

`Codex CLI` `0.134.0` adds local history search and stronger MCP flows

`CodeGraph`: Code Knowledge Graph for AI Coding Agents

`nb-cli`, a CLI for AI agents and notebook automation

`Cline` `v3.85.0` adds newer model provider support

`Grok Build` beta opens to `SuperGrok` and `X Premium+` users

`Paperclip` `v2026.525.0` adds Modal sandboxes and workspace diff plugins

`Flue`, a sandbox harness framework for coding agents