Skip to content

Releases: agents-squads/squads-cli

v0.3.1 — first stable v0.3.x

25 Apr 04:07
d79b06c

Choose a tag to compare

First stable v0.3.x release on npm @latest. Same code as 0.3.0-rc.1 (which burned in on @next).

0.3.0 was skipped because that version slot is reserved by a deprecated historical pre-release. npm enforces version immutability.

Highlights since 0.2.2

Added

  • Conversation protocol — agents talk to each other and use tools mid-conversation. squads run <squad> drives a lead → scan → work → review → verify cycle.
  • Org cyclesquads run with no target runs all squads in waves, with smart-skip for converged work.
  • New commandsreview, credentials, goals, log, plus refinements to init, status, run.
  • Project config system.squads/config.yml for per-project settings. Resolution: env var > config file > defaults.
  • PreToolUse guardrail hooks — agent sessions can be gated by user-defined safety hooks.
  • Demo agent scaffoldsquads init includes starter agents and "what's next" guidance.
  • Growth squad template — added to seed templates.
  • Tier 2 documentation — guides for local-services mode (Postgres, Redis, API, Bridge).

Changed

  • Run engine rewrite — decomposed into smaller modules. Foundation for cloud execution.
  • Role-based timeouts — workers, reviewers, leads have appropriate per-role timeouts (replaces hardcoded 8-min ceiling).
  • Anti-collision rules — multiple squads no longer race to create duplicate release PRs or issues.
  • Prompts extracted — lead briefings, planning instructions, orchestrator prompts moved from TypeScript into templates/prompts/*.md.
  • OIDC trusted publishingrelease.yml and publish.yml publish via GitHub OIDC instead of NPM_TOKEN.

Fixed

  • Telemetry write-only key — restored after being incorrectly removed in March (telemetry had been silent since 2026-03-14).
  • First-run UX — prerequisites check, helpful empty-state for squads list, schedule hint after first run.

Install

npm install -g squads-cli
squads --version    # 0.3.1

Resources

v0.3.0-rc.1 — pre-release burn-in

25 Apr 04:07

Choose a tag to compare

Pre-release

Pre-release candidate published to npm @next for burn-in testing. Same code as v0.3.1.

npm install -g squads-cli@next

Promoted to @latest as v0.3.1 on 2026-04-24. See v0.3.1 release notes for full changes.

v0.2.2 — IDP, observability, tiered architecture, org cycle

28 Mar 16:21
b4c2192

Choose a tag to compare

What's new

Major release adding the IDP (Internal Developer Platform) layer, observability commands, tiered architecture, and org cycle improvements.

IDP (Internal Developer Platform)

  • squads catalog — service catalog: browse, inspect, and validate services
  • squads release — release pre-checks against IDP scorecard
  • Structured agent frontmatter with spec.type, spec.squad, spec.role

Observability

  • squads obs history — execution history with tokens and cost
  • squads obs cost — spend summary by squad and model
  • squads obs sync — sync JSONL execution log to Postgres
  • JSONL-based observability log at .agents/observability/executions.jsonl

Tiered Architecture

  • squads tier — show current architecture tier (Tier 1/2/3)
  • squads services — manage local services (Postgres, Bridge)

Org Cycle

  • Smart skip: re-run squads only when goals/priorities changed
  • Wait on quota + skip completed squads on resume
  • --force flag to re-run all squads

Cognition

  • squads cognition — business cognition engine commands

Quality

  • 1779 tests passing across 87 test files
  • 13/13 Docker fresh-user tests pass

Install / upgrade

npm install -g squads-cli

Full Changelog: v0.2.1...v0.2.2

v0.2.1 — First-run experience reset

13 Mar 19:25
4b59a8c

Choose a tag to compare

What's new

Complete reset of the first-run experience. A new user can now:

npm install -g squads-cli
mkdir my-project && cd my-project && git init
squads init --yes
squads run research/lead

...and get useful AI agent output in under 5 minutes.

Changes

  • 4 core squads for every user: research, company, intelligence, product
  • Default business brief: tracks big AI players (Anthropic, OpenAI, Google, Amazon, Meta, xAI)
  • Auto-commits scaffolding so agents can use git worktrees immediately
  • gh CLI is optional — warning instead of hard error
  • squads run is the unified entry point (replaces squads auto start)
  • Context cascade: SYSTEM.md → SQUAD.md → priorities → directives → state
  • Phase-ordered execution with depends_on in SQUAD.md frontmatter
  • Role-based context injection (scanner/worker/lead/coo profiles)
  • 1732 tests passing across 87 test files
  • Platform-agnostic install hints (no more brew install gh on Linux)

Install / upgrade

npm install -g squads-cli

Full Changelog: v0.6.2...v0.2.1

v0.6.2

21 Feb 07:08

Choose a tag to compare

What's Changed

  • fix: extract hardcoded values to constants, read cooldown from frontmatter by @kokevidaurre in #347

Full Changelog: v0.6.1...v0.6.2

v0.6.1

21 Feb 05:03
7f08536

Choose a tag to compare

What's Changed

Full Changelog: https://github.com/agents-squads/squads-cli/commits/v0.6.1

v0.6.0

20 Feb 23:46

Choose a tag to compare

squads-cli v0.6.0

Security

  • fix(security): escape HTML in OAuth callback responses (#323)
  • fix(security): prevent shell injection in agent spawn (#324)
  • fix(security): restrict auth file permissions to owner-only (#325)

Added

  • Git worktree isolation for concurrent agent execution
  • Multi-provider model routing (Claude, Gemini, GPT, Ollama)
  • Memory injection from daily briefing + cross-squad learnings
  • Event emission on agent completion
  • Ralph verification loop (verify → re-run)
  • squads eval readiness scorer
  • squads deploy command
  • Dashboard ROI metrics, cost projections, baseline command
  • Squad-scoped skills and MCP servers
  • Interactive init, session filtering
  • Light mode terminal palette
  • Privacy indicators in CLI output

Changed

  • Removed pg, ioredis, supabase dependencies (lighter install)
  • Deprecated -e flag, added native Claude execution pattern

Fixed

  • Worktree isolation replaces git checkout -b (no branch collisions)
  • Unset CLAUDECODE env var to allow nested Claude Code execution
  • Memory format updated to ## YYYY-MM-DD: for synthesizer compatibility
  • 15+ additional bug fixes (see CHANGELOG.md)

Tests

  • 14 new test suites covering core commands, database ops, dashboard renderers, LLM CLIs, and more

Installation

npm install -g squads-cli@0.6.0

Full Changelog

https://github.com/agents-squads/squads-cli/blob/main/CHANGELOG.md

v0.5.1 - Model Routing & Security Hardening

28 Jan 01:49

Choose a tag to compare

Highlights

Features

  • Model routing: --model flag for task-difficulty routing (#182)
  • Sprint goals: Display sprint goals in status view
  • Time-windowed stats: Show execution statistics with time windows in trigger command
  • Lead orchestration: Add lead-orchestrated squad execution
  • Security: Pre-commit hook to block PII leakage

Performance

  • 385% faster help command: Lazy-load ioredis to eliminate startup regression (#190)

Fixes

  • Cross-platform path handling in cron/autonomous commands (#207)
  • Scheduler offline error handling in trigger list (#196)
  • Support squad/agent slash syntax in run command
  • Use branch/PR workflow instead of direct main push (#180)
  • Replace tmux with nohup for background execution (#181)
  • Use absolute paths in Claude Code hooks (#189)

Security

  • Remove hardcoded database credentials from db.ts (#195)

UX Improvements

  • Clarify dashboard budget display for different plans (#198)
  • Show help after missing required arguments (#197)
  • Add DEBUG logging to silent error handlers (#201)

Documentation & Testing

  • JSDoc comments for squad-parser.ts, providers.ts, telemetry.ts
  • Unit tests for telemetry, git, memory, auth, lock modules
  • Coverage reporting with @vitest/coverage-v8

🤖 Generated with Agents Squads

v0.5.0

22 Jan 21:42
9713480

Choose a tag to compare

Highlights

New Features

  • Redis distributed locking for safe concurrent memory access
  • Docker production configuration for deployment
  • Guided interactive onboarding with template system
  • Zombie detection and cleanup for tmux sessions
  • Local cooldown check for scheduled runs
  • Bridge API injection with approval instructions

Bug Fixes

  • Handle overnight sessions with proper time comparison
  • Correct command syntax for launchd plists
  • Use TCP socket for non-HTTP service health checks
  • Add --print flag to prevent zombie tmux sessions
  • Resolve lint errors blocking CI
  • Fix TypeScript strict mode errors

Other

  • Migrate ux-tests from hq
  • Mark mem0/engram-mcp as optional
  • Add squads setup templates to AGENTS.md and CLAUDE.md

Full Changelog

v0.4.13...v0.5.0

v0.4.13

19 Jan 20:47

Choose a tag to compare

Multi-LLM Provider Support

Choose your LLM provider during init:

squads init --provider gemini   # Google Gemini
squads init --provider openai   # OpenAI
squads init --provider ollama   # Local models
squads init --provider none     # Planning only

Features

  • Multi-LLM provider selection during squads init
  • Declarative dashboard engine
  • Slack notifications in tonight loop
  • Git hooks for Slack channel sync

Fixes

  • Type definitions for Squad/Agent interfaces
  • Weighted token capacity tracking
  • Non-TTY detection with --yes flag