Goose

C▲ 0 vs last quarter

Open-source (Apache 2.0) AI agent under the Agentic AI Foundation (AAIF) at the Linux Foundation.

Rate this tool Add to compare

Dimension breakdown

Score · confidence

UX / DX

79% conf55

Capability

79% conf62

Reliability

79% conf93

Value

50% conf60

Community

79% conf31

Enterprise / Compliance

50% conf35

Autonomy

50% conf50

Integration

50% conf60

Sources blend review platforms, community sentiment, the Signal Radar and practitioner ratings. Scores re-blend each quarter.

Open-source (Apache 2.0) AI agent under the Agentic AI Foundation (AAIF) at the Linux Foundation. Repo moved to github.com/aaif-goose/goose (AAIF/Linux Foundation donation, April 7 2026). ~49.8K GitHub stars, 474+ contributors, 136 releases — sustained velocity (v1.34 -> v1.36 in 2 weeks).

Runs locally with 25+ LLM providers. v1.36.0 (May 27, 2026) added a '/goal' self-evaluation step before task completion, a local 'goose review' code-analysis command, Hooks with pre/post tool-execution denial, deep linking (goose://resume), and a TUI command interface + diff viewer; v1.35.0 (May 22) shipped the Hooks system, ACP slash commands, unified thinking-effort control, and proactive OAuth token refresh. YAML Recipes for composable workflow automation, headless mode for CI/CD, and 'goose serve' background service distinguish it from competitors.

Strongest open-source data sovereignty story (fully air-gapped capable via Ollama). Block-proven at 12,000 engineers, 40% more code shipped; Block Q1 2026 record results (gross profit $2.91B, +27% YoY) reinforce commercial backing.

AAIF governance materially active: governing-board chair (David Nalley, AWS), 146+ member orgs. No enterprise compliance for the agent itself (no SOC 2, SSO, RBAC, audit logs) — a third-party-review claim of a SOC 2 Type II hosted instance is uncited and unconfirmed by official sources.

No independent SWE-bench validation. No native codebase semantic indexing.

Recommended

Use cases

Not yet assessed — this section fills in as ACES research covers the tool.

Score caps

Risk flags

No codebase indexing
capabilityConditional
No codebase indexing or semantic search capability
Caps Capability at 55
Removed when — Codebase indexing or RAG capability launched
Unvalidated benchmarks
trustConditional
Unvalidated benchmark claims
Caps Autonomy at 70
Removed when — Independent benchmark validation (SWE-bench, Aider leaderboard, etc.) published

Assessment

Status rationale

Tracked because active development under AAIF governance (~49.8K stars, 474+ contributors, 136 releases — v1.34 to v1.36 in 2 weeks; repo moved to github.com/aaif-goose/goose April 7 2026), strong Block commercial backing (Q1 2026 record results, 'intelligence company' strategy), and strongest open-source data sovereignty story. AAIF governance materially active (governing-board chair, 146+ members) — not just a press-release shell.

Assessment is research-based, not hands-on, which caps confidence at the Tracked band. Blocked from advancement by: (1) zero enterprise compliance infrastructure for the agent (no SOC 2, SSO, RBAC, audit logs); (2) no independent benchmark validation; (3) no native codebase semantic indexing.

Watch for

Movement triggers

Upgrade if: (a) Enterprise features ship for the agent — SSO/SAML, audit logs, SOC 2 (would lift compliance out of Band 1); (b) Independent SWE-bench results published (removes unvalidated-benchmarks cap, autonomy could move to 15-16); (c) Native codebase indexing/embeddings shipped (removes no-codebase-indexing cap, context could move to 13-14); (d) Sustained release velocity + viability evidence justifies viability 13->14; (e) Hands-on validation (handsOn -> demo/tested) would raise evidence grade and could support an Assessed signal level. Downgrade if: AAIF/Block reduces investment, major unpatched security vulnerability, or reliability issues (latency, stuck agents, loops) escalate to warrant a reliability-complaints cap.

Caution