AXLR vs MiniMax Code

AXLR 83

Harness reading of the framework line, scored on what a program driving it has to supply. Six local tools plus the host's bookkeeping tools; work modes and an execution policy hide the tools a run may not use, axlr_catalog searches the catalog instead of loading it, forged tools keep the request prefix for the prompt cache, and a context cap of 75,000 prompt tokens opens a successor session (20). local_read, local_search and local_list page with offset, next_offset and max_bytes; StreamEvents takes after, max_events (1 to 256) and wait_ms; a budget per execution in seconds, requests, tool calls and dollars; 32 tool calls a turn and max_session_usd (20). The --exec run ends in one result line with a status, the pending call, every call's outcome and the usage, and exits 0, 3 or 1; service errors carry codes an agent can act on; a budget stop says which limit (budget: requests 20 reached) and leaves the session resumable (16). Idempotency-Key on every turn, approval and direct call with cross-transport replay, expected_revision on decisions, expected_sha256 on worker writes, and an interrupted effectful call left uncertain for an operator rather than retried (19). One command runs a task, but the first launch needs an OpenRouter key or a local server, the service needs a CA, a principals file and remote KMP and MADE before a first session, and the only SDK is the Go library (8). 20+20+16+19+8 = 83.

MiniMax Code 75

Harness reading, for a pipeline driving the CLI. Lightweight mode sends no tool schemas, Plan mode is separate from permissions, and /btw side chats run without blocking the main task. No per-tool allow or deny list was found in the CLI docs (14). json and stream-json output, --timeout, --max-steps, --output-schema and --output-last-message (18). Exit codes per failure class, a status field and an interaction error when a run would need a person (17). --session, --continue, fork and rewind, and a model request is retried when the first response doesn't arrive within 300 seconds (16). One positional argument runs a task and BYOK needs no MiniMax login, but there's no SDK package (10). Total 75.

AXLR 55

Apache-2.0 with a NOTICE (30). No privacy policy or terms on underpassai.com or axlr.underpassai.com; the docs' 'Where data goes' and 'Diagnostics and privacy' sections say that model requests go to OpenRouter, the local server or the native API, that Jev sends text to TypeSafe only when on, that sessions live under $XDG_STATE_HOME/axlr, that payload capture is off by default and redacted when on, and that traces are deleted after 30 days; nothing covers the engine downloads' or the catalog registry's handling of a client's requests (15). No deprecation policy; pre-1.0 changes are dated inline in the docs and the product contract records the 5 October 2026 retirement of the first skill catalogue after the fact (6). No telemetry to disclose, and the docs say in two places what leaves the machine and how to stop payload capture, though no single page says there is no telemetry (17). 30+15+6+17 = 68.

MiniMax Code 63

First-party code is MIT, with the bundled sandbox runtime under Apache-2.0 and the desktop app closed. The repository says the published npm package isn't shown to be built from this source, and npm is at 0.6.12 while the source is at 0.6.5 (22 of 30). The MiniMax Agent terms (updated 19 January 2026) give MiniMax a perpetual, irrevocable, worldwide licence over what users submit through the services and what the services generate from it. The privacy policy of the same date states no retention period and says nothing on training, and the telemetry doc says server-side retention isn't defined. BYOK runs that don't touch MiniMax's services aren't covered by either (8 of 30). No deprecation policy. Removed functions are listed in the changelog without dates ahead (6 of 20). Telemetry is off by default, with each field, each destination and two environment opt-outs documented (20). Total 56.

AXLR 70

Read as a local package, as claude-code, goose, opencode and openhands were. Release archives for Linux, macOS and Windows on amd64 and arm64 with a .sha256 beside each, a GHCR image, an OCI Helm chart and source builds with Go 1.26, all stated in the docs; no Homebrew, npm or apt route (20). CI on main runs vet, race tests and a coverage gate above 80 per cent for both modules, with the sandbox tests required under bubblewrap, and the last ten runs on main to 1 November 2026 all passed; the release workflow tests on six native runners (25). 10 open issues, all opened by the founder on 19 and 21 October 2026 as trackers for service work (work modes, plugins, the sandbox setting, cost limits), none from an outside user and none a crash report, 0 open pull requests and about 100 merged; with 0 stars there is nobody else to report, so the handling of outside reports is untested (20). Tags are semver 0.x and the docs date each behaviour change inline ("until 14 October 2026…"), but the 21 releases have empty notes, there is no CHANGELOG.md, and the product contract records one retirement (the seven 1.0 skill definitions on 5 October 2026) without a notice period (5). Pre-1.0, which the README calls an evolving project (0). 20+25+20+5+0 = 70.

MiniMax Code 63

Local-package reading, for the CLI. An install script for macOS, Linux and WSL and one for Windows, and npm @minimax-ai/code with Node.js 22.19+, 24.2+, 25 or 26 stated. The installer doesn't support Alpine or musl (20). A public 'Source verification' workflow builds and tests on push and pull request, and shields.io reported its checks passing on main on 19 October. The repository's own verification record marks fresh cross-platform acceptance as not run, and the npm releases aren't shown to come from this source (20). 91 issues open against 229 closed and 11 pull requests open against 166 closed, read through shields.io. The repository also takes desktop-app issues, and which open issues are crashes or regressions couldn't be read, so the line scores the counts alone (15). A dated CLI changelog with Added, Improved and Fixed headings, but no breaking-change label, and its latest entry (0.6.7, 10 October) trails npm's 0.6.12 (8). 0.6.x and called a source preview (0). Total 63.

AXLR 62

Harness reading of the framework checklist, as for the other four harnesses. 30 for what leaves the machine by default, 20 for approvals and sandboxing, 15 for prompt-injection posture, 15 for audit and 20 for the security programme. No product telemetry in the console source (the outbound hosts are openrouter.ai, api.anthropic.com, api.openai.com, a configured local server, api.github.com and github.com for engine downloads, and registry.underpassai.com for the signed catalog index) or in the docs; payload capture is off by default since 9 October 2026, traces omit prompts and credentials, model keys are never given to commands or plugins, and Jev sends requests and answers to TypeSafe only when on, which the setup assistant does when TYPESAFE_API_KEY is set, with a notice on stderr (27). Approval cards, a destructive list that always asks, four work modes, a policy level per execution, a bubblewrap or container sandbox with an enforced egress allow list, and a service that approves nothing but reads. Against that, the setup assistant writes autonomous: true, exec_sandbox is off by default, forged tools run with the person's rights, and the README says the console and worker are not sandboxes (13). Forged tools are refused when a file changes since the forge, a policy forbids calls outside the model, and the docs say MCP tools keep their own side effects, but no prompt-injection guidance was found and MCP results reach the model unmarked (6). The service keeps an event journal, audit records, a receipt with the SHA-256 of every call's arguments and outcome, and correlation ids on every event; the console keeps session records, a redacted app log and /session (14). No SECURITY.md, no security.txt on either host, no advisories, no bug bounty; release checksums and a minisign-signed catalog are integrity, not a programme (2). 27+13+6+14+2 = 62.

MiniMax Code 57

Harness reading of the framework checklist. 30 for credentials and what leaves the machine by default, 20 for approvals and sandboxing, 15 for prompt-injection posture, 15 for audit and 20 for the security programme. Provider keys are read from an environment variable and never printed, and stored in config.yaml in the data directory. All three telemetry channels are off by default and the doc lists every field sent. The default smart policy classifies actions through a cloud classifier, and what it sends isn't documented (20). Ask, Auto and Full access in the TUI. Headless defaults to smart and fails on a pending permission request instead of approving it, while full and off skip checks. A bundled sandbox runtime with macOS tests, a read-only filesystem policy for Explore subagents when sandboxing is available, but the default sandbox coverage isn't documented (12). SECURITY.md says permissions and sandboxing don't replace review of untrusted plugins, MCP servers and shell commands. No guidance found for fetched web content (6). Sessions kept locally, /export with credential redaction, /transcript and stream-json events (11). SECURITY.md takes private reports by email and GitHub's private advisories with no response SLA, history and build scans with Gitleaks are recorded, and OSV lists no advisories. No security.txt (8). Total 57.