AXLR vs Kimi Code CLI

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.

Kimi Code CLI 45

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. OAuth tokens sit in 0600 files and shell keys aren't read unless named. Telemetry is on by default and, in the 2.2.0 package, sends events with a device ID, session ID and model name to telemetry-logs.kimi.com. Auto-update is on by default (14). Interactive sessions ask before edits and commands, and a dangerous-command guard covers rm -rf and similar, but kimi -p runs under Never Ask, where everything is approved and the guard is off, with only static deny rules left. No sandbox found, and hooks fail open (7). The MCP page warns about untrusted servers and project .kimi-code/mcp.json entries, and since 2.1.0 project config waits for workspace trust, but no prompt-injection guidance was found for fetched web pages (6). Sessions on disk, kimi export, the kimi vis session visualiser, a diagnostic log and hooks for auditing (11). SECURITY.md names private GitHub advisories and an email, supports only the latest version and gives no response time. No security.txt was found on kimi.com, kimi.ai or moonshot.ai, and OSV, the public mirror of the GitHub Advisory Database, lists no advisories for the package, though 2.1.0 shipped security fixes (7).

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.

Kimi Code CLI 71

Framework reading, adapted to a harness driven by a pipeline. Global tools.enabled and tools.disabled lists, per-agent tool lists, deferred MCP tool loading and a read-only Plan mode (20). stream-json output and a step limit per turn by environment variable, but steps are unlimited by default and there's no wall-time or cost limit for the main turn (13). The server API returns coded envelopes and ACP returns standard JSON-RPC errors, but print mode documents no exit codes outside goal runs (12). Retries only on transient failures and never on an exhausted quota, with --continue, --session, fork, undo and ACP session/load (16). kimi -p needs one argument, with ACP and the REST API as programmatic surfaces but no SDK package in any language (10).

AXLR 79

v0.15.0 on 2026-10-25, with commits on main to 2026-11-01 (30). 21 releases between 4 and 25 October 2026 (20). One maintainer (368 of 399 commits, the rest by two agent accounts), 10 open issues that are his own trackers with no comments, about 100 merged pull requests and no outside report to measure a reply against (15). Not in the MCP registry; the Go module and a proto to generate clients from are the only client code, beside the GHCR image and Helm chart (5). Go 1.26 toolchain pinned, release actions pinned by commit SHA, a go-licenses check that the third-party bundle matches the dependency graph, and CI passing on main (9). 30+20+15+5+9 = 79.

Kimi Code CLI 72

2.2.0 on 14 October 2026 (30). 32 releases on npm in the 90 days to 14 October (20). Pull requests are closed at pace, 2.3k closed against 467 open, and the changelog credits merged pull requests up to #4159 from many contributors, with 742 commits in the 90 days to 18 October. Issues lag, with 1.1k open against 391 closed, and reply times couldn't be read (11). No SDK packages. The ACP server and REST API ship inside the CLI, and Homebrew core is current at 2.2.0 (5). CI with a frozen lockfile, lint and a dependency check (sherif), with no Dependabot file found and the pass state unchecked (6).

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.

Kimi Code CLI 62

MIT (30). The docs have no privacy page for the CLI. The Kimi terms (effective 13 August 2026, NOVASCENT PRIVATE LIMITED, Singapore) let Moonshot use content to train its services unless the user opts out by email, prospectively only, and the Kimi privacy policy says content is used for training. The Kimi Platform privacy policy keeps inputs while the account is active. Neither says which applies to CLI sessions (12). The kimi server tree prints a deprecation notice until the next major version, a table names each deprecated config key and the version that replaced it, and the Python Kimi CLI was archived with a migration guide, though its notice gives no date for when old installations stop working (13). Telemetry is documented as on, with a config key and an environment variable to turn it off, but what it collects isn't listed, and print mode ignored the opt-out until 0.41.0 (9).

AXLR 80

An OpenAPI 3.1 file for the service (15 paths, 23 schemas, mutualTLS security scheme, additionalProperties: false on request bodies), a checked proto for the gRPC service, and the 19 MCP tools built from the same shared schemas (25). No llms.txt on either site (404), and the docs site renders HTML; the Markdown sources are in the repository (5). The product contract tables what each of the four interfaces can and cannot do, the work-mode table states each mode's purpose and tool policy, the transport parity ledger names every verb's arguments and behaviour, and axlr_tools tells the model to look for a tool before forging one (15). Typed inputs throughout, Idempotency-Key 16 to 128 safe ASCII characters, expected_revision a positive integer, policy levels, sandbox profiles and egress values enumerated, budgets as four integers, and the service rejects unknown fields, invalid types and bodies over 4 MiB (14). Examples in every guide, an error shape of code, message and request_id, named codes (409 idempotency_conflict, 400 invalid_cursor, 422 invalid_policy, invalid_budget, sandbox_unavailable, not_ready, invalid_id), a gRPC status mapping, and the worker's five statuses and three exit codes; no single error reference page (14). /v1 routes and an OpenAPI version of 0.1.0 against a 0.15.0 release, and no changelog beyond the dated notes inside the docs (7). 25+5+15+14+14+7 = 80.

Kimi Code CLI 85

Framework reading. The kimi web server serves /openapi.json and /asyncapi.json, generated from its runtime validation schemas, but only from a running server behind auth, and the API is marked experimental. The ACP server's methods and capabilities are tabled (20). llms.txt indexes every page and each page is served as Markdown (10). The permission-mode, goal-mode and MCP pages say when to use each mode and what to avoid (16). The config reference gives type and default for each key, with enums for permission modes, output formats and rule decisions (13). Many examples, the server's error-code bands and ACP error codes are documented, but print-mode exit codes are documented only for goal runs (11). A dated changelog in the docs plus the technical changelog in the repository (15).