Jam MCP vs REA

Jam MCP 45

Status page at status.jam.dev with components for the website, API, extension and integrations, none for MCP (15). 100% uptime shown on the components read (20, no incident history read). No MCP rate limits found (0). No retry guidance (0). No SLA found (0). Not labelled beta (10). 15 + 20 + 0 + 0 + 0 + 10 = 45.

REA 77

Scored as a local stdio package. Official npm package rea-agents, with Node.js ^22.23.4, ^24.22.1 or 26.12.1 and later stated from 6.8.0 (20). The CI workflow runs on every push to main and every pull request, with sharded tests, a Windows native build and package checks. We couldn't see whether main passes, since GitHub's pages and API weren't readable (20 of 25, per the 12 October check). Current issues couldn't be read either, so 15 of 25 as on 12 October. Breaking changes are listed under every release, but the always-bump-minor policy ships them in minor versions, 6.5, 6.6 and 6.8 since the last check (7 of 15). Version 6.8.0 (15). Total 77.

Jam MCP 55

Typed schemas by protocol, not read without sign-in (15 of 25). llms.txt (10). Every tool has a sentence on what it returns, and some say when to use them (15 of 20). Filters documented for getNetworkRequests (statusCode, contentType, host, method, limit, after) (8 of 15). Example prompts, no error responses (5 of 15). Versioned in the registry (1.36.3); no public changelog found (2 of 15). 15 + 10 + 15 + 8 + 5 + 2 = 55.

REA 86

Every tool advertises typed JSON input and output schemas with no recursive references, per docs/mcp-contracts.md (25). No llms.txt on rea.tools. Markdown guides in docs/ and an agent skill installed by setup (5). Descriptions are long and specific. open_binary explains formats, provider binding and snapshot rules, and some say when to use them, such as procedure_assembly (15 of 20). Inputs are typed, but Evidence passed between tools travels as unconstrained JSON objects (11 of 15). An error-contract JSON Schema with codes, retryable and remediation fields, a user-visible error table and contract examples (15). Release Please versions with a dated changelog and a migration guide (15). Total 86.

Jam MCP 60

Registry version 1.36.3 updated 3 November 2026 (30). Dated changelog not found to count, though the registry version implies frequent releases (10 of 20). Help centre and feedback tool, no public changelog (10 of 15). Official registry entry dev.jam/jam (10 of 15, its repository field points at a skills repository). No package to read (0). 30 + 10 + 10 + 10 + 0 = 60. Registry presence read from Anchor Terminal's own daily import of the official MCP registry (/api/v1/registry.json, 6 November 2026), not the registry's API.

REA 88

6.8.0 on 25 October 2026, seven days before this check (30). 16 changelog releases since 3 August, from 3.1.0 (9 August) to 6.8.0, four of them since 12 October (6.5.0, 6.6.0, 6.7.0 and 6.8.0) (20). The changelog credits many outside contributors and closes issues in every release. We couldn't read current reply times on issues (15 of 25). In the official MCP registry as io.github.morluto/rea, latest 6.8.0 published 25 October (15). Exact-pinned dependencies, CI on push, no Dependabot config, per the 12 October check (8 of 10). Total 88.

Jam MCP 50

About 37 tools (5), with a getDetails snapshot that suggests which tools to try next (5 back). Filters and limits on network requests and search (14 of 20). No documented errors (3 of 20). Delete tools exist (deleteJam, deleteFolder, deleteComment); annotations not read (8 of 20). A CLI with the same fields for every tool (15). 10 + 14 + 3 + 8 + 15 = 50.

REA 73

229 tools in the 6.8.0 effect audit, one key per tool (170 at the 6.4.0 tag by the same count, and 172 on main when the 12 October check counted it), all advertised at once (5). Since 6.8.0 input schemas are compact by default, with the full profile behind REA_MCP_INPUT_SCHEMA_PROFILE=full, but there are no toolsets, dynamic loading or read-only subset, so no add-back. Large results return resource_constraint with a ready-to-call inspect_analysis_view continuation and next_offset paging, and get_evidence_bundle takes detail: summary with filters (18 of 20). Typed errors with category, retryable and remediation (20). Every tool's readOnlyHint, destructiveHint and idempotentHint come from an audited effect record (20). One command sets up the named clients and the CLI mirrors the tools, but it is TypeScript only and many tools need long structured inputs (10 of 15). Total 73.

Jam MCP 40

No machine payment protocol (0). Public per-seat price ($14) (10). Free plan for everyone (20). Personal access tokens exist but a person creates them after a browser sign-up (10 of 20). 0 + 10 + 20 + 10 = 40.

REA 60

Free, self-hosted, MIT, nothing to buy from the project, so 20 + 20 + 20. No payment protocol (0). Hopper and IDA are paid products from other vendors and are optional, since Ghidra is free. Total 60.

Jam MCP 48

Closed service; terms URL not found at /terms (404), privacy policy read (10 of 30). The security page says Jam opts out of training on customer data, describes backup retention and custom retention for Enterprise (18 of 30). No deprecation policy found (2 of 20). AI vendors are said to retain data for limited periods; a full subprocessor list wasn't read (15 of 20). 10 + 18 + 2 + 15 = 45.

REA 59

MIT (30). Local software. The README says targets are analysed locally and the model provider's own policy applies to tool results. SECURITY.md describes what the Ghidra session keeps and deletes, and browser captures drop cookies and authorisation headers. No privacy statement beyond that (20 of 30). Breaking changes are dated in the changelog with a 6.2 migration guide, but the policy says minor versions carry no compatibility promise and security fixes go to the latest only (8 of 20). No telemetry in the source, which names no outside host except github.com and Hopper's download origin. There's no written statement (20). Total 78.