LlamaParse API + MCP by LlamaIndex

# Fix list: LlamaParse API + MCP

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/llamaparse, the October 2026 research run, assessed 1 October 2026. Grade C, 59.6 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on LlamaParse API + MCP: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Security & auth, 48 out of 100, up to 9.1 more on the total

Why it scored 48: LlamaCloud keys can be scoped to one project and, since 26 August 2026, set to expire. The MCP takes OAuth or the key as a bearer token (30). No read-only mode, and the unified MCP includes write actions such as creating project API keys with no documented confirmation. Product endpoints narrow what an agent can reach (8 of 20). Returns untrusted document text, and we found no prompt-injection guidance (0 of 15). expand=usage shows what each job cost, and we found no audit log (5 of 15). No security.txt. The trust centre at security.llamaindex.ai renders only with JavaScript, the privacy page says third-party penetration tests are run with summaries under NDA, and the SDK repository routes security reports to support@runllama.ai (5 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 2. Reliability, 55 out of 100, up to 9 more on the total

Why it scored 55: Statuspage at llamaindex.statuspage.io with five components, including separate LlamaParse API and LlamaParse (EU) entries (20). The front page shows two incidents in its recent window, 400 errors for some extract schemas on 29 September (14:19 to 16:04 UTC) and a high error rate for agentic parse in the EU on 17 September (18:48 to 20:30 UTC). The incidents API is blocked by robots.txt and the history page didn't render, so the rest of the 90 days is unread (10). Free is 20 requests a minute, and per the 30 September check uploads are 50 per 5 seconds per project and classify 40 a second, other endpoints unpublished (10 of 15). The billing page documents 402 when credits run out, and we found no 429, Retry-After or backoff guidance (0). Pro and above list 99.9% uptime as a plan feature, with no SLA terms found (5 of 10). Parse and Extract are GA, with Turbo Extract in beta (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 3. Payments & pricing, 40 out of 100, up to 7.5 more on the total

Why it scored 40: No x402, MPP or L402 (0). $1.25 per 1,000 credits and plan prices on the public pricing page, with credits per page per tier published in the docs (20). Free plan with 10,000 credits a month and no card (20). A person signs up in a browser (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 4. Agent ergonomics, 72 out of 100, up to 4.6 more on the total

Why it scored 72: The unified MCP endpoint has 26 tools per the 30 September check, which scores 15. Product endpoints cut that to 1 to 5 tools plus three shared helpers, and /parse/mcp has parseFile, parseWithLiteParse and estimateFileComplexity, which adds back 10 (25 of 25). Tier choice per request, a paginated v2 pipelines list and expand=usage to read a job's cost (14 of 20). The 402 message is clear, and other error bodies weren't checked (10 of 20). No idempotency key and no MCP annotations documented. Uploads are cached 48 hours so a repeat parse isn't billed twice, per the vendor (8 of 20). Official SDKs in Python, TypeScript and .NET (since 14 August 2026) (15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 5. Schema & documentation, 82 out of 100, up to 2.9 more on the total

Why it scored 82: OpenAPI spec at api.cloud.llamaindex.ai/api/openapi.json, per the 30 September check (25). llms.txt at developers.llamaindex.ai, per the same check (10). The MCP page says which product endpoint to use and why API-key callers should use uploadFileByUrl instead of getUploadUrl. Tool descriptions themselves not read (12 of 20). Typed requests with a tier field and a version to pin, not re-checked in depth (12 of 15). The 402 is documented, and we didn't find a full error reference (8 of 15). Dated changelog and dated parse versions such as agentic 2026-09-07 (15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 6. Transparency & trust, 70 out of 100, up to 2.6 more on the total

Made of editorial 57, provenance 82.

Why it scored 70: Closed service, SDKs MIT (15). Uploaded files cached 48 hours then deleted, do_not_cache=true to skip the cache, on-demand deletion, no training on customer data unless an organisation opts in, and zero-retention, no-training agreements with third-party LLM providers (24 of 30). Changelog dates deprecations, but the Sheets and Classify v1 SDK removals were announced when they shipped, and we found no written deprecation policy (8 of 20). North America and EU regions are disclosed. The subprocessor list sits in the JavaScript-only trust centre (10 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: llamaindex.ai, registered 2023-05-12 (3 years) (7 of 15)
- security.txt: not found (0 of 10)

## 7. Maintenance & community, 80 out of 100, up to 1.8 more on the total

Why it scored 80: Python SDK v2.16.0 tagged on 2026-09-08, 23 days ago, and a changelog entry on 1 September (30). Python SDK releases 2.13.0, 2.14.0, 2.14.1, 2.15.0 and 2.16.0 since 22 July (20). Public changelog and GitHub SDK repositories, issue replies not sampled (10 of 15). Current official SDKs in Python, TypeScript and .NET (15). Release-please, CI and test workflows, but 2.14.1 had to restore API surface deleted by a bad regeneration, and two breaking changes shipped as minor versions (5 of 10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## Deductions

Each comes off the total. A fixed and documented problem counts for less at the next check.

- 2026-08-12 to 2026-09-08. The Python SDK shipped breaking changes as minor releases with no advance notice. 2.14.0 renamed files.get() to files.content(), and 2.16.0 removed the classify v1 job methods. Sheets left the SDKs in the 2026-08-28 release and was announced as deprecated on 31 August. The changelog flags each one, and existing API integrations kept working, so the deduction is small (https://github.com/run-llama/llama-cloud-py/blob/main/CHANGELOG.md, https://developers.llamaindex.ai/llamaparse/changelog/)

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- unchecked: incident history before mid-September. The Statuspage API is blocked by robots.txt and the history page didn't render
- unchecked: certifications and the subprocessor list in the JavaScript-only trust centre
- unchecked: the unified MCP tool count, 26 per the 30 September check
- Whether the 99.9% uptime on Pro and above is backed by SLA terms

## Weaknesses

- Breaking SDK changes shipped as minor releases (files.get renamed, classify v1 removed)
- No 429 or Retry-After guidance, and most endpoint rate limits unpublished
- Agentic Plus parse plus Agentic Plus extract costs 95 credits a page, about $119 per 1,000 pages
- Trust centre and certifications need JavaScript, no security.txt
- Two incidents of about 1 hour 45 minutes each on 17 and 29 September

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Connect to a product endpoint such as `https://mcp.llamaindex.ai/parse/mcp` instead of the unified one to load 3 tools plus helpers
- API-key MCP callers can't use `getUploadUrl`; upload with `uploadFileByUrl` or connect with OAuth
- Pin `version` on every parse so tier behaviour stays reproducible
- Add `expand=usage` when fetching a finished job to read the credits it cost
- Treat a 402 as out of credits, not a transient error

## What the review panel asked for

- Publish an error reference
- Document 429 handling
- full error reference

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.