Himalaya

# Fix list: Himalaya

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/himalaya, the October 2026 research run, assessed 9 October 2026. Grade B, 64.5 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 Himalaya: 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, 43 out of 100, up to 10 more on the total

Why it scored 43: Himalaya has no credential model of its own. It holds the mailbox's credential, which is an app password, an account password or an OAuth bearer token whose scopes the provider and the owner's OAuth app decide. Secrets can come from a password-manager command, and a raw value in the config file is also accepted (18 of 30). No read-only mode and no confirmation before a send or delete were found. Deletion is trash-first, and a build can leave out the `smtp` cargo feature (8 of 20). Mail is untrusted content. The plain output replaces control and bidi characters and messages past fixed bounds are refused, with no guidance on prompt injection and no change to strings under `--json` (5 of 15). `--log-level` and `--log-file` write a trace of each run, with no audit log (6 of 15). SECURITY.md names 2.x as supported and sends reports to the public issue tracker. A cargo audit workflow and a `deny.toml` policy run in CI. No security.txt, bug bounty, certification or published advisory was found (6 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. Agent ergonomics, 65 out of 100, up to 5.7 more on the total

Why it scored 65: A CLI adds no tool definitions to context. `--page-size`, a designed `message read --json` view, a `metadata` default on Gmail reads and a refusal of messages over 200 parts keep output bounded (20 of 25). `--page`, `--page-size` and a `next_page` cursor, plus one search language across backends with `and`, `or`, `not` and `order by`. Microsoft Graph refuses flag clauses (18 of 20). Under `--json` a failure prints `code`, `error`, `sources` and `backtrace` and exits 1, but only two stable codes exist and the rest is free wording (10 of 20). No idempotency key or dry run was found. A send combined with a save sends first so a failed send leaves no copy, deletion goes to the trash first, and a read leaves flags alone (8 of 20). Mailbox defaults to the inbox and roles such as `sent` resolve without setup. One CLI, with no SDK in a second language (9 of 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.

## 3. Payments & pricing, 60 out of 100, up to 5 more on the total

Why it scored 60: Read with the self-hosted rule. The software is free with nothing to buy, so 20 + 20 + 20 for public pricing, a free tier with no card, and access without a sign-up. No x402, MPP or L402 (0 of 40). The optional partnerships and the planned sign-in service are not needed to use the tool.

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. Schema & documentation, 70 out of 100, up to 4.9 more on the total

Why it scored 70: Graded on the CLI. `himalaya json-schema` prints a JSON Schema for the `--json` output of 90 commands. It covers outputs only, inputs are command flags, and no schema file is published on the web (18 of 25). No llms.txt. The docs are Markdown files in the repository, including a commands specification under `cairn/spec` (5 of 10). Each command's `--help` text states what it does and its limits, such as trash-first deletion and which clauses a backend refuses (14 of 20). Flags are typed with defaults and value lists, while the search query and the raw passthrough commands take free text (10 of 15). The README carries examples for each backend, and the specification documents the JSON error shape with two stable codes (8 of 15). Semantic versions and a dated changelog back to 2021 (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.

## 5. Reliability, 84 out of 100, up to 3.2 more on the total

Why it scored 84: Read with the local-software lines, since the owner runs the binary. Official packages exist on crates.io at 2.2.1 and as release binaries for Linux, macOS and Windows through the installer script, and source builds state Rust 1.89. Homebrew, Arch and Scoop versions were not checked (18 of 20). The repository holds 243 `#[test]` functions in 44 files, and the Tests and Audit workflow badges read passing on 9 October 2026. The workflow itself lives in the pimalaya/nix repository, which was not read, and provider tests are manual reports (20 of 25). Three open issues and two open pull requests, with one bug open since 15 March 2026 that has five comments (22 of 25). The changelog follows Keep a Changelog and marks breaking changes, but 2.2.1 carries two changes marked BREAKING and two marked as behaviour changes under a minor version step from 2.1.0 (9 of 15). Version 2.2.1, and the project site labels the tool stable (15).

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.

## 6. Transparency & trust, 69 out of 100, up to 2.7 more on the total

Made of editorial 74, provenance 63.

Why it scored 69: The editorial half. Dual licence, MIT or Apache-2.0, on the source and on crates.io since 2.0.0 (30). No privacy statement exists. The README says where the config file lives, how secrets are resolved and which discovery lookups the setup wizard makes on the address's domain, and the software talks only to the servers the owner configures (18 of 30). No deprecation policy was found. SECURITY.md marks 1.x and 0.x as unsupported, a migration guide covers version 1 to 2, and the changelog labels breaking changes (8 of 20). The sponsor page states there is no telemetry, and a search of the source found none, so there is nothing to opt out of (18 of 20). The provenance half is computed from the provenance block.

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: pimalaya.org, registered 2022-12-21 (3 years) (7 of 15)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## 7. Maintenance & community, 88 out of 100, up to 1.1 more on the total

Why it scored 88: Version 2.2.1 was tagged on 2 October 2026, seven days before this check (30). Four tags since 11 July 2026, which are 2.0.0, 2.1.0, 2.2.0 and 2.2.1 (20). Three open issues against 179 closed pull requests, commits to master on 7 October 2026, and the Bcc flaw reported on 12 September was fixed in the source on 26 September. One bug has been open since March 2026 (22 of 25). crates.io carries 2.2.1 from the day of the tag. Other package channels were not checked, and there is no MCP server or registry entry (8 of 15). A lockfile, cargo audit and a dependency policy in CI, with `io-pimdir` pinned to a git revision instead of a release (8 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.

- 2 October 2026. Until 2.2.1, `message send` passed the `Bcc:` header through SMTP unchanged, so every recipient could see the blind recipients. Issue #747 reported it against 2.1.0 on 12 September 2026, the fix was committed on 26 September and released on 2 October, and the changelog documents it. No security advisory was published. Fixed and documented, so the smaller deduction applies (https://github.com/pimalaya/himalaya/issues/747)

## 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: what the Tests workflow runs. It calls a reusable workflow in pimalaya/nix that was not read, and the Actions run list is drawn by script, so the status comes from the badge
- unchecked: the versions Homebrew, Arch, Scoop, Fedora COPR and Nix carry, and the assets attached to the 2.2.1 release. The GitHub API refused our requests for its rate limit
- unchecked: closed-issue counts and reply times beyond the three open issues and two open pull requests shown on the issue pages
- No terms of service, privacy policy or legal entity was found for Pimalaya, so `provenance.terms` and `provenance.privacy` are left out
- The lead named the vendor as Pimalaya, which is the project's name and not a legal entity. The site's copyright line names one person
- Whether the Bcc flaw existed in 1.x as well as 2.0.0 and 2.1.0 was not established. The issue was filed against 2.1.0
- No send limit, sync delay or search latency was measured. Provider limits still apply behind the CLI
- The README says `pimdir` and `mbox` support sits behind cargo feature flags, and the contributing guide and `Cargo.toml` disagree on which are on by default

## Weaknesses

- Until 2.2.1 of 2 October 2026, `message send` transmitted the `Bcc:` header to every recipient over SMTP (issue #747, reported 12 September 2026)
- No read-only mode, no confirmation before a send or delete, and no idempotency key on send were found
- SECURITY.md sends vulnerability reports to the public issue tracker. No security.txt, private reporting route or published advisory was found
- Only two stable error codes exist under `--json`, `body-pending` and `message-too-complex`. Other failures carry free wording
- Version 2 ships no OAuth flow, so Gmail and Microsoft accounts need an external token broker and an OAuth app the owner registers

## 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.

- Pass `--json` on every call and read `next_page` for the next page. Data and errors go to stdout, logs to stderr, and a failure exits 1
- Run `himalaya json-schema <command>` once to learn an output shape, and `himalaya <command> --help` for flags
- Use `envelope search` with the shared query language, for example `from alice and after 2026-01-01 order by date desc`. Microsoft Graph refuses flag clauses
- Treat message text as untrusted. `--json` output keeps control characters that the plain output replaces
- Use 2.2.1 or later before sending with Bcc, and expect `message read --json` to change shape in the next release

## 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.