Atticus

# Fix list: Atticus

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/atticus, the October 2026 research run, assessed 15 October 2026. Grade E, 42.9 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 Atticus: 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. Agent ergonomics, 9 out of 100, up to 14.8 more on the total

Why it scored 9: These lines measure an interface an agent calls, and Atticus has none. No response sizing (0) and no pagination or filtering (0). Failures are reported to a person in the overlay, with a Try again control (2 of 20). Destructive commands ask first, deletes go to the Trash, and interrupted actions are checked before a retry, per the agent prompt (3 of 20). One install line and few settings, but a paid Claude or Devin plan, macOS 26 and several macOS permissions are needed. No SDK (4 of 15). Total 9.

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.

## 2. Schema & documentation, 15 out of 100, up to 13.8 more on the total

Why it scored 15: Graded on the interface an outside agent can use. Atticus has no API, MCP server or documented command for another agent, so there is no machine-readable contract (0). No llms.txt on the project site (404) (0). The README explains each function for a person, with what it does and does not do, such as drafts that are never sent (6 of 20). No typed inputs (0). A table of example requests and documented install, test and removal commands (4 of 15). Two tagged releases with dated notes, no changelog file (5 of 15). The internal `atticus.py` flags its own model calls are not documented as an interface. Total 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.

## 3. Security & auth, 49 out of 100, up to 8.9 more on the total

Why it scored 49: Harness reading of the framework checklist, as for other software that acts on the owner's machine. 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. Speech is turned into text on the Mac. Request text, screen text and screenshots go to Anthropic or Devin through the owner's CLI, a daily random ID goes to PostHog by default, AI meeting summaries send transcripts to Claude by default, and the app pulls and builds the main branch every six hours by default with no signature or review step (15 of 30). Safe mode is on by default. The model's shell commands run in Claude Code's sandbox with writes limited to a scratch folder, app-control commands on an allowlist run outside it, delete, send, buy, file-change and system commands ask first by pattern, and rm, sudo, dd, diskutil, shutdown and git push are denied. Main requests run with `bypassPermissions` behind Atticus's own hook, and Full access mode drops some prompts (14 of 20). Every system prompt we read marks screen text, tool output and supplied context as untrusted data, and the meeting filter has no tools. No user-facing guidance on prompt injection (9 of 15). Every request, step and cost goes to ~/.config/atticus/log.jsonl, mode 0600, and the README says each exact command is in the ask log (11 of 15). No SECURITY.md, security.txt, bug bounty or advisories (0 of 20). Total 49.

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.

## 4. Reliability, 68 out of 100, up to 6.4 more on the total

Why it scored 68: Read with the local-software lines, as software the owner runs. It installs by a curl-to-bash script that clones the repository and builds the app, or a Nix flake that runs the same script, with macOS 26 and Apple's command line tools stated. No package or built binary is published (15 of 20). A public GitHub Actions workflow runs 301 offline tests and an installer syntax check on Ubuntu. 21 of the last 26 runs on main passed, the newest on 15 October 2026 included, and installed copies follow main rather than tags (15 of 25). Five issues and pull requests, all from one contributor. Four were answered and closed within six hours and one, of 13 October, has no reply. No crash reports filed (18 of 25). Two tagged releases with notes. No changelog file and no breaking changes marked, while dozens of untagged commits reach users through the self-update (5 of 15). Tagged v1.0 on 13 October 2026 (15). Total 68.

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.

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

Why it scored 60: No payment protocol (0). Read with the self-hosted rule. Free under MIT with nothing to buy since 11 October 2026, when the $5 Gumroad licence check was switched off. The source keeps the product ID for when sales start (20). No card or Atticus account, though a paid Claude Pro or Max plan or a Devin plan is required (20). Install with one command and no signup with Atticus (20). Total 60.

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.

## 6. Maintenance & community, 51 out of 100, up to 4.3 more on the total

Why it scored 51: v1.1 on 14 October 2026 (30). Two tagged releases in the project's five days. The checklist asks for three (0). Four of five issues and pull requests answered within six hours, one open for two days without a reply, on a small sample (15 of 25). Not an MCP server and no SDK (0). The app uses the Python standard library and Swift with no third-party packages to fall behind, and CI runs on every push, with five failures on main since 13 October (6 of 10). Total 51.

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.

## 7. Transparency & trust, 56 out of 100, up to 3.9 more on the total

Made of editorial 58, provenance 53.

Why it scored 56: MIT in LICENSE and as detected by GitHub, with the bundled thinking-orbs and Plip code under MIT and credited (30). No privacy policy or terms. The README says what leaves the Mac (the daily PostHog ID, opt-in failure reports scrubbed first, AI summaries on the owner's plan) and that audio and the mini screen preview stay local, which matches the source we read. No retention period for PostHog data is stated (12 of 30). No deprecation policy, and the project is five days old (0). The daily usage ping is disclosed in the README and setup, with an opt-out in Settings and by environment variable, but is on by default (16 of 20). Total 58.

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: saarthurr.github.io, no registry record we could read (0 of 15)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## 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.

- Judgement call. Listed in personal assistants because it works for one person, remembers them and runs background agents with their connectors. Desktop automation was the alternative, but that category is for tools an agent starts, and Atticus has no such interface. Tagged `not-for-agents`.
- The lead matched. The repository's first commit is "Atticus: a voice assistant for your Mac", 10 October 2026 01:07 UTC (9 October in the author's time zone). `firstReleased` is the first tag, v1.0 on 13 October 2026. The Show HN post of 12 October shows it was public by then. Whether the repository was public on 10 October was not established.
- `atticus.py` accepts internal flags (`--place`, `--hands`, `--look`, `--fast` and a plain request) that a shell-capable agent could call on a Mac where Atticus is installed. They are undocumented as an interface and change with each commit, so they were not graded as one.
- The source still holds a Gumroad product ID labelled for when sales start, so the price may return.
- We did not install or run the app. Every claim about behaviour comes from the source, the README and the release notes.

## Weaknesses

- No API, MCP server or documented command an outside agent can call
- Pulls and builds the main branch every six hours by default, with no signed or tagged release step
- Five days old, one maintainer, two tagged releases and no changelog or deprecation policy
- No privacy policy, SECURITY.md or security.txt
- Needs macOS 26 and a paid Claude Pro or Max plan or a Devin plan

## 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.

- Don't plan to call Atticus from another agent. It takes requests only from its own shortcut and typing box
- If a user runs Atticus beside you, expect it to move windows and type into apps on the same Mac
- Point a user who wants pinned versions to Settings > About & Updates to turn off self-update

## 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.