From aebffa90cf2690b1f01fe3b85d0f096ad46d1548 Mon Sep 17 00:00:00 2001 From: Nivedit Jain Date: Sat, 25 Jul 2026 14:56:31 +0000 Subject: [PATCH 1/2] docs(agenteye): agents can start the sign-in now MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The CLI page told readers "the skill cannot complete the emailed-code login for you" — no longer true. A coding agent can start a device sign-in and hand the user a link and a short code; the user checks the code matches their terminal, chooses what the agent may do, and approves. Value and contract only, per the public-docs bar: what it does and what the user controls, no mechanics of how approval or scoping are enforced. Counterpart to FailproofAI/agenteye#477. Co-Authored-By: Claude Opus 5 Claude-Session: https://claude.ai/code/session_013uML35iHo8GmymozNwCrJD --- docs/agenteye/cli-and-agents.mdx | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/docs/agenteye/cli-and-agents.mdx b/docs/agenteye/cli-and-agents.mdx index 67e77078..97304504 100644 --- a/docs/agenteye/cli-and-agents.mdx +++ b/docs/agenteye/cli-and-agents.mdx @@ -47,9 +47,13 @@ It is built to run unattended. Confirmation prompts auto-skip when no terminal i Better yet, you should not have to remember any of these flags at all. The **CLI skill** is a small Agent Skill folder named `agenteye-cli` that teaches a coding agent such as Claude Code or Codex to drive the CLI from plain-English requests. Ask "is anything broken today?" and the agent picks the command, runs it as you, and answers in prose. -For Claude Code, drop the `agenteye-cli` folder into `~/.claude/skills/` and it is auto-discovered. Failproof AI Observability provides the folder; there is nothing extra to install, because it only drives the CLI you already installed. Log in yourself first: the skill cannot complete the emailed-code login for you. +For Claude Code, drop the `agenteye-cli` folder into `~/.claude/skills/` and it is auto-discovered. Failproof AI Observability provides the folder; there is nothing extra to install, because it only drives the CLI you already installed. -Because the agent runs the CLI as you, it can do everything your login permits, reads and writes alike: create keys, change settings, resolve incidents. The CLI's "are you sure?" prompt does not fire for an agent, so the skill is written to state the exact command and wait for your OK before any change. You are the confirmation step. +The agent can start the sign-in, too. It hands you a link and a short code; you open the link, check the code matches the one in your terminal, choose what the agent is allowed to do, and approve. The agent picks it up from there. You never copy a code out of your inbox, and the agent never sees one. + +**You decide what it gets.** The approval screen shows the permissions the agent asked for, pre-selected and limited to what you have yourself — you cannot hand over more access than you hold, and you can narrow it before approving. A well-behaved skill asks for read-only and tells you so. Whatever you grant, you can see every signed-in browser and agent and end any of them at any time. + +That matters because the CLI's "are you sure?" prompt does not fire for an agent. The skill is written to state the exact command and wait for your OK before any change — but that is a convention, while the permissions you approve are enforced. ```text you Why did session run-001 fail? From 01c342531b55dbf6e87dc76a7ad7b786ff1e13ba Mon Sep 17 00:00:00 2001 From: Nivedit Jain Date: Sat, 25 Jul 2026 14:58:13 +0000 Subject: [PATCH 2/2] docs(changelog): note the AgentEye CLI agent-sign-in rewrite (#598) Co-Authored-By: Claude Opus 5 --- CHANGELOG.md | 1 + 1 file changed, 1 insertion(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index 37c1b74d..c75570c8 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -24,6 +24,7 @@ - Add a Codex session capture page for AgentEye: what the collector picks up from local OpenAI Codex sessions (CLI, IDE, desktop app), how to turn it on with an `events:add` key, and where captured sessions show up. Value/contract level only, no implementation detail. (#592) - Note in the AgentEye overview that the dashboard and CLI are backed by a REST API callable with a scoped key, so prospects can see a programmatic API exists. Gist/contract level only — no endpoint signatures; the exhaustive HTTP API reference stays in the enterprise docs. (#593) - Add OpenClaw and Hermes session capture pages for AgentEye, and cross-link all three capture pages so a reader landing on any one finds the others. The collector gained both integrations in FailproofAI/agenteye#462 alongside the existing Codex capture, but that PR deferred the public half to "a separate failproofai PR" which was never opened — so befailproof.ai/agenteye has named Codex as the only capturable agent ever since, while the enterprise docs for both shipped correctly. Same value/contract level as the Codex page (#592): what the collector picks up, enabling it with an `events:add` key, where sessions surface, and the privacy caveat — no paths, env vars, or config keys, which stay enterprise-only. Hermes additionally notes that the originating channel (Slack, Telegram, CLI, scheduled run) is recorded, since telling those apart is the reason to capture it. Both registered in the `en` "SDK and CLI" nav group; `bun` could not run in the authoring sandbox, so nav↔file resolution was asserted in both directions across all 27 pages by hand — the file→nav direction being the one `mintlify validate` does not check, and the reason an unregistered page ships with green CI and no sidebar link. (#594) +- Rewrite the AgentEye CLI page's agent section: a coding agent can now start the sign-in itself. The page told readers "the skill cannot complete the emailed-code login for you", which stops being true with FailproofAI/agenteye#477 — the agent runs a device sign-in and hands the user a link plus a short code, the user checks the code matches their terminal, chooses which of their own permissions to hand over, and approves. Also states what the user controls, since that is the part worth knowing before you let an agent act as you: the approval screen is pre-selected with what the agent asked for and capped at what the user holds themselves, and every signed-in browser and agent can be listed and ended at any time. Sharpens the existing note about the CLI's "are you sure?" prompt not firing for an agent — it is a convention the skill follows, whereas the approved permissions are enforced on every request. Value and contract only, per the public-docs bar: no grant type, endpoints, lifetimes, storage, rate limits, or any description of how approval and scoping are enforced — that depth is written separately for the signed-customer set. (#598) ## 0.0.14-beta.3 — 2026-07-20