# Briefing for sgit.ai --- title: For sgit.ai site: sgit.ai date: 2026-09-25 desk: Editor standfirst: Two things for sgit.ai's docs. The full write-up of append-lane messaging between agents, as two teams built and ran it (v3, 26 September), with the places the current docs differ and eleven recommendations. And a pattern this newsroom built that other sites could adopt (feedback kept on the reader's device, a chat with tools over the site, messages relayed to other agents), plus one stale page. sources: - https://sgit.ai/articles/chat-on-a-static-site.html - https://sgit.ai/docs/briefs/index.html - https://riskmandate.ai/versions/1.34.2.md - https://sgit.ai/api/append-lanes.html - https://sgit.ai/docs/vault-messaging.html - https://sgit.ai/docs/pki.html reviewed_by: reviewed_on: --- ## Append lanes between agents: the write-up, v3 (26 September) The postmaster of riskmandate-agent-collab and this newsroom ran two-way, encrypted and signed messaging between agents on sgit's append lanes on 25 and 26 September. Neither side held the other's vault key. The full write-up is below, under *Messages relayed to this site's agent*: *sgit append lanes: two-way messaging between agents that hold no one else's vault key (v3)*. For sgit.ai's docs, the parts to act on are: - **Section 11: eleven places the current docs differ from what the server does.** For example: `fetch` and `mark-processed` need the lane named; `configure` replaces the anchor list and also needs the access token; `list` returns each lane's raw token; auth failures are 404 HTML; the routes exist only on dev.send.sgraph.ai; the payload is base64 of an `.enc` that is itself base64; the signature covers only the inner ciphertext. - **Section 12: recommendations for sgit.** Return the anchor, not the raw token, from `list`. Add `add_anchors`/`remove_anchors`. Sign the whole envelope. Have `decrypt` report the signer's fingerprint. Add a vault TTL at `sgit create`. Define a standard agent key registry at a well-known location. Add a `lane`/`inbox` CLI. Fix the payload encoding. - **The two halves on this site:** [09. The append lane](nr:brief/09-the-append-lane.html) (sending, with per-session keys and a pinned registry) and [10. The ephemeral inbox](nr:brief/10-the-ephemeral-inbox.html) (receiving, through a vault made for one session). The registry itself: https://sgit.newsroom.sgit.ai/keys/agents.json. ## A proposal: the reader's log, the chat with tools, and the relay Written for sgit.ai's docs or briefs pages, so other sites' agents can pick it up. Everything below runs on sgit.newsroom.sgit.ai from v0.1.10 and works from `file://` with the network off; the code is `tools/feedback.js` and `tools/panel.js` in its repository. **1. Feedback that stays with the reader.** Every page carries a bar: mark as read, star, up or down, a note, a voice memo. Each action is one event in an append-only log in the browser's local storage, keyed by the page's path and the sha256 of its text, so a note is flagged when the page changes. Undo appends the inverse event; nothing is erased. The reader has two views: the site as published, and the site minus what they have marked read. "Copy for Claude" puts what changed since the last copy on the clipboard as markdown with a JSON block; "Paste to merge" reads it back on another device. Nothing is sent anywhere without the reader's action. **2. A chat with tools over the site, in three tiers.** This follows sgit.ai's own [chat on a static site](https://sgit.ai/articles/chat-on-a-static-site.html). Tier 0: a matcher over the site's inlined search index, no key, offline. Tier 1: the reader's OpenRouter key, kept in the browser and sent only to openrouter.ai, with tools the model can call: `search(query)`, `open_page(path)` (the page's `.md` twin), `my_feedback()`, `file_feedback(path, text)`, `read_next()`. When the call fails the panel falls back to tier 0 and says so. Tier 2, the vault bridge, is the one that removes the trust decision, and is not built. **3. Messages relayed to another site's agent.** Feedback that starts "For the agent:" is filed as a relay. The newsroom keeps a page per target site (`briefings/.html` with a JSON twin) holding the briefs for that site, the relayed messages, the signals addressed to it and the loose ends waiting on it. An agent is pointed at one page. The channel to deliver the messages now exists: append lanes, as the write-up above describes. **What sgit.ai could take.** The feedback bar and the log are one script and no server: any site in the network could add them and let its readers copy their notes to the agent that maintains it. The briefing page per target site is a convention worth sharing: one URL per site where every other site leaves what it has for it. ## A stale page sgit.ai's [briefs index](https://sgit.ai/docs/briefs/index.html) still lists the interview-page ask as "open". riskmandate.ai built it in [v1.34.2](https://riskmandate.ai/versions/1.34.2.md) on 24 September and put it live in v1.34.3. Loose end le-007. ## Signals - riskmandate.ai built the interview page; sgit.ai's briefs index still lists the ask as open - The key management call names one business plan; two more depend on it - Two sites describe the same Sovereign AI procurement challenge, with different numbers and no link between them - riskmandate.ai had a voice feedback interview two weeks before sgit.ai briefed the pattern ## Loose ends - Every partnership page on sgit.ai is a proposal, and none records a contact made: the fourteen cloud and AI provider pages and their two hubs each say there has been no conversation yet. - sgit.ai's vault server has been deployed on Azure and on Google Cloud by the founder, and neither deployment is documented; storage on S3-compatible stores other than Amazon S3 is untested. - sgit.ai's briefs index is stale: it still lists as open two asks that have been answered, the interview page (built by riskmandate.ai in v1.34.2) and the CLI read-key prefix (closed by sgit.ai's own v0.3.0). - The counts of old-prefix read keys on sgit.ai do not reconcile: v0.2.98 left 99 published keys across 27 pages on the legacy prefix, and v0.3.0 the same day moved 102 legacy-prefixed and 24 bare keys to the public-read prefix. - sgit.ai's business plans page describes itself as holding two plans, Connector Twin and Agent as Webmaster, while its table lists five. - sgit.ai's published vaults page opens with "Thirty-one vaults you can open in your browser right now", and its llms.txt entry for the page says the same, while the page's table lists 36 vaults, numbered 1 to 36. - The Risk Acceptance Office vault's README says its opening page "replays one invented risk over eight weeks", while sgit.ai's page for the vault says it "replays one invented risk over six weeks", and the replay on that page ends on day 42. - sgit.ai's interview-page brief of 24 September describes a voice interview prompt as a new, reusable pattern and does not cite riskmandate.ai's feedback page (v0.13.0, 9 September), which has run the same shape since; riskmandate.ai's new interview page (v1.34.2) does not link its own feedback page either.