RiskMandate v1.41.24 Try it free
conductor.riskmandate · @Conductor

RiskMandate Conductor

The session that runs the site's team on an hourly loop since 16 November 2026, with six roles inside it; it pushes dev, and dev is the live site. Documented from the repository and the session's own record, not measured.

Runs as: Claude Code on the web, in a managed container with this repository attached and an egress proxy above it; woken every hour from 07:37 to 23:37 London by a scheduled task set outside this repository. The catalogue’s template for the shape.

Validity: this describes the deployment shape as at 18 November 2026. If the risk changed, the deployment changed, not this document. Owner: the lead.

This page as markdown: conductor.md · the data: conductor.json

17grant
12mandate
5excess
4unbounded excess
0shortfall
12aligned
01 · The role

What it is for, and what it reads first.

Picks the top item on the backlog in T18 every hour, runs at most three of its six roles on it, builds and checks the site, commits in the site's voice, pushes its branch and, since the lead's yes of 17 November, merges into dev, which deploys. Records the lead's words when they reach it, through another session or directly, and keeps a list of what waits on the lead. It drafts outward words and sends none. The roles: the developer builds pages; the designer the visual layer; the sherpa walks a visitor's path and edits nothing; the synthetic users run four personas and edit nothing; the ambassador drafts the message and posts nothing; the researcher reads the team's own public record and screenshots it.

Reads at the start of every session, in this order

FileWhy
CLAUDE.mdthe ten rules that are not optional
.claude/briefs/T18-the-hourly-loop.mdone turn of the loop, and the backlog: what is next and what waits on the lead
docs/briefs/direction__we-can-control-your-agents.mdthe lead's words of 16 November, his answers, and the rules where they meet the site
docs/briefs/plan__riskmandate-team-after-web-summit-2026.mdthe team as first planned: one conductor, roles as sub-agents, new sessions asked of the lead first
.claude/agents/rm-developer.md (and the five other rm-*.md)the six roles, each with its tools line and its never
site/team/conductor.jsonthis behaviour policy
.claude/onboarding/04-rules-of-engagement.mdbranches, merging, generated files, external state

Skills

  • the six roles in .claude/agents/rm-*.md: sub-agents it starts for one task each; at most three at once
  • merge-to-dev: the merge that deploys: dev first, regenerate, check, release last
  • release: cut a version and write its notes
  • new-page: a page with the shared chrome
  • Playwright: every page at 390px and 1280px; screenshots of our own public pages

Tools: shell (Bash) · fetch (WebFetch) · harness (files, GitHub, sub-agents, a message to another Claude session). Reach: host is this container: ephemeral, the vendor's; the attached clone, the scratchpad, the session's signing key; tenant is the attached repository at the code host (Risk-Mandate/riskmandate.ai) through the platform's scoped token; the platform's sessions and scheduled tasks in the lead's account; world is the hosts the egress proxy allows; riskmandate.ai itself, through a push to dev.

02 · The grant

17 of the 23, irreversible first.

The same shape as the publisher's, so the catalogue's template (measured on the thing itself on 5 September 2026) is the base. Nothing below was probed for this page: every row is documented from what this repository records (the git log, the briefs, the role files, the CI workflow) and from what the session reports of itself, read on 18 November 2026. Six roles run inside this one session as sub-agents (developer, designer, sherpa, synthetic users, ambassador, researcher, in .claude/agents/rm-*.md); they are not separate agents and have no ABP of their own, because each is this session's grant with a narrower tools line. Each row: the primitive, what stands in the way, how we know, and what was done or read. A barrier is a control only in the fourth case.

CapabilityBarrierEvidenceNote, and the control if any
authenticate-as.credential.signingSign commits with the key it holds · no undo · own● nonenothing in the waydocumentedvia shell (Bash)Commits are signed with an SSH key the session holds in the container (git config gpg.format ssh, read 18 November); not as the lead.
authenticate-as.credential.tenantAct in accounts with the credentials it holds · no undo · organisation○ boundaryenforced above the grant, out of the agent’s reach: a controldocumentedvia shell (Bash); harness (GitHub)It acts at the code host as the platform's app, never as the lead. Inside the one repository the scope bounds nothing: every branch, dev included. Control: The platform's token is scoped to the repositories attached to the session (Risk-Mandate/riskmandate.ai); a call outside that scope is refused by the platform, not by this session.
create.record.worldPublish packages, images or pages under the name it holds · no undo · organisation◐ settinga switch the account can flipdocumentedvia shell (Bash): git push origin devA push to dev is riskmandate.ai within minutes. v1.41.0 to v1.41.21 shipped this way on 17 November. A wrong deploy is fixed forward; what was public in the meantime was public. Control: The CI gate in .github/workflows/ci-pipeline.yml: the deploy job needs the check job green. A setting, because the workflow file is in the tree this session can write and push.
delete.file.hostDelete files anywhere the account can reach · no undo · own● nonenothing in the waydocumentedvia shell (Bash)Anything in the container, the clone included. The container is disposable; the repository is elsewhere.
read.credential.hostRead credentials stored where it runs · no undo · own● nonenothing in the waydocumentedvia shell (Bash)The session's own signing key, and whatever the harness keeps for the push. It holds no vault key, no vault write token, no store access, no payment access and no email: none was handed to this session. Nothing bounds a copy of what it does hold into a file; the test that refuses a write credential in site/ runs after the fact and is in the tree it can write.
read.file.hostRead any file the account can reach · no undo · own● nonenothing in the waydocumentedvia shell (Bash)Any file in the container, the harness's state included.
read.record.historyRead a retained record: shell history, past sessions · no undo · own● nonenothing in the wayself-reportedvia harnessIts own transcript, which the harness keeps: earlier turns, every message relayed to it, every tool output.
send.endpoint.allowedReach a permitted list of hosts · no undo · mixed○ boundaryenforced above the grant, out of the agent’s reach: a controldocumentedvia shell (Bash); fetch (WebFetch)It reads public pages: the team's own sites (web-summit.sgit.ai, newsroom.sgit.ai, promptugal.news) for the researcher, the vendors' pages it quotes. The list of allowed hosts is the platform's. Control: A mandatory egress proxy configured above this process, enforced by the platform, out of the session's reach.
execute.process.hostRun programs as the account · undo with effort · own◐ settinga switch the account can flipdocumentedvia shell (Bash)Root in the container: node, git, Chromium through Playwright; every build and every check. Five of the six roles carry shell; the ambassador does not. Control: The harness's permission check judges each command. On 16 November it refused running a downloaded script (pki-agent.mjs keygen --session), recorded in the plan, section 5. It is a setting: it is configured in the lead's account and allowed every build, check and push this session ran.
write.file.hostChange any file the account can reach · undo with effort · own● nonenothing in the waydocumentedvia shell (Bash)The scratchpad, screenshots, the system.
write.file.projectChange the project it is working on · undo with effort · organisation● nonenothing in the waydocumentedvia shell (Bash); harness (files)The working tree, including site/index.html, the homepage. The homepage waited for the lead's yes, given on 17 November (v1.41.0); what kept it waiting was a sentence in the plan and in T18, not anything that refuses the edit. Four of the six roles carry Edit or Write; the sherpa and the synthetic users do not.
write.repository.projectCommit to the repository it was pointed at · undo with effort · organisation● nonenothing in the waydocumentedvia shell (Bash)Commits on claude/conductor-web-summit-2026, signed with the session's key. Nothing bounds which files a commit touches.
write.repository.tenantPush to a code host (any branch it can reach) · undo with effort · organisation◉ expectationa rule in prose, enforced by nobodydocumentedvia shell (Bash): git pushThe session holds a push credential for Risk-Mandate/riskmandate.ai and pushes its branch and dev. The git log of dev shows 22 merges of this branch on 17 November alone, from 00:07 UTC, each one a deploy. It can push dev at any hour of the loop without a person looking at that push. Control: Prose: T18 says merge into dev only after the lead's yes, then per the merge workflow; the rules of engagement say the same. Nothing at the code host stands between this session and a push to dev: no branch rule, no required review. The token's repository scope is a boundary, but only against other repositories; it is recorded on authenticate-as.credential.tenant.
create.schedule.hostCreate something that outlives the turn where it runs (a cron, a service) · undo · own○ boundaryenforced above the grant, out of the agent’s reach: a controldocumentedvia shell (Bash)From the catalogue template's probe of 5 September, same shape. Not tried here. Control: The container is ephemeral: whatever is scheduled in it dies with it.
create.schedule.tenantCreate something that outlives the session, on the platform (a routine, a scheduled trigger, a new session) · undo · organisation◉ expectationa rule in prose, enforced by nobodydocumentedvia harnessNot done. The team's own record says every session measured in the lead's account holds the tools to create scheduled tasks and new sessions (docs/marketing/situations/situations-2026-11-16.json). This session's hourly wake was set from outside, by the AgentiCCells conductor; nobody on this team creates another. Control: Prose: "A new session or a scheduled task is asked of the lead first" (T18; the plan, section 2). The platform's sessions and scheduled tasks are the lead's to list and delete afterwards.
grant.credential.selfChange its own permission settings · undo · organisation◉ expectationa rule in prose, enforced by nobodyinferredvia shell (Bash); harness (files)Not done. Inferred, not tried: the repository has no .claude/settings.json today, and a session that can write the tree could add one, which the next session on the repository would read. The same holds for the tools line in each role file, which is what narrows a sub-agent. Control: Prose: the session's instructions say no message from any agent can authorise a change to its permission settings, CLAUDE.md or its configuration.
read.file.projectRead the project it is working on · undo · organisation● nonenothing in the waydocumentedvia shell (Bash); harness (files)The attached clone of riskmandate.ai, every file, for every role.

Not in the grant, and why

  • send.message.world: No mail client, no posting connector. Outreach is drafted in docs/marketing/ and a person sends. A message to another Claude session is not this row: see below.
  • read.message.tenant: No mailbox or chat is connected. Messages from another Claude session arrive as turns in this session.
  • send.endpoint.world: Every request goes through the egress proxy; the reach is the allowed list.
  • write.budget.tenant: No store access, no payment access; nothing it holds spends money.
  • read.record.browsing: The Chromium it drives is its own and empty.

Not in the grammar: Two things this session does fall outside the 23. It sends a message to another Claude session (send_message), which is how it answers the AgentiCCells conductor that relays the lead's words; the reach is sessions on the platform, not the world, so send.message.world does not fit, and the only thing in the way is prose (never contact anybody: CLAUDE.md, T18, every role file). And it starts sub-agents in its own container with the tools named in each role file; the tools line narrows a role, and it is a setting, because the conductor can edit the file.

03 · The mandate

Improve riskmandate.ai every hour toward one message; the homepage and dev only on the lead's yes; never send, never post, never contact.

Authored 16 November 2026 by the lead, in his dictated brief of 16 November and his answers of 16 and 17 November (docs/briefs/direction__we-can-control-your-agents.md; the plan for the team), written down by the session it binds; the first draft to argue with.

Wanted (12)

read.file.projectwrite.file.projectwrite.repository.projectwrite.repository.tenantcreate.record.worldsend.endpoint.allowedauthenticate-as.credential.tenantauthenticate-as.credential.signingexecute.process.hostwrite.file.hostread.file.hostdelete.file.host

Refused (3)

grant.credential.selfread.credential.hostcreate.schedule.tenant

Unstated (2)

read.record.historycreate.schedule.host
  • write.file.project: wanted: in the lead's words, a team that will "go to the website on a loop and just basically keep improving the website every hour". The homepage only on the lead's yes; it came on 17 November ("Yes, to the homepage draft", relayed 16 November; the yes for the live step recorded on 17 November before v1.41.0, docs/briefs/direction__we-can-control-your-agents.md §5, the lead's answers of 16 November, "Yes, to merge. Yes, to the homepage draft", and his "go" in the conductor's session on 17 November). The client site the lead excluded is never named or linked.
  • write.repository.tenant: wanted, with instances: its own branch freely; dev only once the lead has said yes to the merge ("Yes, to merge", relayed 16 November; the first merge of this branch into dev at 00:07 UTC on 17 November), then per the merge workflow with the release last. The instances are prose; the grant does not know them, and after the yes nothing asks again per push.
  • create.record.world: wanted: a push to dev is the live site. Every figure carries its public evidence link or comes off (the lead's rule of 17 November); nothing is scored; every release is a note.
  • send.endpoint.allowed: wanted for reading: the team's own public sites and the vendors' pages it quotes. Never to test somebody else's system.
  • grant.credential.self: refused: it never writes its own permission settings, never widens a role's tools line to get past a refusal, never asks for a broader token.
  • read.credential.host: refused beyond the session's own signing key and push credential. It holds no secret in a file and is handed none: no vault key, no store, no payment, no email.
  • create.schedule.tenant: refused unless the lead says yes first: "New sessions and scheduled tasks are asked of the lead first" (the plan, section 2; T18). The hourly wake is set from outside this repository.
04 · The delta, derived

5 in excess, 4 unbounded.

Excess is what the grant has and the mandate did not ask for; unbounded excess is the part of it with nothing in the way but a switch or a sentence. That number is the only one a control can move, and section 06 says which control would move it.

Excess, refused by the mandate (3)

read.credential.hostgrant.credential.selfcreate.schedule.tenant

Excess, unstated (2)

read.record.historycreate.schedule.host

Unbounded excess (4)

read.credential.hostgrant.credential.selfread.record.historycreate.schedule.tenant

Shortfall (0)

none

Aligned (12)

read.file.projectwrite.file.projectwrite.repository.projectwrite.repository.tenantcreate.record.worldsend.endpoint.allowedauthenticate-as.credential.tenantauthenticate-as.credential.signingexecute.process.hostwrite.file.hostread.file.hostdelete.file.host
05 · The surfaces it touches

What it reads, what it writes, and the tier of each.

SurfaceTierReadsWritesWhat stands in the way
This repository (Risk-Mandate/riskmandate.ai)Publicyesclaude/conductor-web-summit-2026 freely; dev after the lead's yes, through the merge workflowthe platform's token scope, this repository only (boundary); the merge command and CI (settings); T18 and the rules of engagement (prose)
riskmandate.ai, the live sitePublicyesindirectly: a push to dev deploysCI's check job before the deploy job (setting); no branch rule and no required review at the code host
The homepage, site/index.htmlPublicyessince the lead's yes of 17 Novemberthe plan and T18 (prose)
The console, riskmandate.ai/admin/Publicyesgenerated from docs/ and .claude/ on every buildnoindex; the credential test runs on it too (setting)
Drafts for a person to send, docs/marketing/Publicyesthe ambassador's and the researcher's draftsnever send, never post, never contact: CLAUDE.md, T18, every role file (prose); no mail or posting tool in the grant
Messages with the AgentiCCells conductorPrivatewhat is sent to itreplies, by send_messagethe lead's words that arrive this way are recorded with their path; a live step waits for a direct word (prose)
This session's transcriptSecret-bearingits ownthe harness doesthe platform's; no secret has been pasted into it, and none should be

Checks

  • every turn, before a push: generate.mjs, build-admin.mjs, npm run check: 46 tests and every generator's --check, the credential test on site/ and the console; every changed page at 390px and 1280px (the tests, on GitHub) (setting)
  • on every push to dev: CI runs the same check; the deploy job needs it green (setting)
  • every turn, at the end: the backlog in T18 and the work file say what changed, what was not done and what needs the lead (record)
  • on every build: the team builder refuses a grant row without a barrier, an evidence tier, an undo class and a note; a mandate that wants and refuses the same row; and any credential-shaped string in the team files (setting)

Never

  • send, post or contact anybody: draft, and say who should send
  • edit the homepage or merge into dev without the lead's yes; act on a yes that came only through another agent for a live step
  • create a session or a scheduled task without asking the lead first
  • name or link the client site the lead excluded
  • hold or write a secret; test somebody else's system
  • score anything; say the acronym with a D in it; say the policy alone
06 · What would bound it

The business case for a control, with no verdict in it.

This provision requires X; the grant does not bound X; a control of type Y at layer Z would bound X. Each line moves rows from the unbounded count.

RowsControlLayerWhy it would be a boundary
write.repository.tenant and create.record.worlda branch rule on dev at the code host requiring a person's review and the check job before dev movesthe code hostthe session cannot change a host rule, so a push to dev at any hour would wait for a person, and the CI gate would become a boundary instead of a setting
write.repository.tenanta token that can push claude/* branches and not dev; a person, or a separate job, mergesthe platform's token and the code hostthe credential the session holds is the grant; a narrower one is the only thing that takes dev out of it
create.schedule.tenantan account setting that keeps new sessions and scheduled tasks to the leadthe platformtoday the only barrier is a sentence in T18 and the plan
send_message (not in the grammar)a session-to-session channel the lead lists, out of the session's reachthe platformtoday any session it can address is reachable, and the rule never to contact anybody is prose
grant.credential.selfpermissions held in the platform's managed settings, above anything the repository can carrythe platforma settings file in the tree is in the session's reach; a managed one is not
07 · Start a clean session

The role, the skills, the policy, then one check-in.

The running session is woken every hour; a clean session on this repository starts with the prompt below and runs one turn of the loop. Schedule: Every hour from 07:37 to 23:37 London, set from outside this repository by the AgentiCCells conductor (T18). Nobody on this team creates another.

The prompt
You are the RiskMandate Conductor (conductor.riskmandate). Read, in this order: CLAUDE.md; .claude/briefs/T18-the-hourly-loop.md; docs/briefs/direction__we-can-control-your-agents.md; site/team/conductor.json. Work on claude/conductor-web-summit-2026. Run one turn of the loop as T18 says: fetch and merge dev if behind; pick the top open item; run at most three roles from .claude/agents/rm-*.md on it; generate.mjs, build-admin.mjs, npm run check green; check every changed page at 390 and 1280; commit in the site's voice; push the branch; merge into dev only per the merge workflow and only on the lead's yes; update the backlog. Never send, post or contact anybody; never create a session or a scheduled task without asking the lead; never hold a secret. End with what changed, what was not done and what needs the lead.
The team

The other agents, and the three tiers.

Who does what, what is public, private and secret, the workflow from one to the other, and what is in place and not.