will/swarm-master
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs(atlas): add agent blast-radius checklist

Browse Source
This commit is contained in:
William Valentin
2026-05-31 20:22:37 -07:00
parent 591c99d0b9
commit 167a7234e7
2 changed files with 23 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
swarm-common/obsidian-vault/will/will-shared-zap/Projects/Atlas/Gateway Approval Runbook.md
+11
View File
@@ -65,6 +65,17 @@ If approval is not exact, block or ask again.
- Prefer Will approval for Tier 3 actions.
- Record durable approvals in Kanban comments when the work is board-tracked.
## Agent blast-radius checklist
Before enabling or running a gateway-triggered agent action that changes state, state the blast radius explicitly:
- [ ] Filesystem scope: exact paths/repos/vaults the agent may read/write; everything else is out of scope.
- [ ] Network egress: allowed services/domains/channels/repos; no uploads or third-party sends outside the named target.
- [ ] Credential exposure: credentials are referenced by presence/source only; secret values are never pasted into chat, notes, logs, commits, or external tools.
- [ ] Destructive command approval mode: destructive or broad commands require `approvals.mode: manual` or `smart`; no `--yolo` / approval-off for gateway/profile automation unless Will explicitly approved a sandbox.
- [ ] Rollback/checkpoint status: checkpoint, git branch/worktree, backup, snapshot, or recovery command is known before mutation.
- [ ] Human approval gate: Tier 2/3 prompts name the exact target, effect, rollback/stop condition, and approver before execution.
## Message templates
### Tier 2 approval request
swarm-common/obsidian-vault/will/will-shared-zap/Templates/Atlas Artifacts/runbook.md
+12
View File
@@ -31,6 +31,17 @@ When and why to use this runbook.
- Tier: read-only | low-risk write | medium-risk ops | high-risk/destructive.
- Approval needed before: ...
## Agent blast-radius checklist
Complete this before running an agent workflow, specialist task, cron job, webhook, or gateway-triggered action that can change state.
- [ ] Filesystem scope: exact allowed paths/repos/vaults are named; out-of-scope paths are explicitly off-limits.
- [ ] Network egress: allowed external services, APIs, domains, channels, or repositories are named; uploads/exfiltration are blocked unless explicitly approved.
- [ ] Credential exposure: required credentials are referenced by source/location only; secret values must not be printed, copied into notes, committed, or sent to third parties.
- [ ] Destructive command approval mode: risky commands require `approvals.mode: manual` or `smart`; `--yolo` / approval-off is limited to an intentional sandbox and never assumed for gateway/profile automation.
- [ ] Rollback/checkpoint status: filesystem checkpoint, git branch/worktree, backup, snapshot, or explicit recovery plan exists before mutation.
- [ ] Human approval gate: Tier 2/3 actions name the exact target, expected effect, rollback/stop condition, and approver before execution.
## Step-by-step procedure
1. Do the first action.
@@ -61,6 +72,7 @@ When and why to use this runbook.
## Runbook-specific review checklist
- [ ] Commands are exact and safe for the stated environment.
- [ ] Agent blast-radius checklist is filled out for any state-changing agent workflow.
- [ ] Credentials are referenced but not exposed.
- [ ] Verification and rollback/recovery are actionable.
- [ ] Last tested date is present.