Skip to content

Commit 644d54c

Browse files
committed
Fix extra tab in journal TSV causing column misalignment
1 parent ab1a71a commit 644d54c

1 file changed

Lines changed: 4 additions & 4 deletions

File tree

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
type typeName facet quality_score quality_max quality_level title content
2-
D Decision 5 5 strong Tier-1 audit-gates-are-managed-agents renamed to audit-gates-are-spawned-agent-sessions The original constraint name leaked vendor identity (Anthropic Managed Agents) into a Tier-1 governance commitment that should have been substrate-agnostic. Rename treats the original as a delayed-review correction, not as supersession with old-URI preserved. Cross-references in writings/ and ledger/ kept as historical record; canon/ active references updated to the new URI. The constraint's substantive content stays the same: governance validation requires a spawned clean agent session that operates as fresh context against the artifact under review. What changes is the URI, the name, and the body's framing — substrate is now a runtime-deployment decision, not a constraint commitment.
3-
D Decision 5 5 strong New Tier-1 method doc spawned-agent-session-substrate-options codifies substrate choice Catalogues the substrate options for hosting a spawned agent session: Anthropic Managed Agents, Cloudflare Sandboxes, Daytona, local Docker, etc. Each option's tradeoffs documented across cost, integration complexity, vendor coupling, security posture, and ergonomics. Explicitly notes the Subscription Inclusion Lever — Claude Code authenticated against an Anthropic Max plan via 'claude setup-token' produces CLAUDE_CODE_OAUTH_TOKEN for headless CI use, allowing inference axis to collapse under subscription. Headline arithmetic: 100 audits/day on Max 20x = $223/mo total vs ~$711/mo on Managed Agents (~70% reduction at this volume). Substrate stays vodka-architecture; opinions live in callers.
4-
C Constraint 5 5 strong Vendor names do not appear in Tier-1 canon URIs Operator-as-validator catch from prior session: vendor naming smell in 'audit-gates-are-managed-agents' was not surfaced by tooled validation. Sibling 'governance-validation-via-agents' has the same shape and is flagged as a follow-up rename in scope of this PR or its successor. Tier-1 commitments name what is required structurally; vendor mappings live in method-level catalogues that explicitly enumerate options.
5-
H Handoff 5 5 strong PR A scope: substrate rename + substrate-options method + journal klappy.dev branch claude/substrate-rename-and-options-8f57. Two file changes: (1) rename canon/constraints/audit-gates-are-managed-agents.md to canon/constraints/audit-gates-are-spawned-agent-sessions.md with content rewrite, (2) new canon/methods/spawned-agent-session-substrate-options.md cataloguing substrate options. Journal entry committed alongside. AMS adoption pointer (PR B scope at klappy/agent-messaging-service) depends on this PR's klappy:// URIs landing first. Once PR A merges, PR B for AMS becomes ready to open.
2+
D Decision 5 5 strong Tier-1 audit-gates-are-managed-agents renamed to audit-gates-are-spawned-agent-sessions The original constraint name leaked vendor identity (Anthropic Managed Agents) into a Tier-1 governance commitment that should have been substrate-agnostic. Rename treats the original as a delayed-review correction, not as supersession with old-URI preserved. Cross-references in writings/ and ledger/ kept as historical record; canon/ active references updated to the new URI. The constraint's substantive content stays the same: governance validation requires a spawned clean agent session that operates as fresh context against the artifact under review. What changes is the URI, the name, and the body's framing — substrate is now a runtime-deployment decision, not a constraint commitment.
3+
D Decision 5 5 strong New Tier-1 method doc spawned-agent-session-substrate-options codifies substrate choice Catalogues the substrate options for hosting a spawned agent session: Anthropic Managed Agents, Cloudflare Sandboxes, Daytona, local Docker, etc. Each option's tradeoffs documented across cost, integration complexity, vendor coupling, security posture, and ergonomics. Explicitly notes the Subscription Inclusion Lever — Claude Code authenticated against an Anthropic Max plan via 'claude setup-token' produces CLAUDE_CODE_OAUTH_TOKEN for headless CI use, allowing inference axis to collapse under subscription. Headline arithmetic: 100 audits/day on Max 20x = $223/mo total vs ~$711/mo on Managed Agents (~70% reduction at this volume). Substrate stays vodka-architecture; opinions live in callers.
4+
C Constraint 5 5 strong Vendor names do not appear in Tier-1 canon URIs Operator-as-validator catch from prior session: vendor naming smell in 'audit-gates-are-managed-agents' was not surfaced by tooled validation. Sibling 'governance-validation-via-agents' has the same shape and is flagged as a follow-up rename in scope of this PR or its successor. Tier-1 commitments name what is required structurally; vendor mappings live in method-level catalogues that explicitly enumerate options.
5+
H Handoff 5 5 strong PR A scope: substrate rename + substrate-options method + journal klappy.dev branch claude/substrate-rename-and-options-8f57. Two file changes: (1) rename canon/constraints/audit-gates-are-managed-agents.md to canon/constraints/audit-gates-are-spawned-agent-sessions.md with content rewrite, (2) new canon/methods/spawned-agent-session-substrate-options.md cataloguing substrate options. Journal entry committed alongside. AMS adoption pointer (PR B scope at klappy/agent-messaging-service) depends on this PR's klappy:// URIs landing first. Once PR A merges, PR B for AMS becomes ready to open.

0 commit comments

Comments
 (0)