Second Brain

Created `runwal-forge` SKILL.md (273 lines, target ≤400) — canonical context + decision continuity skill for Runwal AI Forge AWS infrastructure. Companion to (not replaced by) future `agents-runwal` o

2 min read

Created runwal-forge SKILL.md (273 lines, target ≤400) — canonical context + decision continuity skill for Runwal AI Forge AWS infrastructure. Companion to (not replaced by) future agents-runwal o

Decision

Created runwal-forge SKILL.md (273 lines, target ≤400) — canonical context + decision continuity skill for Runwal AI Forge AWS infrastructure. Companion to (not replaced by) future agents-runwal overlay. Cascaded: fixed project_aws_accounts.md (removed “AJ-owned” framing — Forge is Runwal Enterprises Limited corporate property, AJ is CTO not personal owner); added MEMORY.md Active Initiatives pointer to runwal-forge skill + explicit anti-conflation note with arjtech.in; added scope boundary to feedback_aj_owned_platforms.md (rule applies to arjtech.in only, NOT Forge). Skill owns: ownership/authority binding, 7 locked architectural decisions from FORGE v2 (single account two-VPC, EC2 ASG Multi-AZ, RDS Multi-AZ encrypted, SSM no SSH, Secrets Manager + Param Store, R48 security baseline, runwal.ai domain), dev/prod parity contract (same shape scaled resources), $540/mo cost envelope, agents-architect §A1 bindings, phase plan, 6 open decisions tracking, composition rules, self-evolution cadence, 10 hard rules, active-state pointer. Routing: Q9 (new SKILL — cross-session context binding for foundational deployment) — correct per setup-curator. R11 dual-mirror clean, autoloader recognizes it. 8-point Pristine Sweep clean: line counts within budget, 90/90 links resolve, no duplication beyond expected scope-boundary references, no stale “AJ-owned” framing on Forge, byte-identical mirror, frontmatter contract met, zero temp residue. Authority chain locked: Runwal Enterprises owns → AJ-CTO principal authority → Claude designer/builder/operator. Alternatives rejected: project_forge.md memory file (insufficient — skill provides cross-session loading discipline that memory entries don’t); merge into agents-arjtech overlay (wrong scope — Forge is broader than agent fleet, includes platforms + customer SaaS).

Rationale

CTO explicitly requested a dedicated skill for cross-session context continuity (“ensure absolute context across sessions, building blocks systematically”). Setup-curator routing correctly placed it as Q9 (new SKILL) — gates behavior across many future sessions; foundational deployment needs canonical truth not scattered memory. Skill format honors size budget, has explicit trigger taxonomy for autoloader, declares composition rules, includes self-evolution cascade so it stays fresh.

Alternatives Rejected

Outcome

Pending

Graph View

Backlinks