feat(b-0488): KSK (Kinetic Safeguard Kernel) persona map

[AceHack]

Summary

Per-product persona doc for KSK (Kinetic Safeguard Kernel) using the B-0485 template. Closes B-0488 which unblocks B-0492/B-0493 skill-mapping work.

Personas defined

Primary (3):

• ksk-agent-developer — engineers integrating "am I allowed to do this?" checks into AI agents
• ksk-robotics-designer — consent-first robotics / actuator system designers (NVIDIA Thor Homeland-Security clearance lineage)
• ksk-security-engineer — engineers building KSK itself in Lucent-Financial-Group/lucent-ksk

Secondary (1): ksk-clearance-deployer — Homeland-Security / clearance-aware deployers

Adjacent (1): ksk-compliance-auditor — SOC 2 / HIPAA / ISO 27001 auditors consuming KSK signed receipts

Refused (2 — HARD LIMITS):

• ksk-refused-weapons-control — autonomous-weapons / kill-chain designers using KSK as a "consent UI" wrapper. Per methodology-hard-limits.md HARD LIMITS deps: Bump FsUnit.xUnit from 7.1.0 to 7.1.1 #1 + Round 27 — plugin API + governance split + memory-in-repo #3: laundered consent + violates consent-first design intent (PR docs(memory): KSK origin from Aaron+Amara consent-first design; NVIDIA Thor Homeland-Security cleared because actuators; factory at forefront of consent-first AI design #2892).
• ksk-refused-apt-operator — nation-state APT operators using KSK as a privilege oracle (receipt-replay, authorization enumeration, "stealth mode" feature requests). Per mechanical-authorization-check.md: not in the authorization-source list.

Why the refused-persona list is load-bearing

KSK's terminal purpose is human-in-the-loop refusal of impactful AI actions. A weapons-system integration would launder the appearance of consent (the receipts say "authorized" but the operator never had meaningful refusal). An APT integration would use KSK's transparency against its own purpose (mapping authority topology). Both are structural adversaries, not off-target customers.

Acceptance criteria

• Template from B-0485 applied
• Primary personas fully documented (3, exceeds the "grey-hat / ethical researcher" minimum — folded that framing into ksk-security-engineer)
• At least 2 refused personas with explicit HARD LIMITS rationale (R1 + R2)
• Output doc at docs/personas/ksk-personas.md
• B-0488 status: open → in-progress (closes on merge)
• B-0492 composes_with: pointer backfilled (deferred to a follow-up; B-0492 row may not exist yet — needs scoping)

Test plan

• markdownlint-cli2 clean on both files
• Composite-guard used for the commit: test "$(git branch --show-current)" = "<expected>" && git commit
• gh pr create --head explicit ref used
• CI clears
• Auto-merge

🤖 Generated with Claude Code

Co-Authored-By: Claude noreply@anthropic.com

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[Copilot]

Pull request overview

Adds the KSK (Kinetic Safeguard Kernel) per-product persona map using the B-0485 template, and updates the associated backlog row to reflect active work on B-0488.

Changes:

• Add docs/personas/ksk-personas.md defining primary/secondary/adjacent/refused KSK personas (with HARD LIMITS rationale).
• Update docs/backlog/P1/B-0488-ksk-persona-map-2026-05-14.md status from open to in-progress.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
docs/personas/ksk-personas.md	New KSK persona-map document (template-structured), including refused personas + detection/response notes.
docs/backlog/P1/B-0488-ksk-persona-map-2026-05-14.md	Backlog row status update for B-0488.

Comments suppressed due to low confidence (1)

docs/personas/ksk-personas.md:375

• P1 (name-attribution policy): This section repeats direct named attribution (e.g., “Aaron + Amara’s …”) inside a current-state docs/personas/** doc. Per docs/AGENT-BEST-PRACTICES.md “No name attribution in code, docs, or skills” (around lines 671–760), prefer role-refs here and keep named provenance on the allowlisted history surfaces (e.g., link to the relevant memory/**/docs/research/** artifact that preserves the attribution).

2. Compromise the protective lineage (KSK's reputation is its value;
   weapons-system integration destroys it)
3. Violate the consent-first design intent (PR #2892 — Aaron + Amara's
   consent-first design origin)