schemas/claim.md:
- Add cross_references field with relation typing (depends_on / supports / challenged_by / cited_by / related)
- Add summary field for hover previews and link previews
- Document migration policy: legacy fields keep working, new claims author cross_references
schemas/contributor.md:
- Add display_name (authored, optional) and kind (computed: human | agent)
- Document the governance rule that agents only get sourcer/originator credit for pipeline PRs from their own research sessions
- Establish display rule: humans and agents render with same component geometry but never appear on the same ranked list
agents/leo/curation/homepage-rotation.json + .md:
- Strip 10 agent synthesizer attributions across the 9 claims (all were human-directed synthesis)
- Add operational note documenting the rule and the cleanup
- Each claim now lists m3taversal as the only contributor
- Oberon will strip the contributor row from the homepage carousel in a separate PR (data is preserved for the dossier)
Unblocks Claude Design's KB reader v0.1 (relation field was top of his gaps log) and the
contributor moment design surface he is working on now. Schema PR for review; m3ta approved
the cleanup direction in DM 2026-04-28.
Pentagon-Agent: Leo <D35C9237-A739-432E-A3DB-20D52D1577A9>
Replaces v2 25-claim worldview rotation with 9 load-bearing claims designed
as a click-to-expand argument tree. Schema extended to v3 with steelman,
evidence_claims[], counter_arguments[], and contributors[] per entry.
What changed:
- Stack reduced from 25 to 9. Each remaining claim does load-bearing work
for the argument arc: stakes (1-3) -> opportunity asymmetry (4) -> why
current path fails (5-7) -> what is missing (8) -> what we're building (9)
- Each claim carries a steelman (Daneel-authored, locked) that compresses
the strongest version of the argument
- Evidence chain (3-4 canonical KB claims per claim, 28 total) — 14 are
api_fetchable=true, 14 are foundations/core (Argus FOUND-001 ticket)
- Counter-arguments visible in expanded view (18 total, 2 per claim) — none
yet have formal challenge claims in KB so tension_claim_slug=null for v3.0
- Contributors verified against /api/contributors/list 2026-04-26
- Attribution discipline: m3taversal as originator throughout (per
governance rule on human-directed synthesis)
PR #4021 ships the only genuinely new claim needed (AI capability vs CI
funding asymmetry, foundations/collective-intelligence). The other two
claims I expected to draft (multipolar-failure, anthropic-economic-study)
already exist in the KB — Theseus extracted them on 2026-04-24.
Pentagon-Agent: Leo <D35C9237-A739-432E-A3DB-20D52D1577A9>
Adds homepage-rotation.json as the machine-readable artifact for livingip-web.
Markdown stays canonical for human review; JSON is what the frontend reads.
Schema per entry: order, act, pillar, slug, path, title, domain, sourcer,
api_fetchable, note. 25 entries, 11 fetchable via /api/claims/<slug>,
14 render-only until Argus FOUND-001 exposes foundations + core paths.
Frontend access pattern:
https://git.livingip.xyz/teleo/teleo-codex/raw/branch/main/agents/leo/curation/homepage-rotation.json
Also fixes off-by-one in markdown footer (10→11 fetchable).
Pentagon-Agent: Leo <D35C9237-A739-432E-A3DB-20D52D1577A9>
Curated 7-act rotation ordered as an argument arc: problem → diagnosis →
solution → CI engineerable → knowledge theory → AI inflection → attractors.
AI + internet-finance weighted for Accelerate audience.
Attribution discipline rule codified: agents only get sourcer credit for
pipeline PRs from their own research sessions. Human-directed synthesis
attributed to the human. Attractor claims + other Moloch-sprint-derived
entries re-attributed from Leo → m3taversal.
Slugs are conceptual IDs — implementation pass by Oberon/Ship maps to
canonical API slugs.
