Apply Me CRM evaluator-pass punch list + fix claim-checker subagent bug

[hnshah]

The personal-crm-founders run (first real agentic fully-logged run after the PR #10 enforcement tightenings) surfaced a 4-item punch list in its evaluator-pass.md. This PR applies all four. It also fixes a bug in the pagekit-claim-checker subagent that the run exposed.

This is the run-to-repo-improvement loop working exactly as designed: run produces evaluator-pass produces specific repo changes in the very next PR.

Punch list (from runs/personal-crm-founders/evaluator-pass.md)

1. First-page-decision template: falsification prompt

templates/first-page-decision-template.md — added an "If this is a hypothesis: what would falsify it?" sub-field under Confidence basis. Stops hypothesis-level decisions from being silently promoted to conclusions. Required when confidence is "hypothesis"; optional when "data" or "signal."

2. Evaluation scaffold: Source quality field

scripts/new-run.sh evaluation scaffold now includes a ## Source quality section at the top: Real / Training fiction / Mixed. Surfaces provenance prominently rather than burying it in sources/01-source-capture.md. A reader should immediately know whether the run was built on real or invented material.

3. Claim-check: distinguish remove-vs-verify

Claim-check previously collapsed "cut this line" into a single correction category. The audit should preserve three dispositions:

• rewrite
• remove (wrong) — disqualified; do not restore
• remove pending verification — potentially restorable if source confirms

Updated across three surfaces: prompts/07-claim-check.md, .claude/agents/pagekit-claim-checker.md, templates/claim-check-template.md.

4. Evaluation scaffold: weak-section to source-gap mapping

scripts/new-run.sh evaluation scaffold now requires that every section flagged as weak name the specific source material that would fix it. A weak section without a source gap named is a weak section shipping by choice, not by constraint.

Bug fix: claim-checker subagent corrupted the corrected draft

The Me CRM run's working-log reports:

"subagent left inline *[Rewritten: ...]* annotations in body copy; these were stripped manually; 2 new em-dashes introduced by rewrites (lines 41 and 65) were also fixed"

The pagekit-claim-checker subagent was introducing new slop while correcting old slop. Fixed with explicit hard rules in .claude/agents/pagekit-claim-checker.md:

• No inline annotation markers in body copy. Provenance goes in the audit, not the corrected draft.
• No new em-dashes introduced by rewrites (per frameworks/anti-slop.md).
• Self-scan rewrites for flagged patterns before saving the corrected draft.

Mirrored in prompts/07-claim-check.md so chat users get the same enforcement.

Verified

• bash scripts/doctor.sh → PASS
• bash scripts/slop-check.sh → exit 0 clean
• bash scripts/run-check.sh runs/vegan-dog-food-verdel → PUBLISHABLE
• bash scripts/run-check.sh runs/personal-crm-founders → PUBLISHABLE
• Fresh new-run.sh _test scaffold shows ## Source quality and ## Weak section to source-gap mapping sections in evaluation.md
• templates/first-page-decision-template.md shows the falsification prompt
• .claude/agents/pagekit-claim-checker.md contains the "Hard rules for the corrected draft" block