16642bb security: correct the threshold's justification in guard and README
8f40050 Merge pull request #235 from evkir/w6/technical-tails
7633bcb fix(exploit): the scope guard covers every path that touches the target
2e24c9f fix(intel): TLS problems reach the report instead of the knowledge base
40a9ed9 fix(orchestrator): a crossed budget or a broken air gap ends the run
c64eb8e feat(scope): refuse an unscoped exploit phase under --strict-scope
9b66238 Merge pull request #234 from evkir/w4/honest-readme
bf93cbb docs: link the documents README was already citing, and gate the links
ce2789e docs: describe the trust boundary that exists, and gate the count
3a4855e docs: publish the agent score, and gate it against the card
e44a54d chore: release version 1.6.0 and gate the badge that drifted
b00be13 Merge pull request #233 from evkir/w2/honest-provenance
221553c fix(cli): one guard for the two ways a run can say nothing about a model
810f493 Merge pull request #232 from evkir/w2/honest-provenance
8926685 docs(bench): publish the contamination account and the numbers after it
318a675 chore(bench): regenerate both scorecards against live targets
634810d test(architecture): pin what the scorecard may read out of a task
0142601 fix(bench): the scorecard stops naming a model nobody reached
a378b97 feat(bench): the agent engine records whether a model took part
9029fb7 fix(core): a path that takes no model is not blamed on a missing key
496e814 fix(bench): the run manifest stops namin