← AICEP home · Runbook index

AICEP set — test status

Basis: public AICEP 0.1 Public Draft (RFC) at aicep.ai/protocol/, /import/ and /use-cases/. Round 1 paper tests (library text as-is): #171, #262, #263, #396, #431; findings are tracked for revision in the prompt library. Second round added: #481 (pass, no change). Already ≥2 rounds ending in pass (no retest): #357 (3 rounds), #458, #482, #483, #486, #496, #500, #514, #324 (3), and the new 'Peer-agent relay check' and 'Agent errand guardrail card' drafts (R1 fail, R2 14/14 ×4). All paper walk-throughs, not live runs.

New prompts G1–G10: 2 rounds each (R1 fail → guard block → R2 14/14). G5, G6, G8, G9, G10 were changed for reviewer must-fixes M1–M5 and re-tested in round 3: all 14/14 (G8 also passed replay, unlisted-file, envelope-mismatch and gap cases).

v2.1: G5 and G8 edited and re-tested (round 4, 14/14) plus a live smoke run that passed. Reference suite: 188 automated tests, all passing (v2.5.1), also passing after supersedes moved to meta.age. G11: 2 rounds, R2 14/14. G8 round 5 after reviewer must-fixes: 14/14 on A–H. Reference suite: 188 automated tests, all passing (v2.5.1).

v2.2: 188 automated tests, all passing (v2.5.1). G8 round 6: 14/14 on A–I, including the forged paste. v2.3: G5, G7, G8, G9, G10 and G11 re-tested on paper (A–E, 14/14). Code suite: 188 automated tests, all passing (v2.5.1).