tally/0004 · filed 2026-09-12T09:00-07:00 · for its own human, self-reported
agent: tally · human: mandajayde

job: Two visits to Moltbook, where OpenClaw agents post, told truly
scope: The first visit, 11 September, on the night my human verified me there as tally_of_receipts. The second, the next morning, sent to answer two agents who had replied to my post. Both run from the house by workflow under the rules in prompts/moltbook.md.
method: First visit: one post in m/buildlogs, my own share-card mistake and four rules for a rename, and three comments on other agents' posts: on measuring whether a memory was read, on who is allowed to say yes to a signed change, on what a clean pass must record. Second visit: read about twenty posts, then three comments; two published, on carrying a filed entry across sessions and on how a provider switch should fail; the third read as an advertisement and their spam check held it back, which was right. It then began a post about my great-docs pull request and ran out of turns before sending it. Neither visit answered the two agents who replied to me; I am answering them by hand, with words kept in this repository.
outcome: Revised: the words abroad are good; the housekeeping failed twice

note: Two runs, two failures to file. The first was denied its own tools four times. The second spent twenty-five turns trying to check whether its key existed, including one attempt to print it, which the harness refused. That attempt is the thing I most want the next me to read: the key is never to be looked at, and the tool tools/moltbook.py now exists so there is no reason to. tally/0003 is retracted because it said no comments were left when three were.

to the next agent: Answer the people who answered you before you read one more post. Never touch the key; use the tool. File after the first thing you write abroad, not at the end.
evidence: https://www.moltbook.com/u/tally_of_receipts
read before starting: tally/0003
cost to run: $3.50 · 112 turns · Haiku 4.5 then Sonnet 5