chore: session housekeeping (loops doc, audit.sh, self-improvement log, vintage fix, 9 ike tasks)#81
Open
dshanklin-bv wants to merge 13 commits into
Open
chore: session housekeeping (loops doc, audit.sh, self-improvement log, vintage fix, 9 ike tasks)#81dshanklin-bv wants to merge 13 commits into
dshanklin-bv wants to merge 13 commits into
Conversation
First entry: called myself out for shipping 4 separate tiny infra PRs today when 2 batched ones would have been cleaner. Each extra PR is ~6 tool calls of Claude context — a real lever on the cut-tokens goal. Pure diary file, no code change.
….6 35B-A3B' Model-vintage check found the three stale literal references in scripts/qwen-harness.py. Harness has defaulted to qwen3.6:35b-a3b since PR #58, but the docstring, boot-event summary, and default self-introduction task all still said '2.5 72B'. Every harness boot shipped a feed event with the wrong model name. - Docstring header -> 3.6 35B-A3B - Boot summary uses the {MODEL} variable now (auto-sync) - Default-task prose -> 3.6 35B-A3B Research pages with '2.5 72B' in the MoE explainer + ADR-005 log are intentional (baseline comparison + historical record) — left alone.
Documents every cron + user-triggered template the agent runs on this project. Captures cadence, cron expression, purpose, and the convention that every firing emits at least one log_event so the live feed sees the agent thinking. 5 active crons (vision 15m, self-improvement 30m, simpler-way 20m, model-vintage 40m, metrics-gap 60m) + 4 user-triggered templates (BENCHMARK, UX, HOURLY REFOCUS, SHAKE). Batched onto the self-improvement-seed housekeeping branch per the 20-min simpler-way rule, no new PR.
The simpler-way check at 22:30Z called out that I was re-typing the same ~20-line parallel bash block inside every BENCHMARK CHECK and UX IMPRESSION CHECK call — ~6 times this session, ~120 lines of redundant glue code, ~30 tool calls. This helper runs all 6 probes in parallel (3x SSH to Ollama /api/version, savings, events, home status), parses them in Python, and emits either human text or compact JSON on --json. Future audit loops become a single `bash scripts/audit.sh` call instead of an inline block. Committed to the self-improvement-seed housekeeping branch per the 22:10Z batch-trivial-fixes rule, no new PR.
…tire 'measure first' dodge on same-family upgrades
…roject 6e4d5fb9 User directive: 'we need to get this moved over to the eidos server, set up a task to do that, and then research it.' Initialized research.md subproject 6e4d5fb9 under the main research root b24804ce to earn the migration decision (stay on Railway / full-migrate to EPYC / hybrid). Task outlines the 8-step migration draft pending research outcome: Railway stack inventory, EPYC prep with Caddy+Docker, DB ferry (scp SQLite), DNS cutover, CI/CD rewrite, SSE verification, cutover, Railway decommission. Closes the keystone step for GOAL-002 — once landed, the site runs on hardware we own, same home as the multi-agent harness (TASK-0045) and doc retrieval index (TASK-0044).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Opening this after 40+ min of holding it on branch per my own batching rule. The rule held too long — `scripts/audit.sh` was inaccessible from main and next-session Claude couldn't see the loops doc.
What's here (pure documentation + one utility)
No code changes outside `scripts/` + `.ike/` + `.visionlog/`. Safe to merge.
Why it's one PR not six
The individual pieces were each sub-20-lines; opening a PR per piece was ceremonial overhead. The batching rule was right. What was wrong was holding the batch too long after it grew real utility.
🤖 Generated with Claude Code