| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
I had the goal backwards in the lead. Craig's framing is output quality first, with token reduction real but second, and where they conflict quality wins. Anthropic's 80% was a finding rather than a target, and aiming at it directly optimizes the thing we don't care about. P4 falls to that and is dropped: lowering effort buys tokens by spending quality. P5 rises for the same reason, since positive framing over prohibition is what actually targets guardrails working against output.
My token figures were low by about 45%. I converted word counts at a guessed ratio when /context reports real per-file numbers, so I used an estimate because it was easier to compute from inside the repo. The live ratio is 2.28 tokens per word. commits.md is 12,800 tokens rather than the 7,000 I estimated, and claude-rules was near 57,800 per session before today.
I also conflated two loading paths. Memory files come in through the harness at session start. protocols.org and the workflows are read by startup and land in Messages, so they never appear under Memory files at all. They shrink by editing the workflow, not by scoping a rule, and treating them as one surface made protocols.org look like it competed with commits.md for the same fix.
Path-scoping is confirmed working at user level, and the de-duplication holds. That splits the remaining work: anything scopable to a file type or directory ships with no trial, because a glob match is deterministic. Only the semantic rules need the skills route and the stop conditions, and commits.md is the one that matters there at 12,800 tokens of publish machinery that only applies when a commit is in play.
I recorded one caution the confirmation doesn't cover. Path-scoping fires when a matching file is read, so creating a new org file from scratch never triggers todo-format.md. Edits are safe because Edit requires a prior read.
I also wrote up the day's two guard failures, both mine. wrap-org-table reflowed a table into a worse shape and lint-org certified it clean, and the teardown hook consumed a stale sentinel and killed a live work session. The plan leans on mechanical detectors because they have no stake, which is still right, but a detector can be confidently wrong and a green check from a guard that never looked is indistinguishable from one that did.
|
|
|
I wrote three working documents while reviewing the Claude 5 context-engineering post, the Opus 5 prompting guide, and the Fable field guide against what this repo ships downstream.
In proposals.org I measured the always-loaded surface at 32,123 words and ranked six changes, plus two places where a post contradicts something we arrived at deliberately. I also reread the rules as your prompts rather than as agent context, which is where the sharper finding is: 41 execution and hygiene workflows against 6 discovery and design ones, on a system whose bottleneck has moved.
In rollout.org I phased the work and named the seven decisions that gate it. My lead finding is that the harness system prompt already carries most of what the Opus 5 guide recommends adding, so the posts' value here is subtractive. Applying them additively would make the duplication worse.
In metrics.org I split the posts into separable claims and marked which are testable here and which are judgment calls, rather than inventing a metric for the ones an eval harness would be needed to settle. I also stated the pilot's stop conditions in advance, including a zero-tolerance threshold on the load-bearing rules, so I can't renegotiate that threshold later under pressure to make a phase succeed.
|
