Track and surface sub-agent dispatch calibration ratios. Extract from JSONL transcript or read existing ledger. Per-agent clean/blocked/errored/partial trends.
中文简介正在准备中。当前显示原文说明。
Read-only attention reconciliation. Computes actual seven-day project attention from daily notes against explicit Tier-1 intended shares.
中文简介正在准备中。当前显示原文说明。
实测评分 未实测
查看安装方法安装前可以查看交给 AI 代理的指令内容。
This measurement leg reports attention drift. It never edits priorities, daily notes, goals, or runtime state.
Run:
node daemons/nightly-reconcile.mjs --root <aigent-root> --as-of <fire-date>
Use literal output and exit status as the checkpoint evidence.
Read vault/memory/ACTIVE_PRIORITIES.md and locate explicit Tier-1 projects
with intended percentages. Accepted forms:
| Tier | Project | Intended share |
| 1 | [[Project Name]] | 40% |
or:
Tier 1 - [[Project Name]] - intended 40%
Shares are integers from 1 through 100 and the Tier-1 total must be at most 100. Names or aliases must resolve to a project note or an explicit canonical name in the priorities file.
No intended shares returns nonzero:
RECONCILE FAIL reason=intended-share-input-absent projects=0
This is a configuration failure, never permission to invent shares.
vault/daily/YYYY-MM-DD.md notes inside the exact calendar
window as_of-6d .. as_of. A missing day contributes no mentions.actual = project points / all project points * 100, rounded to one
decimal.A zero denominator returns
RECONCILE FAIL reason=no-measurable-project-attention.
actual < 0.40 * intended.Output raw points, actual share, intended share, thresholds, drift decisions, and source paths so the arithmetic is reproducible:
RECONCILE PASS window=YYYY-MM-DD..YYYY-MM-DD notes=<N>
project=<name> points=<N> actual=<N.N>% intended=<N>% threshold=<N.N>% drift=<yes|no>
non_tier_1 points=<N> actual=<N.N>% drift=<yes|no>
Missing notes, shares, measurable attention, or parseable dates are not green.
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Track and surface sub-agent dispatch calibration ratios. Extract from JSONL transcript or read existing ledger. Per-agent clean/blocked/errored/partial trends.
中文简介正在准备中。当前显示原文说明。
Compose the AIgent's body state by reading existing signals. Lazy compute. Never polled. Reports the five pressures + recommended reflex.
中文简介正在准备中。当前显示原文说明。
Generate a structured delegation brief for any task
中文简介正在准备中。当前显示原文说明。
Quarterly skill that walks the skill-index against the actual SKILL.md files on disk, reports drift (skills indexed that no longer exist, skills on disk that aren't enrolled, skills whose descriptions have changed since enrollment). Hardens the auto-enrollment pipeline against silent rot.
中文简介正在准备中。当前显示原文说明。
When a [CADDY] hint fires and the principal is confused why, /caddy-explain walks the scoring transparently — which trigger matched, what points it earned, what the threshold was, what other skills were close. Pairs with the legibility thesis. Caddy's decisions should be inspectable on demand.
中文简介正在准备中。当前显示原文说明。
Enroll a skill into Caddy's index so it can be surfaced when the right prompt arrives. Use when a new skill is created, copied in, or when a skill needs updated triggers. Usage — "/caddy-enroll <path to SKILL.md>".
中文简介正在准备中。当前显示原文说明。