diff --git a/CHANGELOG.md b/CHANGELOG.md index 5830811..6e0040f 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -6,6 +6,7 @@ version in [`.claude-plugin/plugin.json`](.claude-plugin/plugin.json); ## [0.1.0] - 2026-10-15 +- Workflows `unattended-backlog`, `climb-a-score` and `safe-to-close` gain `/switch` and `/remind` steps for usage limits (#235). - Skills split into categories (#227): Plan and decide, Build and orchestrate, Sessions and accounts, Review and report, Repo and safety, Kanban board, plus agents and hooks. One list in `site/data/meta.json` (`categories`, was `map_groups`) drives the site home, `/map`, each page's tag and the README "What's here" tables; a test fails if they drift. - skill pages: more spacing, bullets - New `remind` skill (#226): `/remind when 's session|weekly limit resets`, `when is under N%`, `in 2h ...`, `at 3pm ...`, `list`, `cancel `. Reset times come from `switch`'s usage cache (unknown account, no cache or an already-reset window is an error, never a guess). Due reminders show once at the next prompt on any account through `switch`'s hook (a one-line hook without switch); timed ones also get a one-shot desktop notification (Windows scheduled task with a PowerShell toast, `msg` fallback; `at` on macOS/Linux). `reminders.json` sits next to the switch registry (`~/.claude-remind/` without it). diff --git a/find-me-a-skill/CATALOG.md b/find-me-a-skill/CATALOG.md index cba4372..27bbe5c 100644 --- a/find-me-a-skill/CATALOG.md +++ b/find-me-a-skill/CATALOG.md @@ -29,14 +29,14 @@ Install section. | ponytail | skill | sonnet | medium | `/ponytail` | Lazy senior dev mode: the smallest change that fully solves the task, and a reply a busy human understands in one read. | lazy-fix-handoff | | pr-watch | skill | sonnet | high | `/pr-watch` | Watch the user's open PRs to merge — read bot and human review threads, fix valid ones via pair, reply to or resolve the rest with evidence, report CI; every push, reply, and resolve gated on a yes. | pair-to-pr | | progress-report | skill | sonnet | low | `/progress-report` | Write a short progress report of what an orchestrated or long session did since the last report (never overlapping): done, what worked, what didn't, concerns, decisions made for the user, manual steps. | unattended-backlog, brief-a-teammate | -| remind | skill (user only) | haiku | low | `/remind` | Set a reminder for when an account's session or weekly limit resets, when its usage drops under a percentage, or at a time; list or cancel them. | usage-limit | +| remind | skill (user only) | haiku | low | `/remind` | Set a reminder for when an account's session or weekly limit resets, when its usage drops under a percentage, or at a time; list or cancel them. | unattended-backlog, safe-to-close, climb-a-score, usage-limit | | repo-scrub | skill (user only) | opus | high | `/repo-scrub` | Scan a GitHub repo's full history for secrets and large files, scrub what the user picks, rewrite history safely, then flip it private → public. | go-public | | repo-showcase | skill | sonnet | high | `/repo-showcase` | Make a repo public-facing on its working tree — README rewritten around outcome and demo, stale docs pruned, outdated claims fixed, secrets and private names found, GitHub metadata proposed — then hand off to /repo-scrub. | go-public | | score-loop | skill | opus | high | `/score-loop` | Run an unattended experiment loop that hill-climbs one rerunnable numeric score — local grader-mirroring harness, one hypothesis per commit, keep only gains beyond the noise margin, auto-revert the rest, append-only results ledger; stops on budget, deadline or plateau. | climb-a-score | | session-bridge | skill | sonnet | low | `/session-bridge` | Coordinate with another Claude Code session or agent on the same repo — snapshot its state, check for overlap, send it a gated message. | two-sessions | | sitrep | skill | haiku | low | `/sitrep` | Live status of this session in ≤10 lines — what's running, git state, PRs + CI, asked vs done, what needs the user, and a SAFE/NOT SAFE close verdict. | safe-to-close, two-sessions, usage-limit | | spend-gate | skill | sonnet | medium | `/spend-gate` | Gate any billable action (GPU pods/instances, paid APIs or tiers, cloud credits) behind a cost estimate, a live balance check, and an explicit yes for that action. | paid-tests, climb-a-score | -| switch | skill (user only) | haiku | low | `/switch` | Move this conversation to another Claude account (the previous one, or one by name), or add, remove, rename and list accounts. | usage-limit | +| switch | skill (user only) | haiku | low | `/switch` | Move this conversation to another Claude account (the previous one, or one by name), or add, remove, rename and list accounts. | unattended-backlog, climb-a-score, usage-limit | | teammate-brief | skill | sonnet | medium | `/teammate-brief` | Write an onboarding document for a person joining or picking up a project — what it is, setup steps verified by running them, env var names and where to get them, tasks with "done when", owners, don't-touch list. | brief-a-teammate | | wayfinder | skill (user only) | opus | high | `/wayfinder` | Plan an effort too big for one session as a map of decision tickets (GitHub issue + sub-issues) and resolve them one per session until the way is clear, then hand off to divide. | big-effort | | consult-plan | skill | opus | high | `/consult-plan` | Grill a proposed change, deviation, or new idea against a repo's existing plan (issues, board, roadmap, decisions), then record the outcome. | pitch-an-idea | @@ -114,8 +114,11 @@ human-in-the-loop points: nothing continues without you. 6. manager (agent, opus · high): Briefs an implementer, reviews every round inline and reruns the tests itself, for up to 3 rounds. 7. implementer (agent, sonnet · medium): Implements on its branch, commits small, watches the tests pass and ends with session-handoff. 8. verifier (agent, opus · high): Runs update-progress end to end, answers or parks deviations, then runs next. -9. /progress-report (skill, sonnet · low): Every N merges, writes a two-minute report of what happened since the last one, never overlapping. -10. you: Read the report and do the manual steps it lists. +9. you: When the statusline warns near the limit (90%), type /switch between merges; it is user-only, so Claude can't start it. +10. /switch (skill, haiku · low): The session resumes on the other account; running subagents pause (resume each with SendMessage to its id) and background commands end, hence between merges. +11. /remind (skill, haiku · low): /remind when 's session limit resets pops a toast at the reset time and says it once at your next prompt. +12. /progress-report (skill, sonnet · low): Every N merges, writes a two-minute report of what happened since the last one, never overlapping. +13. you: Read the report and do the manual steps it lists. ### pair-to-pr: One issue to a reviewed PR @@ -148,9 +151,10 @@ human-in-the-loop points: nothing continues without you. 1. you: Ask "can I close this". 2. /sitrep (skill, haiku · low): Reports what's running, git state, PRs and CI, and asked vs done in up to 10 lines, with a SAFE or NOT SAFE verdict. -3. you: Say close and name what may be stopped; anything vaguer gets the list again. -4. /handoff-auto (skill, sonnet · medium): sitrep calls it for mid-flight work; it writes the handoff and prints its absolute path as the last line. -5. you: Start the next session from that path. +3. /remind (skill, haiku · low): Optional: /remind when 's weekly limit resets, so you know when to come back. +4. you: Say close and name what may be stopped; anything vaguer gets the list again. +5. /handoff-auto (skill, sonnet · medium): sitrep calls it for mid-flight work; it writes the handoff and prints its absolute path as the last line. +6. you: Start the next session from that path. ### two-sessions: Coordinate with another session @@ -189,8 +193,11 @@ human-in-the-loop points: nothing continues without you. 2. you: Say yes to the setup once: quota, spend cap, deadline; then it runs unattended. 3. /spend-gate (skill, sonnet · medium): Gates each paid submission or compute step with a cost estimate and a live balance check; the spend cap is a ceiling, not approval. 4. you: Say yes to that one paid action. -5. /handoff-auto (skill, sonnet · medium): When the loop's context degrades, writes a handoff pointing at skilleddocs/experiments.md and /score-loop resume. -6. you: Run /score-loop resume in a fresh session; at the deadline freeze, say yes to the final submission. +5. you: When the statusline warns near the limit (90%), type /switch between experiments, not mid-run; it is user-only, so Claude can't start it. +6. /switch (skill, haiku · low): The session resumes on the other account; running subagents pause (resume each with SendMessage to its id) and background commands end, hence between experiments. +7. /remind (skill, haiku · low): /remind when 's session limit resets pops a toast at the reset time and says it once at your next prompt. +8. /handoff-auto (skill, sonnet · medium): When the loop's context degrades, writes a handoff pointing at skilleddocs/experiments.md and /score-loop resume. +9. you: Run /score-loop resume in a fresh session; at the deadline freeze, say yes to the final submission. ### brief-a-teammate: Hand the project to a teammate diff --git a/site/data/workflows.json b/site/data/workflows.json index a6204bc..65975ae 100644 --- a/site/data/workflows.json +++ b/site/data/workflows.json @@ -73,6 +73,9 @@ {"kind": "agent", "name": "manager", "why": "Briefs an implementer, reviews every round inline and reruns the tests itself, for up to 3 rounds."}, {"kind": "agent", "name": "implementer", "why": "Implements on its branch, commits small, watches the tests pass and ends with session-handoff."}, {"kind": "agent", "name": "verifier", "why": "Runs update-progress end to end, answers or parks deviations, then runs next."}, + {"kind": "human", "why": "When the statusline warns near the limit (90%), type /switch between merges; it is user-only, so Claude can't start it."}, + {"kind": "skill", "name": "switch", "why": "The session resumes on the other account; running subagents pause (resume each with SendMessage to its id) and background commands end, hence between merges."}, + {"kind": "skill", "name": "remind", "why": "/remind when 's session limit resets pops a toast at the reset time and says it once at your next prompt."}, {"kind": "skill", "name": "progress-report", "why": "Every N merges, writes a two-minute report of what happened since the last one, never overlapping."}, {"kind": "human", "why": "Read the report and do the manual steps it lists."} ] @@ -119,6 +122,7 @@ "steps": [ {"kind": "human", "why": "Ask \"can I close this\"."}, {"kind": "skill", "name": "sitrep", "why": "Reports what's running, git state, PRs and CI, and asked vs done in up to 10 lines, with a SAFE or NOT SAFE verdict."}, + {"kind": "skill", "name": "remind", "why": "Optional: /remind when 's weekly limit resets, so you know when to come back."}, {"kind": "human", "why": "Say close and name what may be stopped; anything vaguer gets the list again."}, {"kind": "skill", "name": "handoff-auto", "why": "sitrep calls it for mid-flight work; it writes the handoff and prints its absolute path as the last line."}, {"kind": "human", "why": "Start the next session from that path."} @@ -175,6 +179,9 @@ {"kind": "human", "why": "Say yes to the setup once: quota, spend cap, deadline; then it runs unattended."}, {"kind": "skill", "name": "spend-gate", "why": "Gates each paid submission or compute step with a cost estimate and a live balance check; the spend cap is a ceiling, not approval."}, {"kind": "human", "why": "Say yes to that one paid action."}, + {"kind": "human", "why": "When the statusline warns near the limit (90%), type /switch between experiments, not mid-run; it is user-only, so Claude can't start it."}, + {"kind": "skill", "name": "switch", "why": "The session resumes on the other account; running subagents pause (resume each with SendMessage to its id) and background commands end, hence between experiments."}, + {"kind": "skill", "name": "remind", "why": "/remind when 's session limit resets pops a toast at the reset time and says it once at your next prompt."}, {"kind": "skill", "name": "handoff-auto", "why": "When the loop's context degrades, writes a handoff pointing at skilleddocs/experiments.md and /score-loop resume."}, {"kind": "human", "why": "Run /score-loop resume in a fresh session; at the deadline freeze, say yes to the final submission."} ]