[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78H4G5154VCXVVWFTCC0NEJ":3,"skill-same:01M78H4G5154VCXVVWFTCC0NEJ":90},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":29,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M78H4G5154VCXVVWFTCC0NEJ","open-items",null,"Maintain the repo's open-items workspace under `.agents\u002Fopen-items\u002F` by promoting capture entries, regenerating the open-item index, moving item state, appending notes, summarizing the queue, and using the dev CLI for bulk reads and index regeneration.","aptlyundecided\u002Fbagelwerk","aptlyundecided","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1045064?v=4","01M78H37K1N5D4YHY63NYPBZ89","Alex","01M78H37M6X2X56CJTYZ9DARJS",0,"MPL-2.0","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: open-items\ndescription: Maintain the repo's open-items workspace under `.agents\u002Fopen-items\u002F` by promoting capture entries, regenerating the open-item index, moving item state, appending notes, summarizing the queue, and using the dev CLI for bulk reads and index regeneration.\n---\n\n# Open Items\n\nMaintain the repo's lightweight open-items system.\n\nThe open-items policy lives with this skill, not in a separate repo-root policy file.\n\nV1 uses:\n- capture inbox: `OPEN_ITEMS_CAPTURE.md`\n- workspace root: `.agents\u002Fopen-items\u002F`\n- item files: `.agents\u002Fopen-items\u002Fitems\u002FOI-####.md`\n- derived index: `.agents\u002Fopen-items\u002FINDEX.md`\n- dev CLI (bulk list, validate, regen index, capture snapshot): `npm run open-items -- …` — see § **CLI tooling**\n\nPer-item files are the source of truth. `INDEX.md` is derived.\n\nAt repo root, keep only the capture inbox file: `OPEN_ITEMS_CAPTURE.md`.\n\n## Mandatory first reads\n\nThis skill directory is the policy source of truth for the open-items system.\n\nRead these before making changes:\n- `OPEN_ITEMS_CAPTURE.md`\n- `.agents\u002Fopen-items\u002FINDEX.md` if it exists\n- for **read-only “whole queue”** questions, prefer `npm run open-items -- list` or `list --json` (stdout) before opening every item file — § **CLI tooling**\n- any target item file being changed under `.agents\u002Fopen-items\u002Fitems\u002F`\n- supporting docs in this skill directory when the operation depends on format or mutation rules\n\nSupporting docs in this directory:\n- `ideas.md` — roadmap, CLI\u002Fscript ideas, multi-agent orchestration notes (non-normative)\n- `CAPTURE-FORMAT.md`\n- `ITEM-FORMAT.md`\n- `INDEX-FORMAT.md`\n- `HANDOFF-FORMAT.md`\n- `OPERATIONS.md`\n\n## CLI tooling (repo)\n\nTypeScript CLI: `.pi\u002Fskills\u002Fopen-items\u002Fcli\u002FopenItemsCli.ts` (with `openItemsLib.ts`). **Edit only under `.pi\u002Fskills\u002F`**; run `npm run skills:sync -- --skill open-items` so `.claude\u002F`, `.cursor\u002F`, and `.codex\u002F` copies stay identical. From repo root: `npm run open-items -- \u003Ccommand>` (see root `package.json`). Without a script, from repo root: `npx tsx .pi\u002Fskills\u002Fopen-items\u002Fcli\u002FopenItemsCli.ts \u003Ccommand>`.\n\n| Command | Output |\n| --- | --- |\n| `list` | **stdout:** markdown table of **open** items (id, state, title, summary clip, path). |\n| `list --json` | **stdout:** JSON array of open items (full summary, paths). |\n| `list --all` | Same as `list`, but includes `done` and `archived`. |\n| `validate` | **stderr:** errors and warnings; **exit 1** if any **error** (malformed item, index state mismatch, bad counter). |\n| `index` | Rewrites `.agents\u002Fopen-items\u002FINDEX.md` from item files and advances counter to max id + 1. |\n| `init` | Creates missing `OPEN_ITEMS_CAPTURE.md`, `.agents\u002Fopen-items\u002Fitems\u002F`, and `.agents\u002Fopen-items\u002FINDEX.md` in the selected scope. |\n| `capture` | **stdout:** pending `#` H1 capture titles below the capture boundary (with line numbers). |\n| `where` | **stdout:** resolved repo root, selected scope root, items dir, index path, and capture path. |\n\nOptional for every command: `--root \u003Cabs-or-rel-path>` to point at another repo checkout.\n\nProject-scoped queues: use `--project \u003Cname-or-path>` to target a Flow\u002Fproject-local open-items queue without changing the repo root. Bare names resolve to `flow-library\u002F\u003Cname>`; path-like values resolve relative to `--root`; targets outside `--root` are refused to avoid cross-contamination. Examples:\n\n```powershell\nnpm run open-items -- init --project crossfit\nnpm run open-items -- list --project crossfit\nnpm run open-items -- validate --project flow-library\u002Fcrossfit\nnpm run open-items -- where --project crossfit\n```\n\nA project-scoped queue uses the same storage contract under the selected scope:\n\n```text\nflow-library\u002F\u003Cproject>\u002F\n  OPEN_ITEMS_CAPTURE.md\n  .agents\u002Fopen-items\u002FINDEX.md\n  .agents\u002Fopen-items\u002Fitems\u002FOI-####.md\n```\n\nDefault commands without `--project` continue to use the repo-level queue under repo-root `.agents\u002Fopen-items\u002F`.\n\n### When agents should run the CLI (prefer script stdout over token-heavy file walks)\n\n- **Operator asks for the full queue, a table, or “what is open”** — run `list` or `list --json` and base the answer on script output instead of re-opening every `OI-####.md` manually.\n- **Before or after bulk edits to many items** — run `validate` so counter, index bullets, and item shapes are consistent.\n- **After changing titles or states across several items** — run `index` to regenerate `INDEX.md` deterministically (same rules as manual regen in this skill).\n- **Before promoting capture** — run `capture` to list pending H1 sections explicitly.\n\nStill **open and edit** the specific `OI-####.md` (and `HANDOFF-FORMAT.md`) when recording handoffs, append-only notes, resume, or any prose mutation—the CLI does not replace reading the file you are changing.\n\n**Execution todos:** use `npm run todos -- list` \u002F `list --json` \u002F `validate` for per-item JSON sidecars (see `todo-contract`); `open-items validate` surfaces cross-surface drift with those files.\n\n## Natural-language commands\n\nExamples:\n- \"promote all capture items\"\n- \"refresh the open-items index\"\n- \"move OI-0003 to blocked\"\n- \"append this note to OI-0004\"\n- \"add a handoff to OI-0010\"\n- \"resume OI-0010\"\n- \"summarize the open-items queue\"\n\nInfer the requested action from natural language.\n\n## Storage contract\n\n### Workspace shape\n\n```text\n.agents\u002F\n  open-items\u002F\n    INDEX.md\n    items\u002F\n      OI-0001.md\n      OI-0002.md\n```\n\nCreate missing directories\u002Ffiles as needed.\n\n### Allowed states\n\n- `new`\n- `triaged`\n- `ready`\n- `in_progress`\n- `blocked`\n- `done`\n- `archived`\n\n### Per-item file contract\n\nUse this exact section order:\n\n```md\n# {title}\n\n## id\nOI-0001\n\n## state\nnew\n\n## summary\n{first paragraph or compact summary}\n\n## notes\u002Fdiscoveries\n{remaining notes}\n```\n\n### Index contract\n\n`INDEX.md` includes only open items. Exclude items whose state is:\n- `done`\n- `archived`\n\nRegenerate the index from item files sorted by id ascending.\n\nWrite bullets in this format:\n\n```md\n- OI-0001 — {title} — state: {state} — .\u002Fitems\u002FOI-0001.md\n```\n\nAt the bottom keep the counter block:\n\n```md\n## Counter (next OI id)\nOI-0002\n```\n\nStore the full next id string in the counter.\n\n### Index title policy\n\nAim for titles with 5 words or fewer.\n\nDo not fail if a title is longer.\nFor index rendering only:\n- if the title has more than 5 words, render the first 5 words plus `…`\n\nKeep the full title in the item file.\n\n## Capture inbox contract\n\n`OPEN_ITEMS_CAPTURE.md` is a staging inbox, not an archive.\n\nUse this literal boundary:\n\n```md\nBegin Items Capture\n---\n```\n\nDo not remove or rewrite content above that boundary.\n\nBelow the boundary, each `# H1` section is one capture entry:\n- the `# ...` line becomes the item title\n- the section continues until the next `# ` H1 or end-of-file\n\n### Promotion rules\n\nWhen asked to promote capture items:\n1. process all entries in file order\n2. allocate sequential ids using the `INDEX.md` counter (or bootstrap from `OI-0001`)\n3. create one item file per entry with initial state `new`\n4. set `summary` to the first paragraph after the H1 title\n5. place the remaining body in `notes\u002Fdiscoveries`\n6. remove the processed H1 sections from the capture file\n7. regenerate `INDEX.md`\n8. advance the counter to the next available `OI-####`\n\n## Mutation rules\n\n### Move state\n\nWhen asked to move an item:\n- update only the `## state` section in the item file\n- regenerate `INDEX.md`\n- if state becomes `done` or `archived`, remove it from the open-items index\n- if state becomes `done` or `archived`, delete `.agents\u002Fopen-items\u002Fitems\u002FOI-####.todo.json` if it exists, and delete legacy `.agents\u002Fopen-items\u002Fitems\u002FOI-####.todo.md` if it exists (see `OPERATIONS.md` — Cleanup and todo sidecars; follow `todo-contract` if recreating later)\n\n### Append notes\n\nWhen asked to append notes:\n- append to `## notes\u002Fdiscoveries`\n- preserve existing content\n- regenerate `INDEX.md` only if the title\u002Fstate changed elsewhere\n\n### Record handoff\n\nWhen asked to add a handoff:\n- read `HANDOFF-FORMAT.md`\n- append a canonical handoff block under `## notes\u002Fdiscoveries`\n- preserve existing content and section order\n- do not change item state unless explicitly asked\n\n### Resume from handoff\n\nWhen asked to resume an item:\n- read the latest handoff block if one exists\n- summarize current state, next actions, important artifacts, and risks\u002Fopen questions from that block first\n- if no handoff block exists, fall back to the item summary plus latest notes\u002Fdiscoveries content\n\n### Summarize queue\n\nWhen summarizing or listing the queue, follow **Operator and agent guidance — Queue-style answers** below. Operationally: use open items only, stay compact, exclude `done` and `archived`.\n\n## Operator and agent guidance\n\nNormative for how agents phrase answers to humans and to parallel agents; does not change storage paths or section order in item files.\n\n### When to use what\n\n| Situation | Where to put it |\n| --- | --- |\n| Raw idea not ready for an `OI-####` id yet | `OPEN_ITEMS_CAPTURE.md` below the capture boundary (one `#` H1 section per idea). |\n| Durable tracked unit of work | Promote capture to `.agents\u002Fopen-items\u002Fitems\u002FOI-####.md`, or create an item file directly per `ITEM-FORMAT.md`. |\n| Chronological findings without redefining ownership | Append under `## notes\u002Fdiscoveries` on the item file. |\n| Session boundary so another human or agent can resume | Append a **Handoff** block per `HANDOFF-FORMAT.md` (adds context; does not replace notes). |\n| Executable checklist scoped to one item | `OI-####.todo.json` per `todo-contract` (mutations via `npm run todos -- …`; not the capture inbox). |\n\n### Anti-patterns\n\n- **Eternal capture:** entries that never leave `OPEN_ITEMS_CAPTURE.md` — promote into items or delete intentionally.\n- **Notes wall:** very long `notes\u002Fdiscoveries` with no recent **Handoff** — add a handoff when the story is hard to scan or you switch agents.\n- **Ghost `in_progress`:** `in_progress` with no todo sidecar (neither `.todo.json` nor legacy `.todo.md`) and no handoff naming who is acting — move to `ready`\u002F`blocked` or record a handoff.\n- **Done without verification:** moving to `done` when acceptance or checks named in the item (or `VERIFICATION.md` for the skill) were never satisfied.\n- **Todo scope leak:** unrelated work streams in one `OI-####.todo.json` — split into separate open items.\n\n### Queue-style answers (mandatory fields)\n\nWhen the operator asks to list, show, or summarize open items:\n\n1. **Scope:** open items only — exclude `done` and `archived`.\n2. **Markdown table** with columns `ID`, `Title`, `State`, `Summary` (use each item file’s `## summary`; shorten to one clause in the table when long).\n3. **Who acts next:** immediately after the table, a short subsection. Per item or grouped by state, state whether **human**, **agent**, or **either** should move it forward, inferred from `state`, latest handoff “Next actions”, and open questions.\n4. **Next concrete action:** a short list of **verifiable** next actions (commands to run, paths to open, or explicit mutations like “move OI-0004 to blocked”).\n\nIf the queue is empty, say so and mention `OPEN_ITEMS_CAPTURE.md` for new capture.\n\n## Safety rules\n\n- Do not invent ids out of order.\n- Keep the `INDEX.md` counter block correct.\n- Do not leave promoted capture entries behind.\n- Do not delete content above `Begin Items Capture` \u002F `---`.\n- If an item file is malformed, repair it to the canonical section order before further mutation.\n- Keep todo sidecars aligned with lifecycle: remove `OI-####.todo.json` (and legacy `OI-####.todo.md`) when the owning item is `done` or `archived`; use `paused`\u002Fcleared tasks or **`npm run todos -- delete`** when deferring execution work (details in `OPERATIONS.md`).\n\n## Suggested response shape\n\n### Action\n- what was requested\n\n### Changes\n- files created\u002Fupdated\n- ids created or states moved\n\n### Queue Snapshot\n- compact summary of open items using the **Queue-style answers** table plus **Who acts next** and **Next concrete action** from § Operator and agent guidance\n\n### Follow-ups\n- anything the operator may want to triage next (include stale handoffs or capture backlog when relevant)\n","# Open Items\n\nMaintain the repo's lightweight open-items system.\n\nThe open-items policy lives with this skill, not in a separate repo-root policy file.\n\nV1 uses:\n- capture inbox: `OPEN_ITEMS_CAPTURE.md`\n- workspace root: `.agents\u002Fopen-items\u002F`\n- item files: `.agents\u002Fopen-items\u002Fitems\u002FOI-####.md`\n- derived index: `.agents\u002Fopen-items\u002FINDEX.md`\n- dev CLI (bulk list, validate, regen index, capture snapshot): `npm run open-items -- …` — see § **CLI tooling**\n\nPer-item files are the source of truth. `INDEX.md` is derived.\n\nAt repo root, keep only the capture inbox file: `OPEN_ITEMS_CAPTURE.md`.\n\n## Mandatory first reads\n\nThis skill directory is the policy source of truth for the open-items system.\n\nRead these before making changes:\n- `OPEN_ITEMS_CAPTURE.md`\n- `.agents\u002Fopen-items\u002FINDEX.md` if it exists\n- for **read-only “whole queue”** questions, prefer `npm run open-items -- list` or `list --json` (stdout) before opening every item file — § **CLI tooling**\n- any target item file being changed under `.agents\u002Fopen-items\u002Fitems\u002F`\n- supporting docs in this skill directory when the operation depends on format or mutation rules\n\nSupporting docs in this directory:\n- `ideas.md` — roadmap, CLI\u002Fscript ideas, multi-agent orchestration notes (non-normative)\n- `CAPTURE-FORMAT.md`\n- `ITEM-FORMAT.md`\n- `INDEX-FORMAT.md`\n- `HANDOFF-FORMAT.md`\n- `OPERATIONS.md`\n\n## CLI tooling (repo)\n\nTypeScript CLI: `.pi\u002Fskills\u002Fopen-items\u002Fcli\u002FopenItemsCli.ts` (with `openItemsLib.ts`). **Edit only under `.pi\u002Fskills\u002F`**; run `npm run skills:sync -- --skill open-items` so `.claude\u002F`, `.cursor\u002F`, and `.codex\u002F` copies stay identical. From repo root: `npm run open-items -- \u003Ccommand>` (see root `package.json`). Without a script, from repo root: `npx tsx .pi\u002Fskills\u002Fopen-items\u002Fcli\u002FopenItemsCli.ts \u003Ccommand>`.\n\n| Command | Output |\n| --- | --- |\n| `list` | **stdout:** markdown table of **open** items (id, state, title, summary clip, path). |\n| `list --json` | **stdout:** JSON array of open items (full summary, paths). |\n| `list --all` | Same as `list`, but includes `done` and `archived`. |\n| `validate` | **stderr:** errors and warnings; **exit 1** if any **error** (malformed item, index state mismatch, bad counter). |\n| `index` | Rewrites `.agents\u002Fopen-items\u002FINDEX.md` from item files and advances counter to max id + 1. |\n| `init` | Creates missing `OPEN_ITEMS_CAPTURE.md`, `.agents\u002Fopen-items\u002Fitems\u002F`, and `.agents\u002Fopen-items\u002FINDEX.md` in the selected scope. |\n| `capture` | **stdout:** pending `#` H1 capture titles below the capture boundary (with line numbers). |\n| `where` | **stdout:** resolved repo root, selected scope root, items dir, index path, and capture path. |\n\nOptional for every command: `--root \u003Cabs-or-rel-path>` to point at another repo checkout.\n\nProject-scoped queues: use `--project \u003Cname-or-path>` to target a Flow\u002Fproject-local open-items queue without changing the repo root. Bare names resolve to `flow-library\u002F\u003Cname>`; path-like values resolve relative to `--root`; targets outside `--root` are refused to avoid cross-contamination. Examples:\n\n```powershell\nnpm run open-items -- init --project crossfit\nnpm run open-items -- list --project crossfit\nnpm run open-items -- validate --project flow-library\u002Fcrossfit\nnpm run open-items -- where --project crossfit\n```\n\nA project-scoped queue uses the same storage contract under the selected scope:\n\n```text\nflow-library\u002F\u003Cproject>\u002F\n  OPEN_ITEMS_CAPTURE.md\n  .agents\u002Fopen-items\u002FINDEX.md\n  .agents\u002Fopen-items\u002Fitems\u002FOI-####.md\n```\n\nDefault commands without `--project` continue to use the repo-level queue under repo-root `.agents\u002Fopen-items\u002F`.\n\n### When agents should run the CLI (prefer script stdout over token-heavy file walks)\n\n- **Operator asks for the full queue, a table, or “what is open”** — run `list` or `list --json` and base the answer on script output instead of re-opening every `OI-####.md` manually.\n- **Before or after bulk edits to many items** — run `validate` so counter, index bullets, and item shapes are consistent.\n- **After changing titles or states across several items** — run `index` to regenerate `INDEX.md` deterministically (same rules as manual regen in this skill).\n- **Before promoting capture** — run `capture` to list pending H1 sections explicitly.\n\nStill **open and edit** the specific `OI-####.md` (and `HANDOFF-FORMAT.md`) when recording handoffs, append-only notes, resume, or any prose mutation—the CLI does not replace reading the file you are changing.\n\n**Execution todos:** use `npm run todos -- list` \u002F `list --json` \u002F `validate` for per-item JSON sidecars (see `todo-contract`); `open-items validate` surfaces cross-surface drift with those files.\n\n## Natural-language commands\n\nExamples:\n- \"promote all capture items\"\n- \"refresh the open-items index\"\n- \"move OI-0003 to blocked\"\n- \"append this note to OI-0004\"\n- \"add a handoff to OI-0010\"\n- \"resume OI-0010\"\n- \"summarize the open-items queue\"\n\nInfer the requested action from natural language.\n\n## Storage contract\n\n### Workspace shape\n\n```text\n.agents\u002F\n  open-items\u002F\n    INDEX.md\n    items\u002F\n      OI-0001.md\n      OI-0002.md\n```\n\nCreate missing directories\u002Ffiles as needed.\n\n### Allowed states\n\n- `new`\n- `triaged`\n- `ready`\n- `in_progress`\n- `blocked`\n- `done`\n- `archived`\n\n### Per-item file contract\n\nUse this exact section order:\n\n```md\n# {title}\n\n## id\nOI-0001\n\n## state\nnew\n\n## summary\n{first paragraph or compact summary}\n\n## notes\u002Fdiscoveries\n{remaining notes}\n```\n\n### Index contract\n\n`INDEX.md` includes only open items. Exclude items whose state is:\n- `done`\n- `archived`\n\nRegenerate the index from item files sorted by id ascending.\n\nWrite bullets in this format:\n\n```md\n- OI-0001 — {title} — state: {state} — .\u002Fitems\u002FOI-0001.md\n```\n\nAt the bottom keep the counter block:\n\n```md\n## Counter (next OI id)\nOI-0002\n```\n\nStore the full next id string in the counter.\n\n### Index title policy\n\nAim for titles with 5 words or fewer.\n\nDo not fail if a title is longer.\nFor index rendering only:\n- if the title has more than 5 words, render the first 5 words plus `…`\n\nKeep the full title in the item file.\n\n## Capture inbox contract\n\n`OPEN_ITEMS_CAPTURE.md` is a staging inbox, not an archive.\n\nUse this literal boundary:\n\n```md\nBegin Items Capture\n---\n```\n\nDo not remove or rewrite content above that boundary.\n\nBelow the boundary, each `# H1` section is one capture entry:\n- the `# ...` line becomes the item title\n- the section continues until the next `# ` H1 or end-of-file\n\n### Promotion rules\n\nWhen asked to promote capture items:\n1. process all entries in file order\n2. allocate sequential ids using the `INDEX.md` counter (or bootstrap from `OI-0001`)\n3. create one item file per entry with initial state `new`\n4. set `summary` to the first paragraph after the H1 title\n5. place the remaining body in `notes\u002Fdiscoveries`\n6. remove the processed H1 sections from the capture file\n7. regenerate `INDEX.md`\n8. advance the counter to the next available `OI-####`\n\n## Mutation rules\n\n### Move state\n\nWhen asked to move an item:\n- update only the `## state` section in the item file\n- regenerate `INDEX.md`\n- if state becomes `done` or `archived`, remove it from the open-items index\n- if state becomes `done` or `archived`, delete `.agents\u002Fopen-items\u002Fitems\u002FOI-####.todo.json` if it exists, and delete legacy `.agents\u002Fopen-items\u002Fitems\u002FOI-####.todo.md` if it exists (see `OPERATIONS.md` — Cleanup and todo sidecars; follow `todo-contract` if recreating later)\n\n### Append notes\n\nWhen asked to append notes:\n- append to `## notes\u002Fdiscoveries`\n- preserve existing content\n- regenerate `INDEX.md` only if the title\u002Fstate changed elsewhere\n\n### Record handoff\n\nWhen asked to add a handoff:\n- read `HANDOFF-FORMAT.md`\n- append a canonical handoff block under `## notes\u002Fdiscoveries`\n- preserve existing content and section order\n- do not change item state unless explicitly asked\n\n### Resume from handoff\n\nWhen asked to resume an item:\n- read the latest handoff block if one exists\n- summarize current state, next actions, important artifacts, and risks\u002Fopen questions from that block first\n- if no handoff block exists, fall back to the item summary plus latest notes\u002Fdiscoveries content\n\n### Summarize queue\n\nWhen summarizing or listing the queue, follow **Operator and agent guidance — Queue-style answers** below. Operationally: use open items only, stay compact, exclude `done` and `archived`.\n\n## Operator and agent guidance\n\nNormative for how agents phrase answers to humans and to parallel agents; does not change storage paths or section order in item files.\n\n### When to use what\n\n| Situation | Where to put it |\n| --- | --- |\n| Raw idea not ready for an `OI-####` id yet | `OPEN_ITEMS_CAPTURE.md` below the capture boundary (one `#` H1 section per idea). |\n| Durable tracked unit of work | Promote capture to `.agents\u002Fopen-items\u002Fitems\u002FOI-####.md`, or create an item file directly per `ITEM-FORMAT.md`. |\n| Chronological findings without redefining ownership | Append under `## notes\u002Fdiscoveries` on the item file. |\n| Session boundary so another human or agent can resume | Append a **Handoff** block per `HANDOFF-FORMAT.md` (adds context; does not replace notes). |\n| Executable checklist scoped to one item | `OI-####.todo.json` per `todo-contract` (mutations via `npm run todos -- …`; not the capture inbox). |\n\n### Anti-patterns\n\n- **Eternal capture:** entries that never leave `OPEN_ITEMS_CAPTURE.md` — promote into items or delete intentionally.\n- **Notes wall:** very long `notes\u002Fdiscoveries` with no recent **Handoff** — add a handoff when the story is hard to scan or you switch agents.\n- **Ghost `in_progress`:** `in_progress` with no todo sidecar (neither `.todo.json` nor legacy `.todo.md`) and no handoff naming who is acting — move to `ready`\u002F`blocked` or record a handoff.\n- **Done without verification:** moving to `done` when acceptance or checks named in the item (or `VERIFICATION.md` for the skill) were never satisfied.\n- **Todo scope leak:** unrelated work streams in one `OI-####.todo.json` — split into separate open items.\n\n### Queue-style answers (mandatory fields)\n\nWhen the operator asks to list, show, or summarize open items:\n\n1. **Scope:** open items only — exclude `done` and `archived`.\n2. **Markdown table** with columns `ID`, `Title`, `State`, `Summary` (use each item file’s `## summary`; shorten to one clause in the table when long).\n3. **Who acts next:** immediately after the table, a short subsection. Per item or grouped by state, state whether **human**, **agent**, or **either** should move it forward, inferred from `state`, latest handoff “Next actions”, and open questions.\n4. **Next concrete action:** a short list of **verifiable** next actions (commands to run, paths to open, or explicit mutations like “move OI-0004 to blocked”).\n\nIf the queue is empty, say so and mention `OPEN_ITEMS_CAPTURE.md` for new capture.\n\n## Safety rules\n\n- Do not invent ids out of order.\n- Keep the `INDEX.md` counter block correct.\n- Do not leave promoted capture entries behind.\n- Do not delete content above `Begin Items Capture` \u002F `---`.\n- If an item file is malformed, repair it to the canonical section order before further mutation.\n- Keep todo sidecars aligned with lifecycle: remove `OI-####.todo.json` (and legacy `OI-####.todo.md`) when the owning item is `done` or `archived`; use `paused`\u002Fcleared tasks or **`npm run todos -- delete`** when deferring execution work (details in `OPERATIONS.md`).\n\n## Suggested response shape\n\n### Action\n- what was requested\n\n### Changes\n- files created\u002Fupdated\n- ids created or states moved\n\n### Queue Snapshot\n- compact summary of open items using the **Queue-style answers** table plus **Who acts next** and **Next concrete action** from § Operator and agent guidance\n\n### Follow-ups\n- anything the operator may want to triage next (include stale handoffs or capture backlog when relevant)\n",{"name":5,"description":7},"main","4b62590481058e38cbc5f0d1b781001205b45ff0",".pi\u002Fskills\u002Fopen-items","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk","https:\u002F\u002Fgithub.com\u002Faptlyundecided","alexdwilson.dev","carcinificating","2026-11-12 23:57:35",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add aptlyundecided\u002Fbagelwerk --skill open-items",{"tool":38,"command":39},"bunx","bunx skills add aptlyundecided\u002Fbagelwerk --skill open-items",{"tool":41,"command":42},"pnpm","pnpm dlx skills add aptlyundecided\u002Fbagelwerk --skill open-items","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FSKILL.md",true,[46,52,58,62,66,70,74,78,82,86],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"CAPTURE-FORMAT.md","REFERENCE","参考文档",636,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FCAPTURE-FORMAT.md",{"path":53,"resourceType":54,"resourceTypeDesc":55,"sizeBytes":56,"sourceHtmlUrl":57},"cli\u002FopenItemsCli.ts","SCRIPT","脚本",9333,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002Fcli\u002FopenItemsCli.ts",{"path":59,"resourceType":54,"resourceTypeDesc":55,"sizeBytes":60,"sourceHtmlUrl":61},"cli\u002FopenItemsLib.ts",12385,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002Fcli\u002FopenItemsLib.ts",{"path":63,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":64,"sourceHtmlUrl":65},"HANDOFF-FORMAT.md",1112,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FHANDOFF-FORMAT.md",{"path":67,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":68,"sourceHtmlUrl":69},"ideas.md",5966,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002Fideas.md",{"path":71,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":72,"sourceHtmlUrl":73},"INDEX-FORMAT.md",630,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FINDEX-FORMAT.md",{"path":75,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":76,"sourceHtmlUrl":77},"ITEM-FORMAT.md",807,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FITEM-FORMAT.md",{"path":79,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":80,"sourceHtmlUrl":81},"OPERATIONS.md",3825,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FOPERATIONS.md",{"path":83,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":84,"sourceHtmlUrl":85},"README.md",2994,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FREADME.md",{"path":87,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":88,"sourceHtmlUrl":89},"VERIFICATION.md",1709,"https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fopen-items\u002FVERIFICATION.md",[91,97,104,111,118,125],{"skillNo":92,"name":93,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":19},"01M78H5JMKWBHYA0Y4F7Z4NA00","structured-output-repair","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002Fsrc\u002Fcore\u002Fagent-execution\u002Fskills\u002Fstructured-output-repair\u002FSKILL.md",[],[],{"skillNo":98,"name":99,"nameZh":6,"description":100,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":101,"categories":102,"keywords":103,"publishedAt":19},"01M78H5GC0HM0KCTRZPTMD0TFK","version-release","Use when the user asks to bump\u002Fincrement the package version, prepare a release, produce changelog\u002Frelease notes, or decide whether a change should be called out in the changelog.","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fversion-release\u002FSKILL.md",[],[],{"skillNo":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":110,"publishedAt":19},"01M78H5CP4K6P6A0ADY8DM2XB0","ubiquitous-language","Extract a DDD-style ubiquitous language glossary from the current conversation, flagging ambiguities and proposing canonical terms. Saves to UBIQUITOUS_LANGUAGE.md. Use when user wants to define domain terms, build a glossary, harden terminology, create a ubiquitous language, or mentions \"domain model\" or \"DDD\".","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Fubiquitous-language\u002FSKILL.md",[],[],{"skillNo":112,"name":113,"nameZh":6,"description":114,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":115,"categories":116,"keywords":117,"publishedAt":19},"01M78H58Q2B0XM6YT7AD15AFFZ","todo-monitor","Create and advance a per-open-item agent todo list using the `npm run todos --` CLI without pausing after each task. Follows todo-contract (strict CLI-only mutation) and todo-enforcer rules.","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Ftodo-monitor\u002FSKILL.md",[],[],{"skillNo":119,"name":120,"nameZh":6,"description":121,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":122,"categories":123,"keywords":124,"publishedAt":19},"01M78H55CBZ0K977N7Q1JZ48BM","todo-hard-stop","Execute a per-open-item todo list via the `npm run todos --` CLI under the canonical todo contract, pausing after each completed task for operator feedback.","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Ftodo-hard-stop\u002FSKILL.md",[],[],{"skillNo":126,"name":127,"nameZh":6,"description":128,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":129,"categories":130,"keywords":131,"publishedAt":19},"01M78H50XT4F68HSRPDAG1R579","todo-enforcer","Enforce the per-open-item todo contract (strict CLI-only mutation via `npm run todos --`; no hand-edits to `*.todo.json`) so tracked agent work stays focused, isolated, and linked to an open item.","https:\u002F\u002Fgithub.com\u002Faptlyundecided\u002Fbagelwerk\u002Fblob\u002Fmain\u002F.pi\u002Fskills\u002Ftodo-enforcer\u002FSKILL.md",[],[]]