README.md

@@ -78,7 +78,7 @@ Wiki 位置分兩層：**內容頁**一頁型一個變數——日誌頁 `LOG_{H
 
 以上每個變數未設定都退回 `JSC_WIKI_REPO`，再未設定就詢問（見 `jsc-gitea`）。先讀目前 shell 繼承的環境變數，缺值才詢問。
 
-目錄頁與內容頁從此分屬不同存取庫，兩件事跟著改：`JSC_WIKI_REPO_CONTENTS` 不會退回頁型變數，`JSC_WIKI_REPO_LOG` 之類也頂不了目錄頁的位；目錄頁指向內容頁的連結一律用 `gitea.sh wiki-url` 的絕對網址，`[[頁名]]` 只在同一個 wiki 內解得開，跨庫就是死連結。
+目錄頁與內容頁從此分屬不同存取庫，兩件事跟著改：`JSC_WIKI_REPO_CONTENTS` 不會退回頁型變數，`JSC_WIKI_REPO_LOG` 之類也頂不了目錄頁的位；連結一律寫成 `[{文字}]({連結})`，網址取 `gitea.sh wiki-url` 印出的絕對網址，不自己組路徑。同 wiki 的雙括號寫法一概不用：它只在自己那個 wiki 內解得開，跨庫就是死連結，畫面上還看不出壞掉。連結寫進頁面前先過 `gitea.sh` 旁邊的 `link-check.sh`，結束碼 0 才寫。
 
 ## 相關 domain
 

references/behaviors.md

@@ -7,20 +7,20 @@
 | 項目 | 內容 |
 | --- | --- |
 | 觸發時機 | 一次技能執行留下可重用的教訓時，用 record 模式記錄。要跑某支技能之前，用 consult 模式查過去的教訓。工時與 Token 紀錄不走這支，走 worklog |
-| 關鍵步驟 | record 模式收齊日期、技能、CLI、情境、教訓、下次做法這六欄、從 `git remote get-url origin` 解出 `{owner}/{repo}`、用 `hash-id` 算出完整 40 碼大寫的 `{HASH}`、用 `wiki-repo LEARN` 解出教訓頁存取庫、開 sub agent 讀 `LEARN_{HASH}`、依 `wiki-get` 的退出碼分支（0 在表尾追加一列、4 才用 `templates/learn-page.md` 建頁、7 與 8 停止並回報）、取 `wiki-url` 的絕對網址、同一輪跑 `wiki-contents.sh upsert LEARN 1 {owner}/{repo}` 更新落在 CONTENTS 存取庫的 `LEARN_CONTENTS`、依它的退出碼分流（0 已寫、1 寫入失敗、2 參數錯、3 未設 `JSC_WIKI_REPO_CONTENTS`、4 缺範本、7 金鑰失效、8 其他 API 失敗）、寫入失敗重試一次；consult 模式算出 `{HASH}`、從 CONTENTS 存取庫讀 `LEARN_CONTENTS`、從 LEARN 存取庫讀 `LEARN_{HASH}`、挑出「技能」欄相符的列、整理每一列的「下次做法」交給呼叫端 |
-| 外部呼叫 | `jsc-gitea/tools/hash-id`、`jsc-gitea/tools/gitea.sh wiki-repo LEARN` 與 `wiki-repo CONTENTS` 與 `wiki-url`、`jsc-gitea/tools/wiki-contents.sh upsert`（目錄頁那一列）、`jsc-gitea:wiki`（其餘 wiki 讀寫）、`git remote get-url origin`、`templates/learn-page.md`、`templates/learn-contents.md` |
-| 完成條件 | record 模式要 sub agent 回報兩頁都寫入成功，主代理確認新列在 `LEARN_{HASH}` 上，原有的列一列不少，`wiki-contents.sh` 回 0 並印出 `updated` 或 `added`。consult 模式要兩頁都讀到，或以退出碼 4 回報頁面不存在，或在 5、7、8 停止並回報狀態 |
-| 可驗證跡象 | LEARN 存取庫的 `LEARN_{HASH}` 表尾多一列教訓；CONTENTS 存取庫的 `LEARN_CONTENTS` 上該 repo 那一列的「最後更新時間」換新，且「教訓紀錄」欄是絕對網址，不是 `[[頁名]]`。頁面原本不存在時，會新建 `LEARN_{HASH}` 或 `LEARN_CONTENTS`。consult 模式無寫入跡象，只有回報內容 |
+| 關鍵步驟 | record 模式收齊日期、技能、CLI、情境、教訓、下次做法這六欄、從 `git remote get-url origin` 解出 `{owner}/{repo}`、用 `hash-id` 算出完整 40 碼大寫的 `{HASH}`、用 `wiki-repo LEARN` 解出教訓頁存取庫、開 sub agent 讀 `LEARN_{HASH}`、依 `wiki-get` 的退出碼分支（0 在表尾追加一列、4 才用 `templates/learn-page.md` 建頁、7 與 8 停止並回報）、取 `wiki-url` 的絕對網址並寫成 `[{頁名}]({絕對網址})`、把該網址交給 `link-check.sh` 驗證（0 才寫入、1 有 DEAD 就不寫並回報那幾筆、2 沒給網址、3 未設 `GITEA_HOST`、7 金鑰被拒就停下）、同一輪跑 `wiki-contents.sh upsert LEARN 1 {owner}/{repo}` 更新落在 CONTENTS 存取庫的 `LEARN_CONTENTS`、依它的退出碼分流（0 已寫、1 寫入失敗、2 參數錯、3 未設 `JSC_WIKI_REPO_CONTENTS`、4 缺範本、7 金鑰失效、8 其他 API 失敗）、寫入失敗重試一次；consult 模式算出 `{HASH}`、從 CONTENTS 存取庫讀 `LEARN_CONTENTS`、從 LEARN 存取庫讀 `LEARN_{HASH}`、依列上的絕對網址讀頁、挑出「技能」欄相符的列、整理每一列的「下次做法」交給呼叫端 |
+| 外部呼叫 | `jsc-gitea/tools/hash-id`、`jsc-gitea/tools/gitea.sh wiki-repo LEARN` 與 `wiki-repo CONTENTS` 與 `wiki-url`、`jsc-gitea/tools/link-check.sh`（寫入前驗證連結）、`jsc-gitea/tools/wiki-contents.sh upsert`（目錄頁那一列）、`jsc-gitea:wiki`（其餘 wiki 讀寫）、`git remote get-url origin`、`templates/learn-page.md`、`templates/learn-contents.md` |
+| 完成條件 | record 模式要 sub agent 回報兩頁都寫入成功，主代理確認新列在 `LEARN_{HASH}` 上，原有的列一列不少，寫入前 `link-check.sh` 回 0，`wiki-contents.sh` 回 0 並印出 `updated` 或 `added`。`link-check.sh` 回 1 就不寫目錄頁，改回報 DEAD 清單。consult 模式要兩頁都讀到，或以退出碼 4 回報頁面不存在，或在 5、7、8 停止並回報狀態 |
+| 可驗證跡象 | LEARN 存取庫的 `LEARN_{HASH}` 表尾多一列教訓；CONTENTS 存取庫的 `LEARN_CONTENTS` 上該 repo 那一列的「最後更新時間」換新，且「教訓紀錄」欄是 `[{頁名}]({絕對網址})`，頁面上沒有 `[[...]]` 這種同 wiki 寫法。頁面原本不存在時，會新建 `LEARN_{HASH}` 或 `LEARN_CONTENTS`。連結驗證不過就兩頁都沒有新內容，只有 DEAD 清單的回報。consult 模式無寫入跡象，只有回報內容 |
 
 ## report
 
 | 項目 | 內容 |
 | --- | --- |
 | 觸發時機 | 有人要一段期間的工作總結時用，期間分年、月、週、日四種。單一工作包的紀錄不走這支，走 worklog |
-| 關鍵步驟 | 呼叫端沒指定期間就依 `jsc-ask:ask` 問、跑 `tools/report-range.sh` 取得起訖日與標籤、平行做四件事（`tools/report-template.sh resolve` 解出範本、從 CONTENTS 存取庫讀 `LOG_CONTENTS` 並照列上的絕對網址逐頁讀回工作日誌、年報另外讀 `LEARN_CONTENTS` 並用 `wiki-repo LEARN` 解出教訓內容頁的存取庫、解出 REPORT 的 wiki repo）、把頁面內容存成檔案後跑 `tools/log-aggregate.sh` 算出條目數、涵蓋 repo、花費時間、Token 用量、任務狀態、從存活條目挑出阻塞與未完成工作包、照範本的標題與表格填出報告、用 `hash-id "{REPORT wiki 存取庫}/{period}"` 算出完整 40 碼頁名、讀 `REPORT_{HASH}` 後把本期當成新章節追加在最前、取 `wiki-url` 的絕對網址並依它的退出碼分流（4 是頁還沒寫、5 是頁上沒有 `html_url`、7 與 8 一律停下並回報金鑰或 API 狀態，不得當成 4）、最後跑 `wiki-contents.sh upsert REPORT 2 {HASH}` 更新落在 CONTENTS 存取庫的 `REPORT_CONTENTS`（鍵取第 2 欄的裸 HASH，不取第 1 欄那個帶主機名的連結） |
-| 外部呼叫 | `tools/report-range.sh`、`tools/report-template.sh`、`tools/log-aggregate.sh`、`jsc-gitea/tools/gitea.sh wiki-repo`（CONTENTS、LOG、LEARN、REPORT）與 `hash-id` 與 `wiki-url`、`jsc-gitea/tools/wiki-contents.sh upsert`（目錄頁那一列）、`jsc-gitea:wiki`、`jsc-ask:ask`、`templates/report-{period}.md`、`templates/report-contents.md` |
-| 完成條件 | 回報頁面 URL，或回報跳過寫入與它的原因，或在讀取回 7、8 時停下並回報狀態。`wiki-contents.sh upsert` 要回 0，回 1、2、3、4、7、8 就照該碼回報且不得謊報已寫入。收尾要講出期間標籤、條目數、涵蓋的 repo、範本來源、頁面 URL、帶到下一期的未完成工作包清單，以及 `ELAPSED_MISSING` 與 `TOKEN_MISSING` |
-| 可驗證跡象 | REPORT 存取庫的 `REPORT_{HASH}`（雜湊取自 REPORT wiki 存取庫的 `{owner}/{repo}` 加期間，不是程式碼存取庫）多一個本期章節；CONTENTS 存取庫的 `REPORT_CONTENTS` 該列的「最新一期」、「期數」、「最後更新」換新，且「報表頁」欄是絕對網址、「HASH」欄是不帶連結的裸 HASH。重跑同一期間只換掉那一列，不會多出第二列。本機留下工作日誌頁面內容的暫存檔，供 `log-aggregate.sh` 讀取。沒設定 REPORT wiki repo 時不寫 wiki，只印出報告本文；沒設定 `JSC_WIKI_REPO_CONTENTS` 時內容頁照寫，只有目錄頁那一列沒動 |
+| 關鍵步驟 | 呼叫端沒指定期間就依 `jsc-ask:ask` 問、跑 `tools/report-range.sh` 取得起訖日與標籤、平行做四件事（`tools/report-template.sh resolve` 解出範本、從 CONTENTS 存取庫讀 `LOG_CONTENTS` 並照列上的絕對網址逐頁讀回工作日誌、年報另外讀 `LEARN_CONTENTS` 並用 `wiki-repo LEARN` 解出教訓內容頁的存取庫、解出 REPORT 的 wiki repo）、把頁面內容存成檔案後跑 `tools/log-aggregate.sh` 算出條目數、涵蓋 repo、花費時間、Token 用量、任務狀態、從存活條目挑出阻塞與未完成工作包、照範本的標題與表格填出報告、用 `hash-id "{REPORT wiki 存取庫}/{period}"` 算出完整 40 碼頁名、讀 `REPORT_{HASH}` 後把本期當成新章節追加在最前、取 `wiki-url` 的絕對網址並依它的退出碼分流（4 是頁還沒寫、5 是頁上沒有 `html_url`、7 與 8 一律停下並回報金鑰或 API 狀態，不得當成 4）、章節內與目錄列的連結一律寫成 `[{文字}]({絕對網址})` 並在寫入前全數交給 `link-check.sh` 驗證（0 才寫入、1 有 DEAD 就不寫並回報那幾筆、2 沒給網址、3 未設 `GITEA_HOST`、7 金鑰被拒就停下）、最後跑 `wiki-contents.sh upsert REPORT 2 {HASH}` 更新落在 CONTENTS 存取庫的 `REPORT_CONTENTS`（鍵取第 2 欄的裸 HASH，不取第 1 欄那個帶主機名的連結） |
+| 外部呼叫 | `tools/report-range.sh`、`tools/report-template.sh`、`tools/log-aggregate.sh`、`jsc-gitea/tools/gitea.sh wiki-repo`（CONTENTS、LOG、LEARN、REPORT）與 `hash-id` 與 `wiki-url`、`jsc-gitea/tools/link-check.sh`（寫入前驗證連結）、`jsc-gitea/tools/wiki-contents.sh upsert`（目錄頁那一列）、`jsc-gitea:wiki`、`jsc-ask:ask`、`templates/report-{period}.md`、`templates/report-contents.md` |
+| 完成條件 | 回報頁面 URL，或回報跳過寫入與它的原因，或在讀取回 7、8 時停下並回報狀態。每次寫入前 `link-check.sh` 要回 0，回 1 就不寫該頁並回報 DEAD 清單。`wiki-contents.sh upsert` 要回 0，回 1、2、3、4、7、8 就照該碼回報且不得謊報已寫入。收尾要講出期間標籤、條目數、涵蓋的 repo、範本來源、頁面 URL、帶到下一期的未完成工作包清單，以及 `ELAPSED_MISSING` 與 `TOKEN_MISSING` |
+| 可驗證跡象 | REPORT 存取庫的 `REPORT_{HASH}`（雜湊取自 REPORT wiki 存取庫的 `{owner}/{repo}` 加期間，不是程式碼存取庫）多一個本期章節；CONTENTS 存取庫的 `REPORT_CONTENTS` 該列的「最新一期」、「期數」、「最後更新」換新，且「報表頁」欄是 `[{頁名}]({絕對網址})`、「HASH」欄是不帶連結的裸 HASH，兩頁都找不到 `[[...]]` 這種同 wiki 寫法。連結驗證不過就沒有新章節，也沒有新的目錄列，只有 DEAD 清單的回報。重跑同一期間只換掉那一列，不會多出第二列。本機留下工作日誌頁面內容的暫存檔，供 `log-aggregate.sh` 讀取。沒設定 REPORT wiki repo 時不寫 wiki，只印出報告本文；沒設定 `JSC_WIKI_REPO_CONTENTS` 時內容頁照寫，只有目錄頁那一列沒動 |
 
 ## stats
 
@@ -37,7 +37,7 @@
 | 項目 | 內容 |
 | --- | --- |
 | 觸發時機 | implement 或 maintain 階段每結束一項任務就寫一筆。一項任務是一個工作包、一輪 PR 意見修正，或一次獨立的修正提交。下一項任務開始前就要寫完。規劃階段的筆記不走這支 |
-| 關鍵步驟 | 全程由 sub agent 收集與寫入、平行取得十項事實（`git remote get-url origin` 的 repo、`git branch --show-current` 的分支、PLAN 頁絕對連結、ANALYZE 頁工作包絕對連結、`session-timer.sh report` 的花費時間、`token-usage.sh` 的各 CLI Token、任務狀態、細節與產出、困難與解法、PR 目標分支）、平行解出 LOG wiki repo（只供內容頁）與完整 40 碼大寫的 `{HASH}`、跑 `tools/worklog-target.sh` 取得頁名與本週五日期、用 `templates/log-entry.md` 填出單筆條目檔、跑 `tools/worklog-pending.sh merge` 併入待寫內容、讀 `PAGE` 後依退出碼分支（0 追加在頁尾、4 才建頁、7 與 8 停止且不建頁）、同一輪用 `wiki-repo CONTENTS` 解出目錄頁存取庫、取 `wiki-url` 的絕對網址、跑 `wiki-contents.sh upsert LOG 2 {HASH}` 更新 `LOG_CONTENTS`（鍵取第 2 欄的裸 HASH，不取第 1 欄那個帶主機名的連結）並依 0、1、2、3、4、7、8 各自分流、最後依成敗跑 `worklog-pending.sh commit` 或 `abort` |
-| 外部呼叫 | `jsc-hooks/hooks/session-timer.sh report`、`tools/token-usage.sh`、`tools/worklog-target.sh`、`tools/worklog-pending.sh`、`jsc-gitea/tools/gitea.sh wiki-repo`（LOG 與 CONTENTS）與 `wiki-url`、`jsc-gitea/tools/wiki-contents.sh upsert`（目錄頁那一列）、`jsc-gitea/tools/hash-id`、`jsc-gitea:wiki`、`jsc-ask:ask`、`git remote get-url origin`、`git branch --show-current`、`templates/log-entry.md`、`templates/log-contents.md` |
-| 完成條件 | `MERGED` 的每一筆條目都在 `PAGE` 上，原有條目逐字不動，`wiki-contents.sh upsert` 回 0 且 `CONTENTS` 該列帶著本週五日期、其他列不動，`worklog-pending.sh` 的 `commit` 或 `abort` 其中一個跑過並回報退出碼。`upsert` 回非 0 就照該碼回報，並把步驟七當成失敗處理 |
-| 可驗證跡象 | LOG 存取庫的 `LOG_{HASH}` 頁尾多一筆條目；CONTENTS 存取庫的 `LOG_CONTENTS` 該列的「條目數」與「最後更新」換新，且「日誌頁」欄是絕對網址，不是 `[[頁名]]`，「HASH」欄是不帶連結的裸 HASH。重跑同一頁只換掉那一列，不會多出第二列。`$JSC_HOME/worklog-pending/{HASH}` 底下的待寫檔在 `commit` 後清空，`abort` 後原樣保留；該目錄名是 40 碼大寫十六進位，或尚未遷移的舊暫存那種 8 碼大寫十六進位、`H` 加 7 碼大寫十六進位。本機留下填好的條目檔 |
+| 關鍵步驟 | 全程由 sub agent 收集與寫入、平行取得十項事實（`git remote get-url origin` 的 repo、`git branch --show-current` 的分支、PLAN 頁絕對連結、ANALYZE 頁工作包絕對連結、`session-timer.sh report` 的花費時間、`token-usage.sh` 的各 CLI Token、任務狀態、細節與產出、困難與解法、PR 目標分支）、平行解出 LOG wiki repo（只供內容頁）與完整 40 碼大寫的 `{HASH}`、跑 `tools/worklog-target.sh` 取得頁名與本週五日期、用 `templates/log-entry.md` 填出單筆條目檔、跑 `tools/worklog-pending.sh merge` 併入待寫內容、讀 `PAGE` 後依退出碼分支（0 追加在頁尾、4 才建頁、7 與 8 停止且不建頁）、寫入前把 `MERGED` 裡的每個連結交給 `link-check.sh` 驗證（0 才寫入、1 有 DEAD 就不寫並回報那幾筆、2 沒給網址、3 未設 `GITEA_HOST`、7 金鑰被拒就停下）、同一輪用 `wiki-repo CONTENTS` 解出目錄頁存取庫、取 `wiki-url` 的絕對網址並寫成 `[{頁名}]({絕對網址})`、同樣先過 `link-check.sh` 才跑 `wiki-contents.sh upsert LOG 2 {HASH}` 更新 `LOG_CONTENTS`（鍵取第 2 欄的裸 HASH，不取第 1 欄那個帶主機名的連結）並依 0、1、2、3、4、7、8 各自分流、最後依成敗跑 `worklog-pending.sh commit` 或 `abort` |
+| 外部呼叫 | `jsc-hooks/hooks/session-timer.sh report`、`tools/token-usage.sh`、`tools/worklog-target.sh`、`tools/worklog-pending.sh`、`jsc-gitea/tools/gitea.sh wiki-repo`（LOG 與 CONTENTS）與 `wiki-url`、`jsc-gitea/tools/link-check.sh`（寫入前驗證連結）、`jsc-gitea/tools/wiki-contents.sh upsert`（目錄頁那一列）、`jsc-gitea/tools/hash-id`、`jsc-gitea:wiki`、`jsc-ask:ask`、`git remote get-url origin`、`git branch --show-current`、`templates/log-entry.md`、`templates/log-contents.md` |
+| 完成條件 | 兩次寫入前 `link-check.sh` 都回 0，`MERGED` 的每一筆條目都在 `PAGE` 上，原有條目逐字不動，`wiki-contents.sh upsert` 回 0 且 `CONTENTS` 該列帶著本週五日期、其他列不動，`worklog-pending.sh` 的 `commit` 或 `abort` 其中一個跑過並回報退出碼。`link-check.sh` 回 1 或 `upsert` 回非 0 就照該碼回報，並把步驟七當成失敗處理，讓待寫內容留著 |
+| 可驗證跡象 | LOG 存取庫的 `LOG_{HASH}` 頁尾多一筆條目，條目裡的計畫名稱、工作包編號、PR 目標分支三欄都是 `[{文字}]({絕對網址})`；CONTENTS 存取庫的 `LOG_CONTENTS` 該列的「條目數」與「最後更新」換新，且「日誌頁」欄是 `[{頁名}]({絕對網址})`，頁面上沒有 `[[...]]` 這種同 wiki 寫法，「HASH」欄是不帶連結的裸 HASH。連結驗證不過就兩頁都沒有新內容，待寫檔原樣保留。重跑同一頁只換掉那一列，不會多出第二列。`$JSC_HOME/worklog-pending/{HASH}` 底下的待寫檔在 `commit` 後清空，`abort` 後原樣保留；該目錄名是 40 碼大寫十六進位，或尚未遷移的舊暫存那種 8 碼大寫十六進位、`H` 加 7 碼大寫十六進位。本機留下填好的條目檔 |

skills/learn/SKILL.md

@@ -11,7 +11,8 @@ Close the loop on skill runs: record what a run taught you, consult it before th
 
 - Directory page: `LEARN_CONTENTS`. Content page: `LEARN_{HASH}`, one page per repository.
 - The two pages live in different wikis. `LEARN_{HASH}` goes to `jsc-gitea/tools/gitea.sh wiki-repo LEARN`; `LEARN_CONTENTS` goes to `gitea.sh wiki-repo CONTENTS` (`JSC_WIKI_REPO_CONTENTS`, then `JSC_WIKI_REPO`, then exit 3), which never falls back to the LEARN repo. Exit 3 on either hands the question to `jsc-gitea:wiki`, which owns the resolution order and the wording; exit 2 means the type argument was misspelled, so fix it and rerun.
-- Because they sit in different wikis, the directory row links the lesson page by the absolute URL from `gitea.sh wiki-url <LEARN repo> LEARN_{HASH}`. `[[LEARN_{HASH}]]` resolves only inside one wiki and would dead-link from the directory. `wiki-url` exit 4 means the lesson page is not written yet, so write it first; exit 5 means the page carries no `html_url`, so stop and report it and never assemble the URL by hand; exit 7 or 8 means the token or the API failed, so stop and report that status.
+- Every link this skill writes takes the shape `[{text}]({absolute URL})`. The directory row links the lesson page as `[LEARN_{HASH}](<url>)`, with `<url>` from `gitea.sh wiki-url <LEARN repo> LEARN_{HASH}` — never assembled by hand, and never the same-wiki `[[...]]` form, which resolves inside one wiki only and dead-links from the directory without reporting an error. `wiki-url` exit 4 means the lesson page is not written yet, so write it first; exit 5 means the page carries no `html_url`, so stop and report it and never assemble the URL by hand; exit 7 or 8 means the token or the API failed, so stop and report that status.
+- Check every link before it reaches a page: `jsc-gitea/tools/link-check.sh {url}...`. Only exit 0 permits the write.
 - Compute `{HASH}` from `{owner}/{repo}` with `jsc-gitea/tools/hash-id`. It prints the full 40-character uppercase SHA-1 — no truncation and no prefix rewrite, so never shorten it. Exit 1 means no SHA-1 helper on this machine: stop and report that `sha1sum` or `shasum` has to be installed, and never hand-compute the hash. Exit 2 means the input was empty, so fix the `{owner}/{repo}` parse and rerun.
 - All wiki reads and writes go through `jsc-gitea:wiki`, except the `LEARN_CONTENTS` row, which goes through `jsc-gitea/tools/wiki-contents.sh`.
 
@@ -41,6 +42,16 @@ Run after a skill run that produced a reusable lesson.
      | 7 | The token is invalid or lacks permission, so the old rows are unknown. Stop and report the token problem, and create no page |
      | 8 | Some other API failure. Stop and report that status, and create no page |
 
+   - Check the row's link before writing it. Feed the `wiki-url` output to `jsc-gitea/tools/link-check.sh`; it prints one `{OK|DEAD|SKIP}<TAB>{url}<TAB>{note}` line per URL and resolves Gitea URLs through the API, because a private repo answers a logged-out web request with 404 and would fail a page that is there.
+
+     | Exit | Do |
+     | --- | --- |
+     | 0 | Every link answered. Write the row |
+     | 1 | At least one link is DEAD. Write nothing and report the DEAD lines to the caller |
+     | 2 | No URL reached the script. Pass the URLs and rerun |
+     | 3 | The list holds a Gitea URL but `GITEA_HOST` is unset. Set it and rerun; never skip the check |
+     | 7 | The Gitea token was rejected (HTTP 401/403). Stop and report the token problem. A rejected token makes live pages look missing, and one batch judged on that answer wipes out links that still work |
+
    - Update `LEARN_CONTENTS` in the same pass, and let `jsc-gitea/tools/wiki-contents.sh` do the row work — never hand-edit the directory page. Build one file holding the single row from `templates/learn-contents.md` (the repository name, the absolute link from `gitea.sh wiki-url <LEARN repo> LEARN_{HASH}`, and the update time), then run:
 
      `jsc-gitea/tools/wiki-contents.sh upsert LEARN 1 "{owner}/{repo}" {row file} templates/learn-contents.md`

skills/report/SKILL.md

@@ -74,8 +74,18 @@ Write through `jsc-gitea:wiki`:
 - Repo: the REPORT repo from step 2, line 4. It hosts the content page only; `REPORT_CONTENTS` goes to the CONTENTS repo instead.
 - Page: `REPORT_` plus `gitea.sh hash-id "{owner}/{repo}/{period}"`. Here `{owner}/{repo}` is **the REPORT wiki repo itself** — the value `gitea.sh wiki-repo REPORT` printed — and not the code repo the logs came from. Every other page in this skill set hashes the code repo; this one page does not, because a report spans every code repo whose logs landed in the range, so no single code repo names it. Feed `hash-id` the exact string `{REPORT wiki owner}/{REPORT wiki repo}/{period}`, with `{period}` being the literal `daily`, `weekly`, `monthly` or `yearly` — so year, month, week and day each get their own page. `hash-id` prints the full 40-character uppercase SHA-1: use it whole, never shortened and never prefixed.
 - Read the page first and branch on the exit code the underlying `gitea.sh wiki-get` returned. **Only exit 4 means the page is not there yet** and may be built from scratch. On exit 0 the existing sections are in hand, so append into them. On exit 7 the token is invalid or lacks permission, and on exit 8 the API failed some other way: both leave the earlier periods unknown, so stop, report the status and write nothing — a page rebuilt on top of an unread read loses every period already on it.
+- Check every link before it goes on a page — the ones inside the new section and the row's link alike: `jsc-gitea/tools/link-check.sh {url}...`. It prints one `{OK|DEAD|SKIP}<TAB>{url}<TAB>{note}` line per URL and resolves Gitea URLs through the API, because a private repo answers a logged-out web request with 404 and would fail a page that is there. Only exit 0 permits the write.
+
+  | Exit | Do |
+  | --- | --- |
+  | 0 | Every link answered. Write the section, or the row |
+  | 1 | At least one link is DEAD. Write nothing and report the DEAD lines to the caller |
+  | 2 | No URL reached the script. Pass the URLs and rerun |
+  | 3 | The list holds a Gitea URL but `GITEA_HOST` is unset. Set it and rerun; never skip the check |
+  | 7 | The Gitea token was rejected (HTTP 401/403). Stop and report the token problem. A rejected token makes live pages look missing, and one batch judged on that answer wipes out links that still work |
+
 - Append this period as a new section, newest first. Rerunning the same period replaces that period's section only, leaving the other periods untouched.
-- Refresh the page's row in `REPORT_CONTENTS` with `jsc-gitea/tools/wiki-contents.sh` — never hand-edit the directory page. It sits in the CONTENTS repo, not the REPORT repo, so the row links the report page by the absolute URL from `gitea.sh wiki-url <REPORT repo> REPORT_{HASH}`; `[[REPORT_{HASH}]]` resolves only inside one wiki and would dead-link from here. Build one file holding the single row from `templates/report-contents.md` (the absolute link, the bare `{HASH}`, the period, the newest label, the section count and the update time), then run:
+- Refresh the page's row in `REPORT_CONTENTS` with `jsc-gitea/tools/wiki-contents.sh` — never hand-edit the directory page. It sits in the CONTENTS repo, not the REPORT repo, so the row links the report page as `[REPORT_{HASH}](<url>)`, with `<url>` from `gitea.sh wiki-url <REPORT repo> REPORT_{HASH}`. Every link on the report body and on this row takes that same `[{text}]({absolute URL})` shape; the same-wiki `[[...]]` form resolves inside one wiki only and dead-links from here without reporting an error. Build one file holding the single row from `templates/report-contents.md` (the absolute link, the bare `{HASH}`, the period, the newest label, the section count and the update time), then run:
 
   `jsc-gitea/tools/wiki-contents.sh upsert REPORT 2 "{HASH}" {row file} templates/report-contents.md`
 

skills/worklog/SKILL.md

@@ -27,14 +27,16 @@ Rows 3, 4, 5 and 6 each hit a different source and none of them reads another's
 | --- | --- | --- |
 | 1 | Repository name | Parse `{owner}/{repo}` from `git remote get-url origin`. This is the code repo — never pass it to `wiki-url`, which takes the wiki-hosting repo |
 | 2 | Branch name | `git branch --show-current` |
-| 3 | Plan name | Absolute link to the plan page: `[PLAN_{HASH}](<url>)`. Resolve the hosting repo with `jsc-gitea/tools/gitea.sh wiki-repo PLAN`, then take `<url>` from `gitea.sh wiki-url <that repo> PLAN_{HASH}` — PLAN and LOG may live in different wiki repos, and `[[...]]` only resolves inside one wiki. `wiki-repo` exit 3 (no wiki repo configured for that type) or `wiki-url` exit 4 (page not found) → fill the literal 「無」 for this row and carry on; a `worklog` run triggered from `maintain` normally has no plan page. `wiki-url` exit 5 → stop and report that the page carries no `html_url`; never assemble the URL by hand. `wiki-url` exit 7 (token invalid or no permission, HTTP 401/403) or exit 8 (other API failure) → stop and report the token or API status; never fill 「無」, because that records a page that exists as a page that does not |
-| 4 | Work package id | Absolute link to the work package heading: `[WP-xx](<url>#wp-xx)`. Resolve the hosting repo with `gitea.sh wiki-repo ANALYZE`, then take `<url>` from `gitea.sh wiki-url <that repo> ANALYZE_{HASH}`. A comment-fix round links to the same work package it belongs to. Same branching as row 3: `wiki-repo` exit 3 or `wiki-url` exit 4 → fill 「無」 and carry on; `wiki-url` exit 5 → stop and report that the page carries no `html_url`, never assemble the URL by hand; `wiki-url` exit 7 or 8 → stop and report the token or API status, never fill 「無」 |
+| 3 | Plan name | Link to the plan page in the shape `[PLAN_{HASH}](<url>)`. Resolve the hosting repo with `jsc-gitea/tools/gitea.sh wiki-repo PLAN`, then take `<url>` from `gitea.sh wiki-url <that repo> PLAN_{HASH}` — never assemble it by hand, and never use the same-wiki `[[...]]` form: PLAN and LOG may live in different wiki repos, and `[[...]]` resolves inside one wiki only. `wiki-repo` exit 3 (no wiki repo configured for that type) or `wiki-url` exit 4 (page not found) → fill the literal 「無」 for this row and carry on; a `worklog` run triggered from `maintain` normally has no plan page. `wiki-url` exit 5 → stop and report that the page carries no `html_url`; never assemble the URL by hand. `wiki-url` exit 7 (token invalid or no permission, HTTP 401/403) or exit 8 (other API failure) → stop and report the token or API status; never fill 「無」, because that records a page that exists as a page that does not |
+| 4 | Work package id | Link to the work package heading in the shape `[WP-xx](<url>#wp-xx)`. Resolve the hosting repo with `gitea.sh wiki-repo ANALYZE`, then take `<url>` from `gitea.sh wiki-url <that repo> ANALYZE_{HASH}`. A comment-fix round links to the same work package it belongs to. Same branching as row 3: `wiki-repo` exit 3 or `wiki-url` exit 4 → fill 「無」 and carry on; `wiki-url` exit 5 → stop and report that the page carries no `html_url`, never assemble the URL by hand; `wiki-url` exit 7 or 8 → stop and report the token or API status, never fill 「無」 |
 | 5 | Elapsed time | `jsc-hooks/hooks/session-timer.sh report {session_id}` prints `{sid} {seconds}` and always exits 0. Convert the seconds to h/m and read it at the moment the task ends, so it counts only this task. `0` means the timer holds no start record for that session, not a task that took no time: fill the literal 「無資料」 and say the timer had no record. Never estimate the duration from the transcript |
 | 6 | Token usage | `tools/token-usage.sh <cli> {session_id}` per CLI that ran; it prints `input<TAB>output`. Pass the same `{session_id}` as row 5 so the elapsed time and the token count describe one task. Fill `N/A` in both columns when it prints `N/A`; exit 2 means the CLI name is not one of claude / codex / copilot / antigravity / kiro, so fix the name and rerun |
 | 7 | Task status | One of the literal values 「完成」, 「部分完成」, 「阻塞」 (with reason when blocked). Derive it from the session when the session shows it; otherwise ask via `jsc-ask:ask`, offering those three literals as the options and stating each option's impact scope (「完成」 closes the task, 「部分完成」 leaves the remainder open for the next run, 「阻塞」 records the blocker and hands it back to the operator) |
 | 8 | Details and outputs | One line per changed file or produced page: what changed there and why. A round that changed nothing says what was tried and why it was dropped |
 | 9 | Difficulties and resolutions | One pair per line. Ask via `jsc-ask:ask` when the session does not show them |
-| 10 | PR target branch | Link to the PR page |
+| 10 | PR target branch | Link to the PR page in the shape `[{base branch}]({PR URL})` |
+
+Every link in an entry is text plus an absolute URL — `[{text}]({url})` — with wiki URLs coming from `gitea.sh wiki-url`. The same-wiki `[[...]]` form is out: it resolves inside one wiki only, and a link that fails that way looks like ordinary text on the page instead of reporting an error.
 
 The `{HASH}` in every page name above is computed with `jsc-gitea/tools/hash-id`, which prints the full 40-character uppercase SHA-1 of its input — no truncation to 8 characters and no prefix rewrite. A page name shortened by hand points at a page nobody else writes to.
 
@@ -65,10 +67,22 @@ The `{HASH}` in every page name above is computed with `jsc-gitea/tools/hash-id`
    | 7 | The key is invalid or lacks permission, so the old content is unknown. Stop and report the key problem, write nothing and create no page — a page created here would take the place of a log that is still on the server |
    | 8 | Some other API failure. Stop and report that status, write nothing and create no page. Retry only after the API is back |
 
+   Before the write, hand every URL that `MERGED` carries — plan page, work package heading, PR page — to `jsc-gitea/tools/link-check.sh {url}...`. It prints one `{OK|DEAD|SKIP}<TAB>{url}<TAB>{note}` line per URL and resolves Gitea URLs through the API, because a private repo answers a logged-out web request with 404 and would fail a page that is there. Only exit 0 permits the write.
+
+   | Exit | Do |
+   | --- | --- |
+   | 0 | Every link answered. Write the entries |
+   | 1 | At least one link is DEAD. Write nothing, report the DEAD lines, and go to step 7 as a failure so the pending content survives |
+   | 2 | No URL reached the script. Pass the URLs and rerun |
+   | 3 | The list holds a Gitea URL but `GITEA_HOST` is unset. Set it and rerun; never skip the check |
+   | 7 | The Gitea token was rejected (HTTP 401/403). Stop and report the token problem. A rejected token makes live pages look missing, and one batch judged on that answer wipes out links that still work |
+
    A failed write stops the run and goes to step 7 as a failure — never report the page as written when it was not. Done when every entry in `MERGED` is on `PAGE`, every entry that was already there is still there, and any 4 / 7 / 8 branch was followed as stated.
 6. Update `CONTENTS` in the same pass, and let `jsc-gitea/tools/wiki-contents.sh` do the row work — never hand-edit the directory page.
 
-   `LOG_CONTENTS` lives in the CONTENTS wiki repo that `gitea.sh wiki-repo CONTENTS` resolves (`JSC_WIKI_REPO_CONTENTS`, then `JSC_WIKI_REPO`), which is **not** the LOG repo of step 1 and never falls back to it. Because the two pages sit in different wikis, the row's link to the log page is the absolute URL from `gitea.sh wiki-url <LOG repo> LOG_{HASH}` — `[[LOG_{HASH}]]` resolves only inside one wiki and would dead-link from here. `wiki-url` exit 4 means the step 5 write has not landed yet, so stop and rerun step 5 before this one; exit 5 means the page carries no `html_url`, so stop and report it and never assemble the URL by hand; exit 7 or 8 means the token or the API failed, so stop and report that status.
+   `LOG_CONTENTS` lives in the CONTENTS wiki repo that `gitea.sh wiki-repo CONTENTS` resolves (`JSC_WIKI_REPO_CONTENTS`, then `JSC_WIKI_REPO`), which is **not** the LOG repo of step 1 and never falls back to it. Because the two pages sit in different wikis, the row links the log page as `[LOG_{HASH}](<url>)`, with `<url>` from `gitea.sh wiki-url <LOG repo> LOG_{HASH}` — never the same-wiki `[[...]]` form, which resolves inside one wiki only and dead-links from here without reporting an error. `wiki-url` exit 4 means the step 5 write has not landed yet, so stop and rerun step 5 before this one; exit 5 means the page carries no `html_url`, so stop and report it and never assemble the URL by hand; exit 7 or 8 means the token or the API failed, so stop and report that status.
+
+   Put that URL through the step 5 `link-check.sh` gate before the upsert, with the same exit branches: only exit 0 writes the row, and a DEAD line stops the write and goes to step 7 as a failure.
 
    Build one file holding the single row from `templates/log-contents.md` — the absolute link, the bare `{HASH}` of step 1, this week's Friday date from step 2, the entry count and the update time — then run:
 

templates/learn-contents.md

@@ -2,7 +2,8 @@
 
 > 由 `jsc-log:learn` 維護。這是教訓目錄頁 `LEARN_CONTENTS`。每個存取庫一列；`LEARN_{HASH}` 的 `{HASH}` 依共用 wiki hash 規則產生：取 `{owner}/{repo}` 的完整 SHA-1 四十碼，a-f 轉大寫，不截短、不加前綴。
 > 本頁落在 `JSC_WIKI_REPO_CONTENTS` 解出的專用存取庫，教訓頁 `LEARN_{HASH}` 落在 `JSC_WIKI_REPO_LEARN` 的存取庫，兩者分屬不同 wiki。
-> 所以指向教訓頁的連結一律用 `gitea.sh wiki-url` 取得的絕對網址。`[[LEARN_{HASH}]]` 只在同一個 wiki 內解得開，寫在這裡就是死連結。
+> 連結寫法：教訓紀錄那一欄寫成 `[{頁名}]({絕對網址})`，網址取自 `jsc-gitea/tools/gitea.sh wiki-url`，不用 `[[...]]`。兩頁分屬不同存取庫，`[[...]]` 連不過去，畫面上還看不出壞掉。
+> 連結驗證：每個要寫進本頁的連結先交給 `jsc-gitea/tools/link-check.sh`，退出碼 0 才寫入。出現 DEAD 就不寫，把連不到的那幾筆回報給呼叫端。
 > 寫入語意：一列代表一個存取庫。一律用 `jsc-gitea/tools/wiki-contents.sh upsert LEARN 1 {owner}/{repo} {列檔} {本範本}` 單列整頁寫回——它讀整頁、找得到該存取庫既有的那一列就換掉那一列，找不到才附加。
 > 鍵取第 1 欄的存取庫名稱，不取第 2 欄的連結。存取庫名稱每一輪都一樣，連結會隨主機名與頁名編碼變動。
 > 禁止整頁覆蓋，也不得改動別人的列。

templates/log-contents.md

@@ -2,7 +2,8 @@
 
 > 由 `jsc-log:worklog` 維護。每個日誌頁一列；`{HASH}` 是 `{owner}/{repo}` 的完整 40 碼大寫十六進位 SHA-1，頁內仍依該週五日期整理。
 > 本頁落在 `JSC_WIKI_REPO_CONTENTS` 解出的專用存取庫，日誌頁 `LOG_{HASH}` 落在 `JSC_WIKI_REPO_LOG` 的存取庫，兩者分屬不同 wiki。
-> 所以指向日誌頁的連結一律用 `gitea.sh wiki-url` 取得的絕對網址。`[[LOG_{HASH}]]` 只在同一個 wiki 內解得開，寫在這裡就是死連結。
+> 連結寫法：日誌頁那一欄寫成 `[{頁名}]({絕對網址})`，網址取自 `jsc-gitea/tools/gitea.sh wiki-url`，不用 `[[...]]`。兩頁分屬不同存取庫，`[[...]]` 連不過去，畫面上還看不出壞掉。
+> 連結驗證：每個要寫進本頁的連結先交給 `jsc-gitea/tools/link-check.sh`，退出碼 0 才寫入。出現 DEAD 就不寫，把連不到的那幾筆回報給呼叫端。
 > 寫入語意：一列代表一個日誌頁。一律用 `jsc-gitea/tools/wiki-contents.sh upsert LOG 2 {HASH} {列檔} {本範本}` 單列整頁寫回——它讀整頁、找得到該頁既有的那一列就換掉那一列，找不到才附加。
 > 鍵取第 2 欄的裸 HASH，不取第 1 欄的連結。第 1 欄的連結帶著主機名與存取庫名：`GITEA_HOST` 一換、`JSC_WIKI_REPO_LOG` 改指別的存取庫，或 Gitea 對頁名的編碼有差，連結就跟上一輪寫的不一樣，鍵比不到就走附加，同一個日誌頁多出第二列，舊列從此不再更新。裸 HASH 只跟 `{owner}/{repo}` 有關，這三件事都動不到它。
 > 禁止整頁覆蓋，也不得改動別人的列。

templates/log-entry.md

@@ -4,7 +4,8 @@
 > 任務有三種：一個工作包、一輪 PR 留言修正、一個獨立的修正提交。同一個工作包跑五輪留言修正就是五筆，
 > 標題各自寫清楚是哪一輪，時間與 token 分開記。
 > `HASH` 依共享規則計算；頁名不再寫入年月週數，但頁內仍依本工作週的週五整理內容。
-> {PLAN 頁絕對網址}、{ANALYZE 頁絕對網址} 由 `jsc-gitea/tools/gitea.sh wiki-url` 取得——LOG 與 PLAN／ANALYZE 可能分屬不同存取庫，`[[頁名]]` 跨庫不通。
+> 連結寫法：計畫名稱、工作包編號、PR 目標分支三欄一律寫成 `[{文字}]({絕對網址})`，wiki 頁的網址由 `jsc-gitea/tools/gitea.sh wiki-url` 取得，不用 `[[...]]`。LOG 與 PLAN／ANALYZE 可能分屬不同存取庫，`[[...]]` 跨庫連不過去，畫面上還看不出壞掉。
+> 連結驗證：條目裡的每個連結先交給 `jsc-gitea/tools/link-check.sh`，退出碼 0 才寫入。出現 DEAD 就不寫，把連不到的那幾筆回報給呼叫端。
 > 解不出 wiki 存取庫或頁面不存在時，該欄填「無」，其他欄照填。由 `maintain` 觸發的日誌本來就沒有計畫頁與分析頁。
 
 ---

templates/report-contents.md

@@ -2,7 +2,8 @@
 
 > 由 `jsc-log:report` 維護。年、月、週、日各一頁；`HASH` 取 `{owner}/{repo}/{期間}` 的完整 40 碼大寫十六進位 SHA-1，算法與其他頁面共用，但這裡的 `{owner}/{repo}` 取 REPORT wiki 存取庫，不是程式碼存取庫。
 > 本頁落在 `JSC_WIKI_REPO_CONTENTS` 解出的專用存取庫，報表頁 `REPORT_{HASH}` 落在 `JSC_WIKI_REPO_REPORT` 的存取庫，兩者分屬不同 wiki。
-> 所以指向報表頁的連結一律用 `gitea.sh wiki-url` 取得的絕對網址。`[[REPORT_{HASH}]]` 只在同一個 wiki 內解得開，寫在這裡就是死連結。
+> 連結寫法：報表頁那一欄寫成 `[{頁名}]({絕對網址})`，網址取自 `jsc-gitea/tools/gitea.sh wiki-url`，不用 `[[...]]`。兩頁分屬不同存取庫，`[[...]]` 連不過去，畫面上還看不出壞掉。
+> 連結驗證：每個要寫進本頁的連結先交給 `jsc-gitea/tools/link-check.sh`，退出碼 0 才寫入。出現 DEAD 就不寫，把連不到的那幾筆回報給呼叫端。
 > 寫入語意：一列代表一個報表頁，也就是一個存取庫的一種期間。一律用 `jsc-gitea/tools/wiki-contents.sh upsert REPORT 2 {HASH} {列檔} {本範本}` 單列整頁寫回——它讀整頁、找得到該報表頁既有的那一列就換掉那一列，找不到才附加。
 > 鍵取第 2 欄的裸 HASH，不取第 1 欄的連結。第 1 欄的連結帶著主機名與頁名編碼，`GITEA_HOST` 一換或 Gitea 對頁名的編碼有差，連結就跟上一輪寫的不一樣，鍵比不到就走附加，同一個報表頁多出第二列，舊列從此不再更新。裸 HASH 不受這兩件事影響。
 > 但 `JSC_WIKI_REPO_REPORT` 改指別的存取庫是另一回事：HASH 本身就取自 REPORT wiki 存取庫，換庫等於換頁，四個期間會各多一列。那是換庫的本意，不是鍵失準，舊列請人工清掉。