[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6DSSQVR3DSA6GJKMNMM1ZSK":3,"skill-same:01M6DSSQVR3DSA6GJKMNMM1ZSK":60},{"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":32,"linkOnly":33,"markdownContent":34,"markdownBody":35,"frontMatter":36,"refName":37,"commitSha":38,"skillDirectory":39,"forksCount":40,"repositoryHtmlUrl":41,"repositoryHomepageUrl":42,"ownerHtmlUrl":43,"ownerBlogUrl":44,"ownerDescription":6,"updatedAt":45,"install":46,"resources":59},"01M6DSSQVR3DSA6GJKMNMM1ZSK","debrief",null,"Capture a structured session summary and publish it to the OACP org-memory debrief store as an immutable record. Run at the end of any session.","kiloloop\u002Foacp-skills","kiloloop","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F264514437?v=4","01M6DSS1K2JKZD3YRMXRH54968","Kiloloop","01M6DSS1SMM810F7327W2YX7ZZ",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Fskills\u002Fdebrief\u002Fcodex\u002FSKILL.md",[],[5,19,20,21,22,23,24,25,26,27,28,29,30,31],"memory","session summary","run","publish","immutable","capture","session","summary","record","structured","end","as","oacp","2026-11-10 10:12:30",false,"---\nname: debrief\ndescription: Capture a structured session summary and publish it to the OACP org-memory debrief store as an immutable record. Run at the end of any session.\n---\n\n# \u002Fdebrief — Codex Session Debrief\n\nPublish a structured summary of this session to the central org-memory debrief store:\n\n```text\n$MEMORY_HOME\u002Forg-memory\u002Fdebriefs\u002F\u003Cproject>\u002F\u003CYYYY>\u002F\u003CMM>\u002F\u003CYYYYMMDD>-\u003Cagent>-\u003Csession>.md\n```\n\n`MEMORY_HOME` is the memory home resolved in step 1.\n\nOne session produces exactly one file, written once at session end and never\nrewritten. The store is the full-fidelity session record; curated outcomes\nreach other agents through `agent-memory event write` instead (see \"Curation guard\"\nbelow).\n\n## Arguments\n\n- `\u002Fdebrief` — generate the summary and publish it\n- `\u002Fdebrief --dry-run` — generate and display the summary without writing\n  anything\n\nThe shared writer needs Python 3.9+ and the standard library. Binding status,\ncurated events, and store diagnostics use `agent-memory-cli` (0.1.1+ for\n`event write`); report a missing capability before that operation. The writer\nitself has no OACP CLI dependency.\n\n## Instructions\n\n### 1. Identify the project\n\nResolve the project name with this fallback chain:\n\n1. The `project_name` field in the workspace marker. The marker is gitignored\n   and lives only at the repository's main root, so resolve that root first:\n\n   ```bash\n   ROOT=$(dirname \"$(git rev-parse --path-format=absolute --git-common-dir)\")\n   python3 -c \"import json,sys; print(json.load(open(sys.argv[1]))['project_name'])\" \"$ROOT\u002F.oacp\"\n   ```\n\n2. `basename \"$ROOT\"` — not `git rev-parse --show-toplevel`, which returns\n   the linked-worktree directory when the session is running inside one\n3. The working directory basename, when not in a git repo\n\nThe name must match the workspace directory under `$OACP_HOME\u002Fprojects\u002F`\nbyte-for-byte — the store path segment is case-sensitive. The writer checks\nsegment syntax, not workspace existence: confirm a missing workspace mapping\nbefore publishing unless a repository memory binding establishes the project.\n\nIf the main repository has an `.agent-memory.json` binding, resolve its project\nand home before publication. An ambient `AGENT_MEMORY_HOME` or `OACP_HOME` can\noverride that binding in the CLI; inspect it with\n`env -u AGENT_MEMORY_HOME -u OACP_HOME agent-memory status`, or read the exact\nbinding file and resolve its home. The binding supplies the project without a\nnew confirmation. Do not infer this project from the task worktree's name.\n\nSet `MEMORY_HOME` once to the binding's resolved home, otherwise to\n`${OACP_HOME:-$HOME\u002Foacp}`. Use this same value for ruling lookup, publication,\nevent curation, optional initialization, and doctor. The bound home must be\nwritable before publication; retain the draft if it is unavailable.\n\n### 2. Collect the session identity\n\nThe writer needs five identity values. Gather them before composing the body:\n\n- **`--project`** — from step 1\n- **`--agent codex`** — this runtime's registered project-agent name\n- **`--runtime codex`** — the runtime family writing the record\n- **`--session`** — a short stable session id: **1-32 lowercase letters and\n  digits, never hyphens**. Use the first 8 characters of the harness session\n  UUID with hyphens stripped when Codex exposes it. Otherwise generate one\n  8-character lowercase hexadecimal id once for this run and retain it for\n  every retry. This is what distinguishes multiple sessions by the same agent\n  on the same day. If the session keeps working after a debrief has already\n  landed, use the derived continuation id described in step 8 instead.\n- **`--started-utc` \u002F `--ended-utc`** — session bounds as ISO 8601 UTC with\n  whole seconds, e.g.\n  `2026-08-25T20:04:11Z`. Take these from the clock, never from an estimate;\n  `ended_utc` must not precede `started_utc`, and the date of `started_utc`\n  determines the filename and directory segments.\n\n> **UTC, not local time.** The store path and the `\u003CYYYYMMDD>` filename prefix\n> are derived from `started_utc` in UTC. On a machine whose OS timezone is\n> ahead of or behind UTC, a locally-stamped date lands the record in the wrong\n> month directory. Generate both stamps with\n> `date -u +%Y-%m-%dT%H:%M:%SZ`.\n\n### 3. Review the session\n\nGather what actually happened:\n\n- **Accomplishments** — what was built, fixed, reviewed, or decided\n- **Decisions** — key choices and their rationale\n- **Issues** — tickets created, closed, or updated. If other agents\n  collaborated (parallel runs, review loops), include their closures and PR\n  activity; local `git log` alone misses them.\n- **Files changed** — `git log --name-only --format= --since=\"\u003Cstarted_utc>\"`\n  plus `git status --short`\n- **Git activity** — `git log --oneline --since=\"\u003Cstarted_utc>\"`, in each repo\n  touched for cross-project sessions\n- **Next actions** — carry-forward items, open threads, and blockers that\n  remain live. Before restating a plan, check today's entries in\n  `$MEMORY_HOME\u002Forg-memory\u002Fevents\u002F` and `$MEMORY_HOME\u002Forg-memory\u002Fdecisions.md`\n  for a ruling that already supersedes it; the published record is immutable.\n\nYour conversation context is the primary source. Git activity supplements it;\nit does not replace what you remember of the session.\n\n### 4. Compose the body\n\nWrite the body to a scratch file, then hand that file to the writer. The body\nis everything after the frontmatter — the writer generates the frontmatter\nitself, so do not write a `---` block at the top.\n\n```markdown\n# Session Summary: \u003Cproject> — YYYY-MM-DD\n\n**Runtime**: codex\n**Duration**: ~{{DURATION}} minutes\n\n## What Changed\n\n- bullet points of accomplishments\n\n## Decisions Made\n\n- key decisions with rationale\n\n## Tickets \u002F Issues\n\n- created\u002Fclosed\u002Fupdated references (or \"None\")\n\n## Files Modified\n\n- key files changed\n\n## Next Actions\n\n- [ ] carry-forward items\n\n## Notes\n\n- anything notable (omit the section if nothing)\n```\n\nLeave `{{DURATION}}` literal. The shared writer fills it from the start\u002Fend\nstamps, rounding partial minutes up before hashing. Other placeholders remain\nunchanged.\n\n> **Use a file-writing tool, never a shell heredoc.** Shell expansion corrupts\n> debrief bodies — backticks in a double-quoted heredoc execute as command\n> substitution, and the mangling is silent. Write the body with your runtime's\n> file-writing tool and pass the path.\n\n### 5. Resolve the shared writer\n\nThe catalog has one writer at `skills\u002Fdebrief\u002Fscripts\u002Fwrite_debrief.py`; the\nCodex wrapper does not carry a fork. Resolve the writer from the loaded skill\ndirectory, never from the current project's working directory.\n\n```bash\n# Prefer an installed package, where SKILL.md and scripts\u002F share one directory.\nSKILL_DIR=\"\u003Cdirectory containing the loaded debrief SKILL.md>\"\n\n# In a catalog checkout, codex\u002FSKILL.md sits one level below the shared writer.\nif [ ! -f \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" ] && \\\n   [ -f \"$SKILL_DIR\u002F..\u002Fscripts\u002Fwrite_debrief.py\" ]; then\n  SKILL_DIR=\"$(cd \"$SKILL_DIR\u002F..\" && pwd)\"\nfi\n\n# Try the common project and global Codex install locations.\nif [ ! -f \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" ]; then\n  for d in \"$PWD\u002F.agents\u002Fskills\u002Fdebrief\" \\\n           \"${CODEX_HOME:-$HOME\u002F.codex}\u002Fskills\u002Fdebrief\"; do\n    [ -f \"$d\u002Fscripts\u002Fwrite_debrief.py\" ] && SKILL_DIR=\"$d\" && break\n  done\nfi\n\ntest -f \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" \\\n  || { echo \"debrief writer not found; re-install the skill\" >&2; exit 1; }\n```\n\nStop here if the writer cannot be resolved. Do not hand-compose a record into\nthe store — the frontmatter, the content hash, and the publication primitive\nare the writer's contract, and a hand-written file satisfies none of them.\n\n### 6. Dry run\n\n`\u002Fdebrief --dry-run` stops here. Run the step 7 command with `--dry-run`\nappended: the writer validates the identity and body, composes the exact\nrecord it would publish, prints it, and touches nothing — no directories\ncreated, no files written. Status is reported as `dry-run`.\n\nShow the composed record and the target path, then **stop**. Do not continue\nto step 7, and do not report the session as debriefed.\n\n### 7. Publish\n\n```bash\npython3 \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" \\\n  --project \"\u003Cproject>\" \\\n  --agent codex \\\n  --runtime codex \\\n  --session \"\u003Csession>\" \\\n  --started-utc \"\u003Cstarted>\" \\\n  --ended-utc \"\u003Cended>\" \\\n  --body-file \"\u003Cpath-to-body>\" \\\n  --oacp-dir \"${MEMORY_HOME:?resolve it in step 1}\" \\\n  --json\n```\n\nThe writer validates the schema, fills `{{DURATION}}`, computes the content\nhash, stages privately, publishes with an atomic no-replace link, and reads back to\nconfirm. Exit codes: `0` published (or an idempotent re-publish of a\nbyte-identical record), `1` validation error, `2` publication failure.\n\nPublication is failure-atomic — the canonical path only ever holds a complete,\nverified record. On any failure the path stays absent; at worst a private\n`.stage.*` file remains, and a later successful publication of the same record\nsweeps it. The writer only ever publishes an inode it created itself, so a\n`.stage.*` file that appears from anywhere else is never adopted —\n`agent-memory doctor --home \"${MEMORY_HOME:?resolve it in step 1}\"` reports it\nand it is safe to delete by hand.\n\n### 8. Handle a collision\n\n`--session` collisions are the one failure that needs a judgment call:\n\n- **Byte-identical record already published** → reported as `idempotent`.\n  Nothing to do; the run already landed.\n- **Different record at the same path** → hard failure. A published debrief is\n  never replaced. Re-publish under a **new session identifier**; the existing\n  record stays as it is.\n\nCorrections and follow-ups are new artifacts too — a later debrief, or an\nevent referencing the original — never an edit to a landed file.\n\n> **A session that keeps working after its own debrief.** Publish a\n> continuation segment under the same project, agent, and runtime: suffix the\n> session id with the segment number (`b31ffdea` → `b31ffdea2`, then\n> `b31ffdea3`), set `started_utc` to the previous record's `ended_utc`, and set\n> `ended_utc` to now. The suffixed form stays inside the 1-32 lowercase\n> alphanumeric grammar. Open the body by naming the parent record and the span\n> it covers; never backfill the landed record or skip the continuation.\n\n### 9. Confirm\n\nPrint the canonical path, the status, and the content hash from the writer's\nJSON output.\n\n## Curation guard\n\nRaw debriefs never enter `recent.md` or `events\u002F`. Only curated folds do. If\nthe session produced an outcome other agents or projects need, write it as a\nseparate event:\n\n```bash\nagent-memory event write --home \"${MEMORY_HOME:?resolve it in step 1}\" \\\n  --type decision --project \"\u003Cproject>\" --agent codex --slug \"\u003Cslug>\" \\\n  --source-ref \"\u003CYYYYMMDD>-codex-\u003Csession>\" --body-file \"\u003Cpath-to-event-body>\"\n```\n\nSet the event's `source_ref` to the debrief filename stem\n(`\u003CYYYYMMDD>-\u003Cagent>-\u003Csession>`) so the two can be reconciled later.\n\n## Notes\n\n- Keep each section to the facts a later reader needs. If the session was\n  trivial, say so rather than padding.\n- Facts over narrative: the store is read by curation tooling, not for prose.\n- Debrief files are never auto-loaded at session start; they are permanent\n  history, not context.\n- The store does **not** have to exist first — the writer provisions its own\n  tree. `agent-memory org init --home \"${MEMORY_HOME:?resolve it in step 1}\"`\n  is optional setup.\n- `agent-memory doctor --home \"${MEMORY_HOME:?resolve it in step 1}\"` validates\n  store *setup* — layout, staging artifacts, symlinks.\n  It never opens debrief files; schema and hash correctness are the writer's\n  contract, verified at publication.\n","# \u002Fdebrief — Codex Session Debrief\n\nPublish a structured summary of this session to the central org-memory debrief store:\n\n```text\n$MEMORY_HOME\u002Forg-memory\u002Fdebriefs\u002F\u003Cproject>\u002F\u003CYYYY>\u002F\u003CMM>\u002F\u003CYYYYMMDD>-\u003Cagent>-\u003Csession>.md\n```\n\n`MEMORY_HOME` is the memory home resolved in step 1.\n\nOne session produces exactly one file, written once at session end and never\nrewritten. The store is the full-fidelity session record; curated outcomes\nreach other agents through `agent-memory event write` instead (see \"Curation guard\"\nbelow).\n\n## Arguments\n\n- `\u002Fdebrief` — generate the summary and publish it\n- `\u002Fdebrief --dry-run` — generate and display the summary without writing\n  anything\n\nThe shared writer needs Python 3.9+ and the standard library. Binding status,\ncurated events, and store diagnostics use `agent-memory-cli` (0.1.1+ for\n`event write`); report a missing capability before that operation. The writer\nitself has no OACP CLI dependency.\n\n## Instructions\n\n### 1. Identify the project\n\nResolve the project name with this fallback chain:\n\n1. The `project_name` field in the workspace marker. The marker is gitignored\n   and lives only at the repository's main root, so resolve that root first:\n\n   ```bash\n   ROOT=$(dirname \"$(git rev-parse --path-format=absolute --git-common-dir)\")\n   python3 -c \"import json,sys; print(json.load(open(sys.argv[1]))['project_name'])\" \"$ROOT\u002F.oacp\"\n   ```\n\n2. `basename \"$ROOT\"` — not `git rev-parse --show-toplevel`, which returns\n   the linked-worktree directory when the session is running inside one\n3. The working directory basename, when not in a git repo\n\nThe name must match the workspace directory under `$OACP_HOME\u002Fprojects\u002F`\nbyte-for-byte — the store path segment is case-sensitive. The writer checks\nsegment syntax, not workspace existence: confirm a missing workspace mapping\nbefore publishing unless a repository memory binding establishes the project.\n\nIf the main repository has an `.agent-memory.json` binding, resolve its project\nand home before publication. An ambient `AGENT_MEMORY_HOME` or `OACP_HOME` can\noverride that binding in the CLI; inspect it with\n`env -u AGENT_MEMORY_HOME -u OACP_HOME agent-memory status`, or read the exact\nbinding file and resolve its home. The binding supplies the project without a\nnew confirmation. Do not infer this project from the task worktree's name.\n\nSet `MEMORY_HOME` once to the binding's resolved home, otherwise to\n`${OACP_HOME:-$HOME\u002Foacp}`. Use this same value for ruling lookup, publication,\nevent curation, optional initialization, and doctor. The bound home must be\nwritable before publication; retain the draft if it is unavailable.\n\n### 2. Collect the session identity\n\nThe writer needs five identity values. Gather them before composing the body:\n\n- **`--project`** — from step 1\n- **`--agent codex`** — this runtime's registered project-agent name\n- **`--runtime codex`** — the runtime family writing the record\n- **`--session`** — a short stable session id: **1-32 lowercase letters and\n  digits, never hyphens**. Use the first 8 characters of the harness session\n  UUID with hyphens stripped when Codex exposes it. Otherwise generate one\n  8-character lowercase hexadecimal id once for this run and retain it for\n  every retry. This is what distinguishes multiple sessions by the same agent\n  on the same day. If the session keeps working after a debrief has already\n  landed, use the derived continuation id described in step 8 instead.\n- **`--started-utc` \u002F `--ended-utc`** — session bounds as ISO 8601 UTC with\n  whole seconds, e.g.\n  `2026-08-25T20:04:11Z`. Take these from the clock, never from an estimate;\n  `ended_utc` must not precede `started_utc`, and the date of `started_utc`\n  determines the filename and directory segments.\n\n> **UTC, not local time.** The store path and the `\u003CYYYYMMDD>` filename prefix\n> are derived from `started_utc` in UTC. On a machine whose OS timezone is\n> ahead of or behind UTC, a locally-stamped date lands the record in the wrong\n> month directory. Generate both stamps with\n> `date -u +%Y-%m-%dT%H:%M:%SZ`.\n\n### 3. Review the session\n\nGather what actually happened:\n\n- **Accomplishments** — what was built, fixed, reviewed, or decided\n- **Decisions** — key choices and their rationale\n- **Issues** — tickets created, closed, or updated. If other agents\n  collaborated (parallel runs, review loops), include their closures and PR\n  activity; local `git log` alone misses them.\n- **Files changed** — `git log --name-only --format= --since=\"\u003Cstarted_utc>\"`\n  plus `git status --short`\n- **Git activity** — `git log --oneline --since=\"\u003Cstarted_utc>\"`, in each repo\n  touched for cross-project sessions\n- **Next actions** — carry-forward items, open threads, and blockers that\n  remain live. Before restating a plan, check today's entries in\n  `$MEMORY_HOME\u002Forg-memory\u002Fevents\u002F` and `$MEMORY_HOME\u002Forg-memory\u002Fdecisions.md`\n  for a ruling that already supersedes it; the published record is immutable.\n\nYour conversation context is the primary source. Git activity supplements it;\nit does not replace what you remember of the session.\n\n### 4. Compose the body\n\nWrite the body to a scratch file, then hand that file to the writer. The body\nis everything after the frontmatter — the writer generates the frontmatter\nitself, so do not write a `---` block at the top.\n\n```markdown\n# Session Summary: \u003Cproject> — YYYY-MM-DD\n\n**Runtime**: codex\n**Duration**: ~{{DURATION}} minutes\n\n## What Changed\n\n- bullet points of accomplishments\n\n## Decisions Made\n\n- key decisions with rationale\n\n## Tickets \u002F Issues\n\n- created\u002Fclosed\u002Fupdated references (or \"None\")\n\n## Files Modified\n\n- key files changed\n\n## Next Actions\n\n- [ ] carry-forward items\n\n## Notes\n\n- anything notable (omit the section if nothing)\n```\n\nLeave `{{DURATION}}` literal. The shared writer fills it from the start\u002Fend\nstamps, rounding partial minutes up before hashing. Other placeholders remain\nunchanged.\n\n> **Use a file-writing tool, never a shell heredoc.** Shell expansion corrupts\n> debrief bodies — backticks in a double-quoted heredoc execute as command\n> substitution, and the mangling is silent. Write the body with your runtime's\n> file-writing tool and pass the path.\n\n### 5. Resolve the shared writer\n\nThe catalog has one writer at `skills\u002Fdebrief\u002Fscripts\u002Fwrite_debrief.py`; the\nCodex wrapper does not carry a fork. Resolve the writer from the loaded skill\ndirectory, never from the current project's working directory.\n\n```bash\n# Prefer an installed package, where SKILL.md and scripts\u002F share one directory.\nSKILL_DIR=\"\u003Cdirectory containing the loaded debrief SKILL.md>\"\n\n# In a catalog checkout, codex\u002FSKILL.md sits one level below the shared writer.\nif [ ! -f \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" ] && \\\n   [ -f \"$SKILL_DIR\u002F..\u002Fscripts\u002Fwrite_debrief.py\" ]; then\n  SKILL_DIR=\"$(cd \"$SKILL_DIR\u002F..\" && pwd)\"\nfi\n\n# Try the common project and global Codex install locations.\nif [ ! -f \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" ]; then\n  for d in \"$PWD\u002F.agents\u002Fskills\u002Fdebrief\" \\\n           \"${CODEX_HOME:-$HOME\u002F.codex}\u002Fskills\u002Fdebrief\"; do\n    [ -f \"$d\u002Fscripts\u002Fwrite_debrief.py\" ] && SKILL_DIR=\"$d\" && break\n  done\nfi\n\ntest -f \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" \\\n  || { echo \"debrief writer not found; re-install the skill\" >&2; exit 1; }\n```\n\nStop here if the writer cannot be resolved. Do not hand-compose a record into\nthe store — the frontmatter, the content hash, and the publication primitive\nare the writer's contract, and a hand-written file satisfies none of them.\n\n### 6. Dry run\n\n`\u002Fdebrief --dry-run` stops here. Run the step 7 command with `--dry-run`\nappended: the writer validates the identity and body, composes the exact\nrecord it would publish, prints it, and touches nothing — no directories\ncreated, no files written. Status is reported as `dry-run`.\n\nShow the composed record and the target path, then **stop**. Do not continue\nto step 7, and do not report the session as debriefed.\n\n### 7. Publish\n\n```bash\npython3 \"$SKILL_DIR\u002Fscripts\u002Fwrite_debrief.py\" \\\n  --project \"\u003Cproject>\" \\\n  --agent codex \\\n  --runtime codex \\\n  --session \"\u003Csession>\" \\\n  --started-utc \"\u003Cstarted>\" \\\n  --ended-utc \"\u003Cended>\" \\\n  --body-file \"\u003Cpath-to-body>\" \\\n  --oacp-dir \"${MEMORY_HOME:?resolve it in step 1}\" \\\n  --json\n```\n\nThe writer validates the schema, fills `{{DURATION}}`, computes the content\nhash, stages privately, publishes with an atomic no-replace link, and reads back to\nconfirm. Exit codes: `0` published (or an idempotent re-publish of a\nbyte-identical record), `1` validation error, `2` publication failure.\n\nPublication is failure-atomic — the canonical path only ever holds a complete,\nverified record. On any failure the path stays absent; at worst a private\n`.stage.*` file remains, and a later successful publication of the same record\nsweeps it. The writer only ever publishes an inode it created itself, so a\n`.stage.*` file that appears from anywhere else is never adopted —\n`agent-memory doctor --home \"${MEMORY_HOME:?resolve it in step 1}\"` reports it\nand it is safe to delete by hand.\n\n### 8. Handle a collision\n\n`--session` collisions are the one failure that needs a judgment call:\n\n- **Byte-identical record already published** → reported as `idempotent`.\n  Nothing to do; the run already landed.\n- **Different record at the same path** → hard failure. A published debrief is\n  never replaced. Re-publish under a **new session identifier**; the existing\n  record stays as it is.\n\nCorrections and follow-ups are new artifacts too — a later debrief, or an\nevent referencing the original — never an edit to a landed file.\n\n> **A session that keeps working after its own debrief.** Publish a\n> continuation segment under the same project, agent, and runtime: suffix the\n> session id with the segment number (`b31ffdea` → `b31ffdea2`, then\n> `b31ffdea3`), set `started_utc` to the previous record's `ended_utc`, and set\n> `ended_utc` to now. The suffixed form stays inside the 1-32 lowercase\n> alphanumeric grammar. Open the body by naming the parent record and the span\n> it covers; never backfill the landed record or skip the continuation.\n\n### 9. Confirm\n\nPrint the canonical path, the status, and the content hash from the writer's\nJSON output.\n\n## Curation guard\n\nRaw debriefs never enter `recent.md` or `events\u002F`. Only curated folds do. If\nthe session produced an outcome other agents or projects need, write it as a\nseparate event:\n\n```bash\nagent-memory event write --home \"${MEMORY_HOME:?resolve it in step 1}\" \\\n  --type decision --project \"\u003Cproject>\" --agent codex --slug \"\u003Cslug>\" \\\n  --source-ref \"\u003CYYYYMMDD>-codex-\u003Csession>\" --body-file \"\u003Cpath-to-event-body>\"\n```\n\nSet the event's `source_ref` to the debrief filename stem\n(`\u003CYYYYMMDD>-\u003Cagent>-\u003Csession>`) so the two can be reconciled later.\n\n## Notes\n\n- Keep each section to the facts a later reader needs. If the session was\n  trivial, say so rather than padding.\n- Facts over narrative: the store is read by curation tooling, not for prose.\n- Debrief files are never auto-loaded at session start; they are permanent\n  history, not context.\n- The store does **not** have to exist first — the writer provisions its own\n  tree. `agent-memory org init --home \"${MEMORY_HOME:?resolve it in step 1}\"`\n  is optional setup.\n- `agent-memory doctor --home \"${MEMORY_HOME:?resolve it in step 1}\"` validates\n  store *setup* — layout, staging artifacts, symlinks.\n  It never opens debrief files; schema and hash correctness are the writer's\n  contract, verified at publication.\n",{"name":5,"description":7},"main","ea78d5cce0208a15b59f6f2c49e729ba71a3c5af","skills\u002Fdebrief\u002Fcodex",1,"https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills","https:\u002F\u002Foacp.dev\u002Fskills\u002F","https:\u002F\u002Fgithub.com\u002Fkiloloop","https:\u002F\u002Fkiloloop.com","2026-10-31 14:39:52",{"source":8,"skillName":5,"commands":47,"prompt":57,"zipAvailable":58},[48,51,54],{"tool":49,"command":50},"npx","npx skills add kiloloop\u002Foacp-skills --skill debrief",{"tool":52,"command":53},"bunx","bunx skills add kiloloop\u002Foacp-skills --skill debrief",{"tool":55,"command":56},"pnpm","pnpm dlx skills add kiloloop\u002Foacp-skills --skill debrief","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Fskills\u002Fdebrief\u002Fcodex\u002FSKILL.md",true,[],[61,72,77,98,104,113],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":32},"01M6DSWR7WDRG7C3D31R4QKW5T","\u003Cskill-name>","\u003Cone-line description>","https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Ftemplate\u002Fcodex\u002FSKILL.md",[],[68,69,70,71],"skill name","name","\u003Cskill name>","line",{"skillNo":73,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":32},"01M6DSWMRPCHQH4HJJF3MATTXZ","https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Ftemplate\u002Fclaude\u002FSKILL.md",[],[68,69,70,71],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":32},"01M6DSWH83XKNGJSKX2XWQCRKJ","wrap-up","End-of-session cleanup for Codex — inspect stale artifacts, run debrief, write significant org-memory events, run a delta-first self-improve pass, then publish bounded approved memory and repository changes.","https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwrap-up\u002Fcodex\u002FSKILL.md",[],[84,85,86,21,19,87,88,89,22,90,91,92,25,93,94,95,96,29,5,97],"wrap up","up","wrap","cleanup","artifacts","repository","first","improve","write","stale","session cleanup","delta","self","codex",{"skillNo":99,"name":100,"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":101,"categories":102,"keywords":103,"publishedAt":32},"01M6DSWEQBS2CE6F2H8T3VQNZY","claude","https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwrap-up\u002Fclaude\u002FSKILL.md",[],[],{"skillNo":105,"name":79,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":32},"01M6DSW99EPXDVAG8GHR785CC5","End-of-session cleanup, debrief, org-memory events, self-improve, commit, and push in one command.","https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Fskills\u002Fwrap-up\u002FSKILL.md",[],[84,85,86,96,19,110,87,111,91,112,25,94,29,5],"commit","push","command",{"skillNo":114,"name":115,"nameZh":6,"description":116,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":117,"categories":118,"keywords":119,"publishedAt":32},"01M6DSW4FRNRKRQ0TRMZAYNMKZ","self-improve","Audit and improve Codex operating guidance: skills, curated project memory, AGENTS.md, and relevant configuration. Use for explicit self-improve requests, routine wrap-up drift checks, stale or conflicting guidance, or an approved cleanup. Route release-wide OACP migrations and context-budget audits to audit-oacp-skills. Do not hand-edit generated Codex memories or installed plugin\u002Fsystem artifacts.","https:\u002F\u002Fgithub.com\u002Fkiloloop\u002Foacp-skills\u002Fblob\u002Fmain\u002Fskills\u002Fself-improve\u002Fcodex\u002FSKILL.md",[],[120,96,91,121,87,122,123,124,125,126,127,128,85,129,130,131,132,133,31,19],"self improve","project","release","audit","Budget","system","edit","context","generated","do","route","drift","memories","operating"]