end
Triage-first session close-out COORDINATOR for any project on this machine.
Skill by julianhernandez2155
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/julianhernandez2155-end
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/end
What's inside
- Skill name
end- Description
- Triage-first session close-out COORDINATOR for any project on this machine. Use for `/end` or a generic close-out request ("close out the session", "end the session", "I'm done for now", "let's close out") when the user wants the current session triaged and routed. First decides whether the session is worth capturing — a true scratch session (a quick question with no decision, discovery, or config/tool-surface change) is closed clean with nothing manufactured. If substantive, it DELEGATES by invoking other skills: `decision-log` for STATUS/daily-log/ADR work and `session-handoff` (`/wrap`) for Workspace knowledge capture, rather than re-writing their output itself; cross-session facts go to memory. Offers (never forces) to bootstrap STATUS/log scaffolding if a project lacks it. DRAFT-FIRST — shows everything before writing, never commits to git.
- License
- MIT
- Instructions
- 129 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# end (/end) The evening half of the daily lifecycle pair (mirror of `/catchup`). The single front door for closing a working session. **It triages, then delegates — it does not reimplement** what `decision-log` and `session-handoff` already do. That delegation is the whole point: one skill to remember, no duplicated routing logic, no overlap drift. ## Step 1 — TRIAGE: is this worth routing at all? Judge by what was *learned or changed* this session, not just by whether files moved. **Scratch session** — close clean only when ALL of these hold: a quick question or explanation, with no decision, no discovery/gotcha, no pivot, no config/tool-surface change, and no file work you did this session. → Say plainly: **"Scratch session — nothing worth routing. Closing clean."** Optionally offer *one* line: "Want a single-line log entry anyway? (y/n)". Then **stop** — no delegation, no manufactured insights. **Substantive — route it (Step 2)** if ANY of these hold, *even with zero file edits*: - A decision, pivot, or discovery/gotcha the user flagged. - A **config / tool-surface / security change** — installed or configured an MCP, changed a setting or permission, set up a token or auth scope. These persist beyond the session, so they must leave at least a one-line environment breadcrumb (in the daily log or a memory entry) — **with any secret/token value redacted: refer to it by name only (e.g. "the Firecrawl API key"), and never reprint the literal secret value anywhere, even while explaining that you're redacting it** (workspace no-secrets rule). A config change is the opposite of scratch precisely because it outlives the session. - A new cross-session fact, a multi-step build, or the user explicitly asks to capture. - Files you created/edited *this session* (≥1 meaningful change). **Do NOT count as substance:** pre-existing uncommitted/dirty state you didn't touch this session, or generated/transient artifacts (`__pycache__/`, build output, lockfile churn). Flag them as a housekeeping note if useful, but on their own they don't make a session substantive. ## Step 2 — Detect context, then route ONLY what applies Walk up from cwd to classify where you are. **First, reconcile git against STATUS — this is what stops drift at the source.** Before routing any STATUS/log work, in each scaffolded project touched this session: - Run `git log --oneline -10` + `git status --short`, and read STATUS.md's `_Last updated: YYYY-MM-DD_` line. - Run `git log --since="<that date>" --oneline` (and eyeball the date of `git log -1`). **If commits landed after STATUS's last-updated date — or commits this session moved past what STATUS's "What's next" describes — STATUS is behind and the delta MUST be captured.** This is the exact failure that breaks tomorrow's `/catchup`: work shipped in the same session as a decision, and STATUS still reads "about to start." - Don't rely only on what surfaced in conversation. A session that *executed* a plan often shipped more than was discussed — the commit subjects since the STATUS date are the ground truth for what actually changed. Feed that delta into the `decision-log` hand-off so STATUS's "What's next" reflects what's genuinely left, not what was already done. - If `git status` shows uncommitted content you touched this session, flag it in the close-out so it's not silently lost (the user commits; you don't). Then route each surfaced piece by delegating: | What surfaced this session | Route → delegate to | Only when | |---|---|---| | Project state changed (tasks done, milestone, new "next") | **`decision-log`** Update Mode → STATUS + daily log | always, for any project | | A decision worth recording | **`decision-log`** `/decide` (new ADR) | a deliberate trade-off was made | | A reversal of a prior decision | **`decision-log`** `/supersede NNNN` | this overrides an accepted ADR | | A cross-session fact (status, scope, contact changed) | **memory** entry under `~/.claude/projects/<encoded-cwd>/memory/` + MEMORY.md line | the fact will matter next session **and clears the specificity gate below** | | project-specific fact OR reusable Team-KB pattern | **`session-handoff`** (`/wrap`) | cwd is inside the SU Workspace (detect `ExampleApp/` or `Team Knowledge Base/` walking up) | | Where you stopped / next step | breadcrumb | always — make sure the log's **Next** line and STATUS **What's next** are set so tomorrow's `/catchup` reads cleanly | Rules that keep routing honest: - **Delegate by invoking, don't re-author.** For STATUS/log/ADR work, **invoke the `decision-log` skill** and let it produce and gate its own drafts — do NOT hand-write the ADR / STATUS / log markdown yourself inside `/end`. Re-authoring it here silently re-encodes decision-log's templates and drifts the moment that skill changes. Your job is the triage, the routing decision, and the hand-off — not the artifact. Same for Workspace knowledge: invoke `session-handoff`, don't reproduce its four-block output. - **Scope the `session-handoff` hand-off to knowledge only** — SU `learnings/`, Team-KB `wiki/`, and memory. STATUS / daily-log / ADRs are `decision-log`'s territory. If you let `session-handoff` *also* draft project STATUS/ADRs, you double-propose the same artifact. One owner per artifact. - Route **only** what applies to where you are. In a bare external code project, the SU-learnings / Team-KB routes simply don't fire — STATUS/log + memory + breadcrumb are the whole job. Don't offer Workspace routes outside the Workspace. - If multiple pieces surfaced, handle each separately — don't cram a pattern into a log entry or a decision into memory. - **Specificity gate before anything graduates to memory.** A memory entry must be a *specific, reusable* fact that will still be actionable cold next session — not a vague activity summary. Keep: "the Confluence export fails closed on inherited read-restrictions via `restrictions.py`." Reject: "worked on the export tool today" (that's a log line, not a memory). If the candidate doesn't name a concrete system/file/decision/contact and a reason it matters later, route it to the daily log instead and write nothing to memory. This keeps `MEMORY.md` high-signal. ## Step 3 — Bootstrap offer (project has no scaffolding yet) If this is a code project being built but `docs/STATUS.md` + `docs/log/` don't exist, and there's something worth tracking, **offer** (don't force): > "This project has no STATUS/log yet. Want me to scaffold it (via `decision-log` Init Mode) so > `/catchup` and `/end` have something durable to read? (y/n)" On **yes** → run `decision-log` Init Mode (it builds STATUS + decisions/ + today's log from existing context, draft-first). On **no** → fall back to a single session-draft breadcrumb and move on. Never scaffold silently. **Missing VISION in an already-scaffolded project.** If `docs/STATUS.md` + `docs/decisions/` already exist but there's no `docs/VISION.md` (the enduring charter — the top layer), and this was a substantive session, **offer** (don't force): > "This project has STATUS/ADRs but no VISION.md (its enduring north star). Want me to author it > now via `decision-log` (it'll ask ~3 questions, then load-only forever after)? (y/n)" On **yes** → run `decision-log` (its Step 1 backfill routes straight into the one-time VISION interview, draft-first). On **no** → leave it; the session-start hook and `/catchup` will keep the gap visible next time. Author once, never nag mid-session. ## Step 4 — Approval gate, no commit - Everything is **draft-first**: show proposed writes (delegated skills handle their own draft/approve prompts) and wait for explicit approval before anything hits disk. - **Never `git add` / `git commit`.** Consistent with the user's git discipline and with every other skill in this harness. Leave staging to the user. ## Step 5 — Clear the routed session-draft(s) If this close-out consumed one or more unattended captures from `~/.claude/session-drafts/` (the SessionEnd-hook output), then **after the user approves the writes**, move those draft files into `~/.claude/session-drafts/routed/` (create it if missing). This is what stops `/catchup` from re-surfacing them as "pending" forever. If nothing was approved/written, leave the drafts in place. Never delete a draft — archiving to `routed/` preserves the trail. ## Relationship to the rest of the harness
From julianhernandez2155/claude-code-harness, MIT.
More from julianhernandez2155/claude-code-harness
- 0.479score
context-map
julianhernandez2155
Build a comprehensive, skimmable context map of an arbitrary project folder — auto-detecting whether it is a CODEBASE or a RESEARCH/NOTES folder and mapping…
SkillCFreshness: CClaude Code +9 - 0.479score
deep-research
julianhernandez2155
Multi-platform deep research with reputation-gated sources, two-hop trust expansion, parallel platform fan-out (Web + GitHub + Reddit + X +…
SkillCFreshness: CClaude Code +9 - 0.476score
html-visual
julianhernandez2155
Generate an interactive single-file HTML visualization (mockup, wireframe, ERD, flowchart, chart, slides, architecture diagram, dashboard, timeline, mindmap…
SkillCFreshness: CClaude Code +9 - 0.474score
curate-apply
julianhernandez2155
Review the background curator's latest nightly proposals and apply the ones the user approves.
SkillCFreshness: CClaude Code +9 - 0.474score
qmd
julianhernandez2155
Bootstrap QMD search instructions from the installed qmd CLI.
SkillCFreshness: CClaude Code +9 - 0.474score
codex-review
julianhernandez2155
Generate a paste-ready Codex review prompt for the current branch/diff.
SkillCFreshness: CClaude Code +9 - 0.474score
plan-grill-codex
julianhernandez2155
High-stakes pre-implementation planning gate tuned for this harness's context stack.
SkillCFreshness: CClaude Code +9 - 0.474score
design-review
julianhernandez2155
Empirical, browser-driven design review of front-end changes — drives a live UI with Playwright and observes actual behavior, rather than reasoning from code.
SkillCFreshness: CClaude Code +9 - 0.474score
excalidraw
julianhernandez2155
Draw professional diagrams on a live Excalidraw canvas via MCP.
SkillCFreshness: CClaude Code +9 - 0.474score
folder-architect
julianhernandez2155
Use this skill to design or audit a project's folder/file architecture, grounded in the user's Team Knowledge Base (the patterns/workflows/concepts under…
SkillCFreshness: CClaude Code +9 - 0.474score
catchup
julianhernandez2155
Start-of-session briefing — "where did I leave off?" Reads (never writes) the nearest docs/STATUS.md + recent docs/log/ entries + the newest…
SkillCFreshness: CClaude Code +9 - 0.462score
decision-log
julianhernandez2155
Use this skill to maintain a per-project decision + status + session log under `docs/`.
SkillCFreshness: CClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Jun 8, 2026 (4 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from julianhernandez2155
- 0.479score
context-map
julianhernandez2155
Build a comprehensive, skimmable context map of an arbitrary project folder — auto-detecting whether it is a CODEBASE or a RESEARCH/NOTES folder and mapping…
SkillCFreshness: CClaude Code +9 - 0.479score
deep-research
julianhernandez2155
Multi-platform deep research with reputation-gated sources, two-hop trust expansion, parallel platform fan-out (Web + GitHub + Reddit + X +…
SkillCFreshness: CClaude Code +9 - 0.476score
html-visual
julianhernandez2155
Generate an interactive single-file HTML visualization (mockup, wireframe, ERD, flowchart, chart, slides, architecture diagram, dashboard, timeline, mindmap…
SkillCFreshness: CClaude Code +9 - 0.474score
catchup
julianhernandez2155
Start-of-session briefing — "where did I leave off?" Reads (never writes) the nearest docs/STATUS.md + recent docs/log/ entries + the newest…
SkillCFreshness: CClaude Code +9 - 0.474score
plan-grill-codex
julianhernandez2155
High-stakes pre-implementation planning gate tuned for this harness's context stack.
SkillCFreshness: CClaude Code +9 - 0.474score
curate-apply
julianhernandez2155
Review the background curator's latest nightly proposals and apply the ones the user approves.
SkillCFreshness: CClaude Code +9
Cite this page
end skill. Agent Mart. https://agentmart.dev/skills/julianhernandez2155-end (last updated 2026-10-24)