Skip to content
Back to skills

Orient

ASecurity

At end-of-session, dispatches a read-only sub-agent to survey recent work and drafts a vaguely-specific priming prompt the next session can paste to orient itself and plan its next sprint. Produces orientation, not prescription — candidate directions are framed as questions, not decisions.

  • 56 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 20, 2026
ai-agentssqlgit

Security analysis

A100/100

Scanned September 20, 2026

npx -y skills add griffinwork40/agent-afk --skill orient --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Orient?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Orient
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/griffinwork40-orient/badge)](https://www.skillsdirectory.com/skills/griffinwork40-orient)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: orient
description: At end-of-session, dispatches a read-only sub-agent to survey recent work and drafts a vaguely-specific priming prompt the next session can paste to orient itself and plan its next sprint. Produces orientation, not prescription — candidate directions are framed as questions, not decisions.
context: load
---

## Sub-agent contract
/contract

Dispatch one read-only research sub-agent to survey recent work across four narrow targets, then produce a single markdown prompt block the user can paste into a fresh session.

**Sub-agent surveys:**
- Memory: the **`memory_search` tool** (2–3 keyword queries on the current work; FTS5 syntax) + hot memory at `~/.afk/state/memory/HOT.md`. Do **not** grep `~/.claude/projects/**/memory/` — that path is vestigial and empty on both trees; the archive is SQLite and only `memory_search` reaches it. Report which stores you consulted.
- Git activity in this repo + sibling repos the session touched (`git log --oneline -10`, open PRs, dirty state)
- `experiments/` docs (if present) relevant to the current work
- Telemetry/state files the session interacted with (e.g. `~/.afk/agent-framework/*.jsonl`)

**Return exactly this shape (≤ 60 lines total):**
1. One-line project identity + one-line "where we are" (current phase / latest milestone)
2. 3–5 bullets on what shipped recently — commits, PRs, empirical closures; be specific
3. Memory pointers by filename only (next session's memory system auto-loads — don't restate)
4. 2–4 **candidate directions** framed as questions or possibilities, **never** as a decided plan
5. Closing line inviting the next session to run `/ground-state` first, then propose a plan

**Discipline the contract enforces:** orientation, not execution. Handoff briefs drift toward prescription by default; the "questions-as-possibilities" framing is the guardrail so the receiving session plans with fresh judgment.

Relay the prompt to the user. On request, save to `/tmp/orient-<ISO8601>.md`.

**Skip when:** no substantive work to carry forward, user wants a bare `/compact`, or user already has a decided plan (use `/mint` or draft directly).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.