draft-okrs
Skill0 starsChanged 6 months agoMITView on GitHub
Draft quarterly OKRs from a North Star. Use when the user asks to draft OKRs, write OKRs, plan a quarter, set objectives, draft KRs / Key Results, or invokes /tandem:draft-okrs. Operates as Founder hat. Reads 00-Strategy/NORTH-STAR.md, the previous quarter's OKRs, and recent retros; produces 00-Strategy/OKR-YYYY-Qx.md following the kit's strategy template.
In this repository32 other files it gives its agents.
- _00-Project-Management/CLAUDE.mdCLAUDE.md
- autopilotSkill
- close-out-storySkill
- close-phaseSkill
- coreSkill
- critiqueSkill
- curate-toolkitSkill
- documentSkill
- draft-epicSkill
- draft-prdSkill
- execute-batch-parallelSkill
- execute-batchSkill
- execute-storySkill
- execution-strategistSkill
- fill-claude-mdSkill
- installSkill
- modeSkill
- monthly-retroSkill
- path-scope-exampleSkill
- peer-reviewSkill
- plan-driftSkill
- refine-backlogSkill
- reflectSkill
- run-testplanSkill
- session-startSkill
- split-into-featuresSkill
- split-into-storiesSkill
- start-phaseSkill
- triage-backlogSkill
- updateSkill
- weekly-monitorSkill
- write-outcomesSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 16
Follow that prompt verbatim.
- Distinctive: line 20
If the user didn't supply context, ask: "Are we drafting for a fresh quarter, or replacing OKRs that feel disconnected?
- Distinctive: line 44
Show the draft to the user in chat before writing the file: Wait for edits before saving.
- Very specific: line 53
Do NOT mark the OKR file ready or in-progress — Founder approval (next session) handles that.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Tandem: draft-okrs (Founder hat)
Operate as Founder hat. The user is starting a new quarter (or replacing OKRs that drifted from the work).
This skill is the slash-command wrapper for the kit's canonical OKR-drafting prompt. The prompt is the source of truth; this file is the entry point.
Source of truth
@_00-Project-Management/92-Prompts/01-draft-okrs-from-northstar.md
Follow that prompt verbatim. The sections below add only the slash-command-specific glue (input resolution, layout detection, post-write handoff) — do not re-declare the prompt's content here.
Inputs needed
- If the user didn't supply context, ask: "Are we drafting for a fresh quarter, or replacing OKRs that feel disconnected? And which quarter's filename should I use — e.g. OKR-2026-Q3?"
- Quarter slug format:
YYYY-Qx(e.g.2026-Q3). Used in the output filename.
Load into context
The canonical layout is under _00-Project-Management/. Use Read / Glob to detect existence rather than assuming; treat missing files as "not present" rather than throwing.
- North Star —
_00-Project-Management/00-Strategy/NORTH-STAR.mdif present. If absent, ask the user to paste their North Star text inline before drafting. - Previous quarter's OKRs —
_00-Project-Management/00-Strategy/OKR-*.md(most recent by filename). If absent, treat this as the first OKR set. - Customer journey —
_00-Project-Management/00-Strategy/CUSTOMER-JOURNEY.mdif present (optional context). - Recent retros —
_00-Project-Management/14-Retros/*.md, most recent 1–2 by filename (optional). - SOP —
_00-Project-Management/90-Standards/SOP.mdfor OKR rules + frontmatter contract. - OKR template —
_00-Project-Management/91-Templates/OKRS.template.mdif present; if absent, scaffold from the structure described in the source prompt and note the gap in the end-of-session summary.
Task
- Re-read the North Star. If it has shifted from what last quarter assumed, stop and ask the user before drafting.
- Review the previous quarter's OKRs (if any): which KRs hit ≥70%, which missed, and for each miss whether it was wrong-target / wrong-action / wrong-quarter.
- Draft a new
OKR-YYYY-Qx.mdper the source prompt's rules:- ≤ 3 Objectives.
- 2–3 KRs per Objective, each a measurable number or binary state.
- Confidence column (0–100%) per KR — honest probability of hitting.
- "What we are deliberately NOT doing this quarter" section.
- Explicit North Star linkage per Objective.
- Show the draft to the user in chat before writing the file. Wait for edits before saving.
- On save: set frontmatter
status: not-started,created_at: <ISO 8601 now from system clock>, all other timestamp fields empty strings. - Final question to the user before saving: "Which of these will hurt to drop in 4 weeks?" — if they can't answer, the set isn't focused enough; iterate.
Output rules
- Three Objectives is the ceiling. Two or one is fine.
- KRs must be numbers or binary states. "Improve X" is not a KR. "X reaches 100 by Sept 30" is.
- KR confidence > 90% is not ambitious. < 30% is not realistic. Aim 50–70%.
- Do NOT mark the OKR file
readyorin-progress— Founder approval (next session) handles that.
Non-negotiable rules from CLAUDE.md
- Frontmatter timestamps (quoted ISO 8601 with offset, from system clock).
- Status enum (closed set of 9 —
not-startedon creation). - Templates rule — use
91-Templates/OKRS.template.mdverbatim if present; do not redraft section headings from memory. - Strategy linkage — every Objective must reference a North Star section.
End-of-session summary (always emit)
- File written:
_00-Project-Management/00-Strategy/OKR-YYYY-Qx.md - Objectives drafted: X
- KRs drafted: Y
- Mean KR confidence: Z%
- Status:
not-started(awaiting Founder approval)
Next command
Next: /tandem:draft-prd — turn one of these Objectives into a PRD.
Or, if you're going straight to execution-shaped strategy: /tandem:draft-epic — turn a KR directly into an Epic.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

