Skip to content

phuryn/product-engineering / app-stack/skills/embedded-agents-agent-sdk

embedded-agents-agent-sdk

Verified

Only when the Agent SDK is asked for, an agent needs Claude Code's own tools, or the app already runs it: embed Claude agents in a Next.js app with the Agent SDK.

examples/knowledge-agent/knowledge/okrs.md

View on GitHub
# OKRs — Objectives and Key Results

## What OKRs are

OKRs are a goal-setting framework popularized by Intel and Google. An **Objective** is a qualitative, inspirational statement of what you want to achieve. **Key Results** are 3–5 quantitative, time-bound measures that prove the objective was achieved. If every Key Result is hit, the Objective is, by construction, accomplished.

## Structure

- **Objective:** A short, memorable sentence. Qualitative. Ambitious but achievable.
- **Key Results:** Measurable outcomes. Each KR has a numeric target and a deadline. KRs are *results*, not activities — "ship feature X" is not a KR, "increase weekly active users from 40k to 60k" is.

A canonical OKR has exactly one Objective and 3–5 Key Results. More than 5 is a signal that the Objective is too broad or the KRs are tasks in disguise.

## What good Key Results look like

- **Outcome, not output.** "Reduce p95 latency from 800ms to 300ms" is an outcome. "Migrate the API to Rust" is an output.
- **Numeric with a baseline.** Always state where you start, where you want to end, and by when. Without a baseline you cannot measure progress.
- **Independent.** A KR should not be a sub-step of another KR on the same Objective.
- **Owned.** A single person is accountable for each KR, even if a team delivers it.

## Scoring and ambition

OKRs are scored 0.0 to 1.0 at the end of the cycle. **Hitting 0.7 on average is the target, not 1.0.** A team that consistently scores 1.0 is setting goals that are too easy. A team that scores 0.3 is either over-reaching or under-resourced. The 0.7 norm is what makes OKRs a stretch tool rather than a commitment tool.

## Cadence

- **Quarterly company OKRs** set direction.
- **Team OKRs** ladder up to company OKRs and are also quarterly.
- **Weekly check-ins** track confidence ("on track / at risk / off track") not just numbers.

## Common failure modes

- **OKRs become a task list.** When KRs are activities ("launch the feature", "run the campaign"), the framework collapses into a project plan.
- **Too many OKRs.** More than 3 Objectives per quarter per team dilutes focus. Pick the few that matter.
- **OKRs tied to compensation.** As soon as bonuses are linked to OKR scores, teams sandbag — set easy goals to guarantee the payout. Keep them separate from performance reviews.
- **Cascading without translation.** Copy-pasting the CEO's OKRs down the org produces nonsense. Each level translates the parent OKR into its own outcomes.

## When OKRs are the right tool

OKRs fit organizations that need alignment across many teams, ambitious stretch targets, and a regular rhythm of re-prioritization. They are a poor fit for very small teams (overhead exceeds value) and for work that is purely operational/run-the-business (KPIs do that job).
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 470kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall