# Work Plan

> Turn a task into a detailed, read-only execution plan by always launching one gpt-5.6-sol xhigh analyst and producing a self-contained handoff suitable for a lower-capability execution model. Use when the user wants a plan before implementation.

- **Type:** Skill
- **Install:** `agentstack add skill-xcaeser-work-skill-work-plan`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [xcaeser](https://agentstack.voostack.com/s/xcaeser)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [xcaeser](https://github.com/xcaeser)
- **Source:** https://github.com/xcaeser/work-skill/tree/main/skills/work-plan
- **Website:** https://skills.sh/xcaeser/work-skill

## Install

```sh
agentstack add skill-xcaeser-work-skill-work-plan
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Work / 3. Plan

Produce an implementation-ready plan without changing source files.

Make the plan protect reliability, restraint, modular boundaries, the simplest
complete implementation, human authorship, clean breaks over obsolete
compatibility, meaningful tests, and proof over status reports. Prefer existing
dependencies and durable architecture over speculative packages or stopgaps.

## Fixed route

- If the exact subagent model, effort, or goal tools are unavailable, return `blocked` with the missing capability. Do not substitute a model or perform the analyst's work in the parent.
- Launch exactly one `gpt-5.6-sol` subagent at `xhigh` effort with `fork_turns: none`.
- Give it a fun call sign, the exact task statement, source of truth, read-only ownership, constraints, acceptance checks, and validation requirements.
- Require it to inspect independently and return concrete evidence, not a status report or a claim that an unproved step is “routine.”
- Require it to call `create_goal` with the exact analysis goal before inspection and `update_goal complete` only after the required evidence is collected.
- Let the analyst run until it returns the plan, a real blocker, or a user-input request. Do not cancel it because a polling window or arbitrary wall-clock interval elapsed; a wait timeout is not a failure.
- The analyst must not edit, commit, deploy, spawn, or delegate.
- Scale depth to the task. Inspect only relevant paths, direct consumers, and evidence needed for the handoff; return the plan without process narration and stop when every required field is supported.
- Report the successful launch as a compact table with `Agent`, `Working on`, `Goal`, `Ownership`, and `Model`.

## Build the plan

The parent defines the mission. The analyst inspects only relevant context, traces current behavior through real consumers, and resolves evidence-backed decisions. Keep early alternatives independent; do not let the first elegant reduction become the plan without checking it against the exact task. The parent verifies paths, symbols, commands, and assumptions before returning the handoff.

For test work, inspect the implementation, existing tests, public APIs, and
actual user flows before proposing a test. Put the four-line test plan before
any test implementation assignment, and name tests deliberately omitted when
no real behavior justifies them.

Confirm that implementation serves a defined user outcome rather than merely
being cheap to generate. If the plan includes prototypes or variants, require
each to answer a named question and include the evidence, selection rule, or
kill criterion that determines what survives.

For qualitative decisions, record the fundamentals, references, context, and
tradeoffs that make the chosen direction appropriate. Do not use preference or
trend-following as the entire rationale.

```markdown
## Work / 3. Plan

**Mission:** 

**Source of truth:** 

**Current behavior and evidence**
- 

**Done when**
- 

**Does not count**
- 

**Decisions, constraints, and non-goals**
- 

**Taste rationale**
- 

**Alternatives rejected**
- 

### Implementation map
| Step | File / symbol | Exact change | Reason / invariant |
|---|---|---|---|
| 1 |  |  |  |

**Executor assignments:** 

### Validation
| Command or flow | Expected evidence | Proves |
|---|---|---|
|  |  |  |

**Testing**
- **Behavior being protected:** 
- **Why it matters:** 
- **Test level:** 
- **Exact observable assertion:** 
- **Tests deliberately not added:** 

**Risks or material unknowns**
- 

**Ready:** Yes | No — 
```

Assume the plan may be executed by a lower-capability model with no access to hidden context. Make it self-contained: name exact files and symbols, ordered edits, patterns to follow, invariants, edge and failure cases, acceptance evidence, and commands with expected results. Resolve judgment-heavy decisions in the plan instead of delegating them to the executor.

A plan is not ready when it merely restates the task, reduces it to an unproved assumption, omits a required acceptance check, or reports progress without evidence. If a route is blocked, state the exact gap and propose another round only for a materially new mechanism. Do not launch an executor or edit files from this skill. Hand a verified ready plan to `$work` for execution.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [xcaeser](https://github.com/xcaeser)
- **Source:** [xcaeser/work-skill](https://github.com/xcaeser/work-skill)
- **License:** MIT
- **Homepage:** https://skills.sh/xcaeser/work-skill

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-xcaeser-work-skill-work-plan
- Seller: https://agentstack.voostack.com/s/xcaeser
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.