# Outcome Eval

> Outcome evaluation (OFC/striatum). Invocala dopo aver completato un'azione concreta, prima di considerarla chiusa. Use after each substantive action to verify it produced the expected effect — not "should have", actually did. Triggers anche su "fatto X", "completato Y", "il comando è andato a buon fine?", quando hai un risultato ma non l'hai ancora confrontato con l'aspettativa.

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

## Install

```sh
agentstack add skill-tommasobbianchi-claude-cognitive-skills-outcome-eval
```

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

## About

# Outcome-Eval — OFC/striatum (valutazione esito)

**Principio**: l'azione che credi di aver fatto e l'effetto reale spesso divergono. Verifica.

1. **Aspettativa**: cosa doveva succedere come effetto dell'azione? (1 riga, concreto: "il file X esiste con contenuto Y", "il processo Z è attivo", "l'endpoint risponde 200")
2. **Realtà**: ispeziona ora lo stato reale. Lo strumento per ispezionare dipende dall'effetto atteso (ls, ps, curl, git status, log, screenshot, journalctl, /sauronseye).
3. **Confronto**: aspettativa = realtà? Sì / No / Parziale.
4. **Se No o Parziale**: l'azione non è andata come pensavi. Diagnosi prima di passare oltre — non accumulare debiti di verifica.

Output: 1 riga "atteso X, osservato Y → ok / diverge / parziale".

## Source & license

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

- **Author:** [tommasobbianchi](https://github.com/tommasobbianchi)
- **Source:** [tommasobbianchi/claude-cognitive-skills](https://github.com/tommasobbianchi/claude-cognitive-skills)
- **License:** MIT

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:** yes
- **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-tommasobbianchi-claude-cognitive-skills-outcome-eval
- Seller: https://agentstack.voostack.com/s/tommasobbianchi
- 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%.