SpecForge
You are the manager; an AI agent does the work. Agents are fast — and they make confident mistakes. Write a clear brief, run the agent, read its trace, and catch what went wrong by fixing the brief, not the output. You are scored on what you caught, never on speed.
How are you feeling?
A case · you decide
Ayesha's research brief
Ayesha, 17 — using an AI agent to gather sources for a history essay on the 1918 flu
Ayesha's brief says: 'Find sources about the 1918 flu.' The agent returns eight sources in neat citation format. One is a book whose title she cannot find in any library catalogue, and two are about the 2009 flu.
A neat citation is not evidence that a source exists.
A vague brief gives the agent room to drift. 'About the 1918 flu' does not say what she needs.
What does she do about the book she cannot find?
How should she rewrite the brief?
The second run returned five sources, each with an archive link. Ayesha opened every one before it went into her notes.
Who made this?
“In reality, every technology reflects decisions made by designers, companies, and communities, and carries the values and power of those choices.”
- Who benefits from this technology?
- Whose needs are ignored?
Brain break
What next? Pick a door
- More like this ReviewForge Which test catches the bug? KindForge Will this program compile? ChallengeForge Does this entry clear the rubric?
- A bit harder ShipForge Will this app pass review? FlowDojo What steps make up your routine? LocaleForge How many characters is “phở,” really?
- Something different FirstIdeaQuest How did humans first think in symbols? TranslateDojo What does this saying really mean? AnchorQuest Which thinking trap is this?