referee
Judge a pending data answer before it ships — run tripwires and the checklist, optionally an independent judge pass, and return a verdict (PASS…
Skill by danjamk
Chalk Labs
Need help deploying referee?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/danjamk-referee
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: data
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/referee
What's inside
- Skill name
referee- Description
- Judge a pending data answer before it ships — run tripwires and the checklist, optionally an independent judge pass, and return a verdict (PASS, PASS-WITH-CAVEAT, PROVISIONAL, REFUSE). Use when the user says "referee this", "check this answer", "is this number right", or before sharing an analysis.
- License
- MIT
- Instructions
- 22 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Referee Spec: `<base>/../../kit/REFEREE.md`. ## Deterministic pass 1. Run `referee/tripwires.sql`; record each value against its threshold. 2. Walk `referee/checklist.md`; every "no" gets a stated consequence. 3. Assign the verdict. ## Judge pass (independent) Launch a subagent with **only**: the final answer text, the numbers and their stated frame, `referee/checklist.md`, `meaning/caveats.yaml`, and `meaning/refusals.yaml`. Do not give it the reasoning, the SQL drafts, or the conversation. Ask it to return a verdict and the checklist items it believes fail, with reasons. If the judge has not been calibrated on this domain's goldens (see REFEREE.md §7), report its verdict as **advisory** next to the deterministic one. Where they disagree, show both. ## Output The verdict ships with the answer. Log it in the checklist's verdict log.
From danjamk/analytics-context-engine, MIT.
More from danjamk/analytics-context-engine
- 0.565score
analyze
danjamk
Answer a data question through a governed context pack — load context, resolve terms, declare the frame, run tripwires, query read-only, reconcile, referee…
SkillAFreshness: AClaude Code +9 - 0.565score
check
danjamk
Validate a data domain's context pack — references resolve, measured facts are not stale, no golden leaks into exemplars, tripwires run, and goldens reproduce.
SkillAFreshness: AClaude Code +9 - 0.565score
recipe
danjamk
Run, refine, or capture a recipe — a remembered analysis for a recurring report, with its objective, audience, sources, rules, steps, watch-for list, output…
SkillAFreshness: AClaude Code +9 - 0.565score
upgrade
danjamk
Upgrade a context engine instance (operator core or domain pack) made by an older analytics-context-engine kit version — read the changelog entries since its…
SkillAFreshness: AClaude Code +9 - 0.559score
bootstrap
danjamk
Set up a context engine for data analysis — create the operator's private core and a first domain pack, interview for definitions and rulings, seed goldens…
SkillAFreshness: AClaude Code +9 - 0.559score
new-domain
danjamk
Add a new data domain (database, export, API) to an existing context engine — register its sources, profile it for grains and traps, run the rulings…
SkillAFreshness: AClaude Code +9 - 0.559score
propose
danjamk
Stage a change to a data domain's meaning (definition, reading choice, refusal, caveat, recipe) as a reviewable proposal with blast radius and golden impact…
SkillAFreshness: AClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 8
- Last commit (this folder)
- Oct 2, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from danjamk
- 0.565score
analyze
danjamk
Answer a data question through a governed context pack — load context, resolve terms, declare the frame, run tripwires, query read-only, reconcile, referee…
SkillAFreshness: AClaude Code +9 - 0.565score
check
danjamk
Validate a data domain's context pack — references resolve, measured facts are not stale, no golden leaks into exemplars, tripwires run, and goldens reproduce.
SkillAFreshness: AClaude Code +9 - 0.565score
recipe
danjamk
Run, refine, or capture a recipe — a remembered analysis for a recurring report, with its objective, audience, sources, rules, steps, watch-for list, output…
SkillAFreshness: AClaude Code +9 - 0.565score
upgrade
danjamk
Upgrade a context engine instance (operator core or domain pack) made by an older analytics-context-engine kit version — read the changelog entries since its…
SkillAFreshness: AClaude Code +9 - 0.559score
bootstrap
danjamk
Set up a context engine for data analysis — create the operator's private core and a first domain pack, interview for definitions and rulings, seed goldens…
SkillAFreshness: AClaude Code +9 - 0.559score
new-domain
danjamk
Add a new data domain (database, export, API) to an existing context engine — register its sources, profile it for grains and traps, run the rulings…
SkillAFreshness: AClaude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get referee running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
referee skill. Agent Mart. https://agentmart.dev/skills/danjamk-referee (last updated 2026-10-24)