oracle-validation
Compare runtime or kernel output with the project oracle, isolate the first divergence, and produce an evidence-ready verdict.
Skill by linjuhon
Chalk Labs
Need help deploying oracle-validation?
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/linjuhon-oracle-validation
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
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/oracle-validation
What's inside
- Skill name
oracle-validation- Description
- Compare runtime or kernel output with the project oracle, isolate the first divergence, and produce an evidence-ready verdict. Use when an active gate requires numeric truth or correctness comparison.
- License
- MIT
- Instructions
- 34 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# oracle-validation Use this skill when comparing kernel output against the contract oracle. ## Responsibilities - compare actual versus expected data - identify the first failing coordinate - keep tolerance and identity explicit - separate correctness failure from missing observation ## Required Output - comparison scope - expected values - actual values - tolerance - first divergence - pass/fail verdict ## Hard Rules - Do not widen the compare range until the current first failure is explained. - Do not treat partial agreement as pass. - Do not change the oracle to match the implementation. ## Entry And Return Read the active gate and its required fixture or freeze identity before comparing. Return the smallest reproducible comparison command, first failing coordinate, and expected-versus-actual result. Leave PASS/FAIL adjudication to the gate contract unless the gate explicitly delegates that verdict.
From linjuhon/codex-personal-skills, MIT.
More from linjuhon/codex-personal-skills
- 0.524score
ask-matt
linjuhon
Choose one skill or workflow for a task across personal, Matt Pocock, Codex, and plugin capabilities.
SkillBFreshness: BClaude Code +9 - 0.524score
documentation-organization
linjuhon
Organize project documentation into a clear authoritative structure, classify source documents, preserve traceability, and create indexes or handoff maps.
SkillBFreshness: BClaude Code +9 - 0.524score
patch-decomposition
linjuhon
Decompose fork-versus-base changes into one reviewable plugin patch with hunk ownership, dependencies, activation boundaries, verification, and rollback…
SkillBFreshness: BClaude Code +9 - 0.524score
upstream-rebase
linjuhon
Rebase an accepted, dependency-ordered patch onto an already locked upstream revision with conflict, verification, and rollback evidence.
SkillBFreshness: BClaude Code +9 - 0.521score
binary-identity
linjuhon
Verify that source, build configuration, binary, and runtime identity match before treating output as evidence.
SkillBFreshness: BClaude Code +9 - 0.521score
cuda-kernel-debug
linjuhon
Narrow a CUDA kernel correctness failure to the first failing coordinate and one semantic owner using minimal probes.
SkillBFreshness: BClaude Code +9 - 0.521score
route-operations
linjuhon
Route multi-step document, artifact, browser, GitHub, Gmail, website, or image operations when the correct managed capability or write authorization is unclear.
SkillBFreshness: BClaude Code +9 - 0.521score
evidence-packaging
linjuhon
Package completed bounded execution output into one evidence claim with identity, observations, limitations, and receipt.
SkillBFreshness: BClaude Code +9 - 0.521score
git-worktree-protection
linjuhon
Protect dirty worktrees and narrow file ownership during execution, evidence archival, and cleanup.
SkillBFreshness: BClaude Code +9 - 0.521score
route-planning
linjuhon
Route multi-step planning, research, design decisions, specifications, and ticketing when the next planning skill is unclear.
SkillBFreshness: BClaude Code +9 - 0.521score
quality-evaluation
linjuhon
Run and interpret the project's quality matrix after correctness and route gates are closed.
SkillBFreshness: BClaude Code +9 - 0.521score
route-diagnostics
linjuhon
Route multi-step bug, regression, CI, CUDA, or performance diagnosis when failure isolation requires choosing a method.
SkillBFreshness: BClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Aug 8, 2026 (3 months ago)
- Last release
- Aug 8, 2026 (3 months 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 linjuhon
- 0.524score
patch-decomposition
linjuhon
Decompose fork-versus-base changes into one reviewable plugin patch with hunk ownership, dependencies, activation boundaries, verification, and rollback…
SkillBFreshness: BClaude Code +9 - 0.524score
upstream-rebase
linjuhon
Rebase an accepted, dependency-ordered patch onto an already locked upstream revision with conflict, verification, and rollback evidence.
SkillBFreshness: BClaude Code +9 - 0.524score
ask-matt
linjuhon
Choose one skill or workflow for a task across personal, Matt Pocock, Codex, and plugin capabilities.
SkillBFreshness: BClaude Code +9 - 0.524score
documentation-organization
linjuhon
Organize project documentation into a clear authoritative structure, classify source documents, preserve traceability, and create indexes or handoff maps.
SkillBFreshness: BClaude Code +9 - 0.521score
cuda-kernel-debug
linjuhon
Narrow a CUDA kernel correctness failure to the first failing coordinate and one semantic owner using minimal probes.
SkillBFreshness: BClaude Code +9 - 0.521score
append-only-gate-governance
linjuhon
Legacy append-only gate governance instructions retained for compatibility with older references.
SkillBFreshness: BClaude 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 oracle-validation 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
oracle-validation skill. Agent Mart. https://agentmart.dev/skills/linjuhon-oracle-validation (last updated 2026-10-24)