agentleFS
Sign inStart free
wjgoarxivwjgoarxiv/litopencode/skills/visual-qa/SKILL.md

visual-qa

Skill2 starsChanged 45 days agoMITView on GitHub

Use a concise OpenCode-native evidence gate for rendered surfaces, with exact blocked outcomes and lazy detailed references.

20 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 20 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use only an existing project or host executor.

    line 40
  2. Distinctive:

    No browser install, network authority, MCP route, authentication persistence, host-config mutation, or write authority.

    line 55

“Common” is measured: kept by at least 6 in 10 of the 20 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

#contract.activation

contract_schema_version: litopencode.skill_contract.v1
skill_id: visual-qa
runtime_class: runtime-skill
static_documentation: true
feature_ids: [visual-qa, doctor-install]
entry_routes: ["chat.message visual-qa", "native skill visual-qa", "tool.execute.after advisory"]

Visual QA

This is static documentation for the native-installed LitOpenCode validator. Do not execute commands from this file automatically. It is not a browser executor, does not make a browser callable, and grants no capture authority.

#contract.inputs

FieldRequired contract
targetExact surface, state, viewport, interaction, and binary observable
artifactsBounded captures with an evidence-eligible beta design hash, source/capture hashes, and freshness
reviewerHost-proven independent identity when the tier requires review

#contract.mode_matrix

ModeEvidenceOutcome
smokeCritical inventory and one real channelPASS, FAIL, or BLOCKED
fullComplete inventory plus independent reviewPASS only with material evidence
reference-fidelityFull evidence plus reference targetsSemantic and visual checks must pass

#contract.procedure

  1. Probe callable current-session capability before promising capture.
  2. Use only an existing project or host executor; define the observable before running it.
  3. Accept only caller-authorized open material descriptors; read, recheck, and close each. Validate bytes, hashes, pointers, source freshness, passing required channels, findings, and cleanup.
  4. Reconcile the frozen contract source hash and review policy. Require host-proven independent review over immutable inputs whenever the contract says so; self-attestation cannot establish provenance.
  5. Return the highest-precedence honest verdict. BLOCKED outranks FAIL and cannot become PASS by prose.

#contract.outputs

Return PASS with paths and hashes, FAIL with findings, or an exact BLOCKED receipt. Only litfamily.evidence-manifest/v1beta1 bound to the caller's exact open-descriptor set is eligible. A root path alone blocks: Node lacks portable openat, so containment cannot be proved across root swap/restore. Nested labels block; v1alpha1 is diagnostic only. Full and reference-fidelity remain blocked without host-owned reviewer provenance.

#contract.evidence

Require the canonical litfamily.design-contract/v1beta2 Design Contract. Valid litfamily.design-contract/v1beta1 documents remain accepted and evidence-eligible. Validate material evidence with schemas/evidence-manifest-v1beta1.json; alpha is migration-only and cannot PASS evidence. Read references/capture-playbook.md for channel capture and references/complete-contract.md only for detailed schemas, tiers, commands, reviewer rules, and failures.

#contract.hard_stops

  • No browser install, network authority, MCP route, authentication persistence, host-config mutation, or write authority.
  • A path, HTTP response, similarity score, or same-context assertion is not material visual evidence.
  • Missing renderer, safe authentication, reviewer provenance, timeout, or cleanup is BLOCKED.

#contract.anti_patterns

Reject nonexistent, non-image, stale, self-attested, root-escaping, or symlink evidence; reject browser criteria downgraded to non-browser surfaces.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free