ENOCH WORLD / Public operating procedure / Version 1.1
Use an AI skill with a clear scope and a reviewable result.
A practical procedure for selecting an ENOCH workflow, checking access, supplying authorized inputs and recording what the work actually delivered.
Access status: This SOP is public guidance for the six documented catalog workflows. Five reviewed ENOCH skill packages are installed for ENOCH’s use; Traffic Attribution has a separate public guide. No public ChatGPT share link or plugin listing has been verified for these skills. Use the procedure with material and connections actually available in your own authorized environment.
The examples are task templates. They do not install a skill, connect an account or initiate a paid service.
Use the free 16-workflow catalog
Before adapting this SOP, inspect the ENOCH Agent Discovery Index for the public workflow MCP service, tested catalog scope and limits. The five-workflow paid kit and free sixteen-workflow service are different editions.
Two focused, downloadable AI evidence kits
ENOCH GEO & LLM Discovery Kit — $79: original source-to-page SOP, eight prompts, citation-test plans, checklists and SHA-256-verified ZIP. Synthetic exercises; no guaranteed rankings.
ENOCH AI Traffic Attribution & Evidence Kit — $99: attribution SOP, campaign-tagging guide, offline session analyzer, synthetic fixtures and six passing unit tests. AI source tags are not authenticated provider identity.
Both are one-time licensed internal-use products with manual secure-link delivery within two business days after order verification. The separate $149 five-workflow bundle remains available. Explore the free 16-workflow service before choosing.
Read the SOP free; use the structured pack in your own work
This page provides the operating procedure and a simple run record at no charge. The ENOCH AI Agent Skills & SOP Dataset — Operational Reference Edition, v1.0.0 adds reusable, versioned files: 5 workflow records, 8 SOP-step records, 15 prompts, 10 designed input/output scenarios and 1 documented publication case, plus schemas, receipt templates, source notes, an integrity validator and two HTML templates.
The five included workflows are Dispatch, GEO Agent, Enterprise Diligence, SOP Delivery and Paid Samples. Traffic Attribution remains a separate public guide. Use the portable Markdown, JSON/JSONL and HTML records with your own authorized tools; the dataset does not install a personal skill or grant a service entitlement.
$149 USD, one-time, for one legal entity’s internal use. The entity may adapt the methods and build its own products. Manual secure-link delivery to the checkout email within two business days after order verification; no physical shipping.
Match the skill to the outcome
- ENOCH ONE Dispatch
- Clarify the request, scope the work and select a specialist workflow.
- ENOCH GEO Agent
- Turn approved research into discoverable content and evaluate the available acquisition evidence.
- ENOCH Enterprise Diligence
- Review a proposal or claim set, identify evidence gaps and define a meaningful enterprise pilot.
- ENOCH SOP Delivery
- Organize authorized execution, quality review and handover using task-specific procedures.
- ENOCH Paid Samples
- Inspect an explicitly priced sample and reconcile access, payment and delivery through a configured service.
The operating procedure
-
Choose the outcome
State the deliverable, audience and intended use. Identify who will accept the result and the criteria they will use. A research brief might require dated primary sources; a website change might require an accessible page, correct links and an accurate public status.
Record: Objective, deliverable, audience, destination and acceptance criteria.
-
Inspect the scope and inputs
Read what the selected workflow covers. List the sources, files, systems and decisions it needs. Confirm that the inputs may be used for the intended purpose. Record unavailable material and unresolved claims instead of treating the source collection as complete.
Record: Input inventory, source dates, permitted use and known gaps.
-
Check release and service access
If an approved shareable release or verified service connection is available, check its version, usage terms, required tools and limits before using it. Confirm that it is accessible to your account. For a plugin, use the actual published listing or supported sharing route supplied with the release.
A catalog page, an installed owner workflow, a hosted tool endpoint and a paid entitlement are separate forms of access. If a required connection is unavailable, the run can prepare a draft or report the missing dependency within the agreed scope.
Record: Release or service version, verified access path and prerequisites still needed.
-
Use your own authorized connections
Use only accounts, files and systems you are permitted to access. Supply the material needed for the task and keep credentials in the intended connection or credential manager. Sharing or installing a workflow does not transfer another person’s account access.
Record: Authorized sources and connected tools, without copying credentials into the run record.
-
Run a bounded task
Specify what the agent may read, draft, edit or publish, the destination of the result and any spending limit. Begin with a defined deliverable. Procedures and acceptance checks should be adapted to the actual task and reviewed by an appropriate practitioner.
For paid samples, inspect the exact quote, resource, price and usage terms before an authorized transaction. Record payment and delivery separately; an uncertain payment needs reconciliation before another attempt.
Record: Permitted actions, limits, the task request and actual execution state.
-
Review evidence and limitations
Check claims against their sources and verify the criteria agreed at intake. Separate observed results, calculations, owner-reported information, proposals and unknowns. Record failed checks and missing inputs. A structurally valid file or a drafted scenario is not evidence that a real service run succeeded.
Record: Output references, checks performed, results, reviewer findings and unresolved questions.
-
Verify delivery and publish within scope
Complete the relevant checks and release the accepted result within the granted authority. Open the resulting page or file as its intended audience where possible. For a website change, check its links and layout; for a service, verify the returned result and access boundary. Keep a previous accepted version or practical recovery path.
Record: Delivered location, verified access, acceptance decision, release version and recovery information.
-
Record the outcome and update when needed
Keep the skill version, important source versions, completed actions and remaining dependencies together. Recheck affected work when the skill, connected service, source material or intended use changes. Give unfinished actions a clear owner and next step.
Record: Final status, follow-up owner and the event that should trigger another review.
The same method, used in ENOCH’s public release
The bundle documents one actual catalog/SOP publication: five installed definitions inspected, the two public pages prepared and verified, Enterprise → AI & Security navigation links added, a header anchor corrected, and existing pricing/API sections preserved. The ten teaching scenarios are designed examples and remain unexecuted.
Read the single documented ENOCH publication case. Its evidence concerns that bounded site release; it does not establish separate execution of all five workflows, a paid transaction or a public plugin installation.
A simple run record
Copy these fields into your own task notes. Include links or identifiers for approved evidence instead of duplicating sensitive material.
- Task and date
- A short name, the responsible person and when the run occurred.
- Skill and version
- The workflow used, release version and actual access path.
- Objective and acceptance
- The requested output and the criteria for accepting it.
- Inputs and scope
- Authorized sources, dates, permitted actions and spending or publishing limits.
- Execution
- What was run, what was prepared, and what remains blocked or unattempted.
- Evidence and checks
- Source references, test results, reviewer findings and limitations.
- Delivery and next step
- The output location, acceptance decision, recovery path and follow-up owner.
A useful first request
“Use [skill name] to prepare [deliverable] for [audience] using [approved inputs]. Work within [scope and permitted actions]. Return the result, supporting sources, checks performed and anything still needed for acceptance.”
When the task involves discovery or audience growth
Keep human activity, automated requests and owner testing separate when the available data supports that distinction. Preserve the metric definition, date range, timezone and data source. Views, sessions, visitors and requests measure different things.
A crawler request does not establish that an AI answer recommended a page. A referral can support an observed click source; the original search term or AI prompt may still be unavailable. A traffic peak recorded during testing is an observation to investigate, not proof of an agent’s causal effect.
Use documented campaign links, known test sessions, authorized server logs and available platform reports to investigate attribution. Mark unavailable keywords and uncertain identity as unknown. Report genuine audience outcomes against a suitable baseline with the assumptions visible.
Before offering a skill to other users
Prepare a release containing only material approved for that audience. Provide its version, scope, usage terms, dependencies and example tasks. Test it with an appropriate recipient or separate authorized account so the result reflects the recipient’s actual access.
For public ChatGPT distribution, OpenAI’s documented route is a plugin package with the applicable checks, review and explicit publication. A skill-only plugin can distribute workflows; live tools require their own supported service connection. Add the verified release link to the catalog once publication is complete.
OpenAI: Build skills · OpenAI: Upload, review and publish a plugin
Place ENOCH AI resources in the Marketplace of Ideas
List ENOCH AI products, datasets, agent workflows, SOPs and clearly AI-related research or publications in the Marketplace of Ideas, while retaining useful topical collections. State the actual formats, license scope, evidence status and delivery or access conditions. Collection membership does not publish or activate an unready draft.
Keep future descriptions in this order: role and use → documented or designed example, clearly labeled → deliverable → price and access.
Traffic Attribution Agent: additional evidence rules
Use the ENOCH Traffic Attribution Agent for source, campaign and landing-page audits. Query raw referrer domains and UTM tags as well as processed channel labels; a “direct” session can still carry an AI-associated tag.
Tags are recorded signals, not authenticated identity. A verified assistant fetch requires authorized request logs and operator verification. Preserve owner statements about testing separately, leave unknown sources unresolved, and never infer private prompts or distinct people from sessions.
Check date range, timezone, row limits and count reconciliation before publishing sanitized aggregates. Read the October 30 case and method.