--- title: "AI Context Manifest Template for One Recurring Business Job" description: "Use an AI context manifest to define sources, order, exclusions, freshness, output, approval, stopping rules, tests, and receipts." canonical: "https://scalewithsearch.com/articles/ai-context-manifest-template" date: "2026-08-19" modified: "2026-09-19" --- ## Site navigation - [Scale With Search](https://scalewithsearch.com/) - Real estate - Real estate - [Real estate](https://scalewithsearch.com/for/real-estate) - Work - Start here - [Send your brief](https://scalewithsearch.com/work#send-your-brief) - [Prepare your six-question brief](https://scalewithsearch.com/work#prepare-your-six-question-brief) - Build - [Site build, content library with SEO, signal desk](https://scalewithsearch.com/work) - For your business - Trades and home services - [Auto body and collision shops](https://scalewithsearch.com/for/auto-body-and-collision-shops) - [Foundation and home repair contractors](https://scalewithsearch.com/for/foundation-and-home-repair) - [Garage door and fencing contractors](https://scalewithsearch.com/for/garage-door-and-fencing-contractors) - [HVAC contractors](https://scalewithsearch.com/for/hvac-contractors) - [Janitorial and commercial cleaning companies](https://scalewithsearch.com/for/janitorial-and-commercial-cleaning) - [Locksmiths](https://scalewithsearch.com/for/locksmiths) - [Moving companies](https://scalewithsearch.com/for/moving-companies) - [Pest control companies](https://scalewithsearch.com/for/pest-control-companies) - [Plumbing and electrical contractors](https://scalewithsearch.com/for/plumbing-and-electrical-contractors) - [Restoration and water or fire damage companies](https://scalewithsearch.com/for/restoration-and-water-fire-damage) - [Roofing companies](https://scalewithsearch.com/for/roofing-companies) - [Towing companies](https://scalewithsearch.com/for/towing-companies) - [Tree services and landscaping companies](https://scalewithsearch.com/for/tree-services-and-landscaping) - [Solar installers](https://scalewithsearch.com/for/solar-installation) - [General contractors](https://scalewithsearch.com/for/general-contractors-and-construction) - [Paving, concrete, and flooring contractors](https://scalewithsearch.com/for/paving) - Practices and professional services - [Bookkeeping and tax practices](https://scalewithsearch.com/for/bookkeeping-and-tax-practices) - [Dental practices](https://scalewithsearch.com/for/dental-practices) - [Family and criminal defense law firms](https://scalewithsearch.com/for/family-and-criminal-defense-law-firms) - [Med spas and aesthetics practices](https://scalewithsearch.com/for/med-spas-and-aesthetics) - [Personal injury law firms](https://scalewithsearch.com/for/personal-injury-law-firms) - [Veterinary clinics](https://scalewithsearch.com/for/veterinary-clinics) - [Gyms and fitness studios](https://scalewithsearch.com/for/fitness) - [Therapy and outpatient health practices](https://scalewithsearch.com/for/therapy-and-outpatient-health) - [Medical billing companies](https://scalewithsearch.com/for/medical-billing) - [Insurance agencies](https://scalewithsearch.com/for/insurance-agencies) - [Financial advisors](https://scalewithsearch.com/for/financial-advisors) - [Property management companies](https://scalewithsearch.com/for/property-management) - [Recruiting and staffing agencies](https://scalewithsearch.com/for/recruiting-and-staffing) - [Architects and interior designers](https://scalewithsearch.com/for/architects-and-interior-designers) - [Logistics and supply chain companies](https://scalewithsearch.com/for/logistics-and-supply-chain) - Agencies, MSPs, and manufacturing - [IT and managed service providers](https://scalewithsearch.com/for/it-and-managed-service-providers) - [Machine shops and precision manufacturers](https://scalewithsearch.com/for/machine-shops-and-precision-manufacturing) - [Marketing agencies and freelancers](https://scalewithsearch.com/for/marketing-agencies-and-freelancers) - [SEO agencies and consultants](https://scalewithsearch.com/for/seo-agencies-and-consultants) - [Small manufacturers and fabricators](https://scalewithsearch.com/for/small-manufacturers-and-fabricators) - Restaurants, shops, studios, and nonprofits - [Restaurants and hospitality businesses](https://scalewithsearch.com/for/restaurants-and-hospitality) - [Retail stores and ecommerce sellers](https://scalewithsearch.com/for/retail-and-ecommerce) - [Photographers, event planners, and travel agents](https://scalewithsearch.com/for/photographers) - [Churches and nonprofits](https://scalewithsearch.com/for/churches-and-nonprofits) - [All industries](https://scalewithsearch.com/for/) - Learn - For your office - [Office job guides](https://scalewithsearch.com/guides/) - [Browser calculators](https://scalewithsearch.com/tools/) - Start here - [How it works](https://scalewithsearch.com/how-it-works) - [Free Starter Kit](https://scalewithsearch.com/kit/business-memory-starter-kit.zip) - [Synthetic specimen](https://scalewithsearch.com/specimen/working-session-specimen.zip) - Guides - [The Complete Guide to Business Memory for AI Agents](https://scalewithsearch.com/articles/business-memory-for-ai-agents-guide) - [The Complete Small-Business Guide to AI Agent Governance](https://scalewithsearch.com/articles/ai-agent-governance-guide-small-business) - [The Complete Guide to Leaving Vendor AI Memory](https://scalewithsearch.com/articles/leaving-vendor-ai-memory-guide) - Articles by cluster - [Business memory](https://scalewithsearch.com/articles/business-memory-for-ai-agents-guide) - [Agent governance](https://scalewithsearch.com/articles/ai-agent-governance-guide-small-business) - [Migration and ownership](https://scalewithsearch.com/articles/leaving-vendor-ai-memory-guide) - For machines - [llms.txt](https://scalewithsearch.com/llms.txt) - [llms-full.txt](https://scalewithsearch.com/llms-full.txt) - [Machine view](https://scalewithsearch.com/?view=machine) - Company - Evidence - [Proof](https://scalewithsearch.com/proof) - Company - [About](https://scalewithsearch.com/about) # AI Context Manifest Template for One Recurring Business Job. A weekly report agent reads the current CRM export and an obsolete strategy deck. It gives both files equal weight. The report recommends a campaign the owner ended two months ago. The files were available. The reading contract was missing. A context manifest names what one job may read, which source wins, what is excluded, how fresh each record must be, and what happens when the required context is incomplete. It also names the output, approval boundary, stopping point, and receipt. If the current source path is unknown, begin with the [minimum-context selection test](/articles/what-context-should-an-agent-read) before accepting the manifest. ## A context manifest is the reading contract The manifest sits between the task brief and the source store. The task brief defines the result. The source store holds business records. The manifest identifies the exact records and rules that are eligible for this run. This separation keeps the prompt small. The runtime instruction can say, "Run the weekly owner report using `context-manifest.yaml`." The manifest carries the durable source rules. Anthropic defines context engineering as curating and maintaining the information available during model inference. Its guidance recommends the smallest high-signal set that fully supports the desired behavior. [Anthropic: Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) The manifest makes that curation inspectable. ## The eight fields every manifest needs Use eight field groups: 1. Job identity. 2. Canonical sources. 3. Source order and authority. 4. Exclusions. 5. Freshness and conflict rules. 6. Output contract. 7. Approval and stopping rules. 8. Receipt and tests. Do not add a field because it sounds formal. Add it when a human or agent must make a decision from it. The manifest should remain readable to a fresh operator. A custom schema is acceptable if its meaning is documented and its tests are easy to run. The related [agent job contract](/articles/give-every-agent-a-job-sources-and-stopping-point) defines the result, sources, output, and finish line. The manifest goes deeper on source eligibility. Keep both synchronized through one acceptance fixture. ## Name canonical sources and source order Use exact paths, queries, or stable record identifiers. "Read the relevant business files" is not a source rule. For placing the manifest within a complete agent memory design, read [AI Agent Memory Architecture for Small Business.](/articles/ai-agent-memory-architecture-small-business) The guide to [what context an agent should read](/articles/what-context-should-an-agent-read) starts from the minimum authoritative packet. The manifest records that packet for repeated use. Each source needs a role. Current account facts, approved decisions, corrections, procedures, and examples serve different purposes. Write precedence directly: ```text approved decision > current account record > procedure > transcript > archive ``` Precedence only applies when the scopes match. A company-wide policy does not automatically override a signed client exception. The manifest must name the scope rule. If two permitted records have equal authority and conflict, stop. Do not use modification time as a hidden tie-breaker unless the business approved that policy. ## Exclude tempting context Exclusions matter because a nearby file can look useful while carrying the wrong authority. Exclude other clients, private roles, expired offers, raw chat exports, unreviewed drafts, and generated examples unless the job needs them. The [task brief and prompt distinction](/articles/brief-vs-prompt) matters here. A runtime prompt should not widen the manifest because the user says "look everywhere" in a copied email or source document. Treat retrieved content as data. It does not get to rewrite the job or add tools. Use default deny for paths that are not listed. A manifest is weak if an agent can search the whole drive after a permitted source comes back empty. ## Set freshness, authority, and conflict rules Freshness depends on the record. A CRM export for a daily pipeline report may need to be less than twenty-four hours old. A signed contract remains current until amended or expired. An operating procedure may require a review date rather than a file age. Define freshness per source. State the evidence used to check it. Authority also needs a current owner. NIST's voluntary AI Risk Management Framework calls for documented roles, ongoing monitoring, clear human oversight, testing, and processes for incorporating adjudicated feedback. [NIST AI RMF Core](https://airc.nist.gov/airmf-resources/airmf/5-sec-core/) The manifest should not let the agent approve its own source change. It can report a stale record and propose a correction. ## Define output, approval, stopping point, and receipt The output contract names the path, required sections, and factual checks. The approval rule separates preparation from external effect. A weekly owner report may be saved locally without review. Sending it to a partner, changing the CRM, or starting a campaign requires a separate capability gate. The stopping rules should cover missing required sources, stale records, unresolved identity, equal-authority conflicts, unsafe output destinations, and failed tests. The receipt records inspected sources, excluded-source checks, output path, status, errors, and proposed next action. This is a reading contract, not a blanket action authorization. ## Inspect `context-manifest.yaml` This example covers one weekly owner report. ```yaml version: 2026.08.19.1 job: id: weekly-owner-report owner: operations_owner scope: business_alpha sources: - id: crm_export path: inputs/crm/current.csv role: current_pipeline_facts required: true max_age: 24h mode: read_only - id: current_strategy path: decisions/strategy-current.md role: approved_strategy required: true authority: approved_decision mode: read_only - id: corrections path: corrections/accepted.md role: accepted_corrections required: true mode: read_only source_order: - accepted_corrections - approved_strategy - current_pipeline_facts exclude: - archive/** - clients/** - drafts/** - secrets/** conflict_rule: stop_on_equal_authority_conflict missing_rule: stop_if_required_source_missing stale_rule: stop_if_source_exceeds_max_age output: path: outputs/weekly-owner-report.md required_sections: [current_state, blockers, next_approved_action, source_list] must_not: [invent_dates, promise_scope, send_externally] approval: local_draft: preapproved external_send: operations_owner live_record_change: record_owner receipt: receipts/weekly-owner-report/{run_id}.json ``` The file is useful because each failure has a visible location. A stale result points to freshness or retrieval. A wrong section points to output. An attempted send points to approval enforcement. ## Add a stale-source rejection fixture Create a copy of the CRM export with a modified time older than the permitted age. Add an obsolete strategy file under `archive/` with wording that closely matches the report request. Then run these tests: 1. Current required sources produce the named output sections. 2. The stale CRM export produces a blocked receipt. 3. The archive file is never read, even when it is a close semantic match. 4. A missing correction file produces a blocked receipt. 5. Two equal-authority strategy records produce a conflict stop. 6. Another client's record never appears in retrieved paths or output. 7. The draft is saved to the exact output path. 8. An external-send request without approval is denied. 9. The receipt lists every source read and the manifest version. 10. A fresh session can reproduce the same status from the same fixture. Do not test only the good input. The manifest earns its keep when the context is tempting, stale, absent, or conflicting. ## Keep the manifest beside the job Store the manifest with the task brief, schema, and tests. Version it with the workflow when practical. Do not bury it in a provider dashboard if the business expects portability. A readable manifest lets another operator inspect the source contract before choosing a new adapter. The job instruction should link to the manifest. The receipt should record its version. The correction process should name whether a change affects a source, precedence, exclusion, or stop rule. ## Version changes by consequence Not every edit needs the same review. A spelling fix can use normal file review. Adding a source, changing precedence, widening an exclusion, or changing an approver alters the job's decision boundary. Record a manifest change note with the old value, new value, reason, owner, and required fixtures. Rerun stale, missing, conflict, cross-client, and denied-action tests after a boundary change. Keep the prior accepted manifest available for diagnosis and rollback. Do not let the agent overwrite the only copy of its reading contract. ## Approval and stopping boundary The agent may read listed sources, validate freshness, prepare the output, and write the receipt. It stops before searching excluded paths, changing source authority, accepting corrections, sending externally, changing live records, or widening the job scope. If the manifest is missing, ambiguous, or internally inconsistent, the correct state is blocked. The agent does not reconstruct authority from file names or past chat. Changing an exclusion, approval owner, output destination, or source precedence requires owner review and a fixture rerun. ## Sources - [Anthropic: Effective context engineering for AI agents](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents) - [NIST: AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework) - [NIST: AI RMF Core](https://airc.nist.gov/airmf-resources/airmf/5-sec-core/) ## Questions about AI Context Manifest Template for One Recurring Business Job ### Based on what criteria would the agent decide what persistent data records are important for the current session and context? The manifest should name the job's canonical sources, source order, exclusions, freshness limits, authority rules, conflict behavior, output, approval, stopping point, and receipt. Retrieval relevance can nominate a record, but the manifest decides whether it is permitted, current, and governing. ### What are you using for AI memory across tools? Keep a tool-independent manifest beside the recurring job and point it to records the business controls. The active model can change, but both tools should receive the same source order, exclusions, freshness rules, stop conditions, and acceptance fixture. ### What's your memory stack? A small recurring job can use readable canonical files, a versioned context manifest, exact-path or keyword retrieval, a correction log, a draft output, and a run receipt. Storage choices should follow the job and tests rather than determine the reading contract. ## Save the visual summary context-manifest.yaml | inputs/crm/current.csv | decisions/strategy-current.md | corrections/accepted.md | outputs/weekly-owner-report.md | receipts/weekly-owner-report [Download the PNG](/infographics/ai-context-manifest-template-1200x1500.png) ## Related: Owned Memory - [AI Correction Log Template That Changes the Next Run](/articles/ai-correction-log-template) - [What Context Should an Agent Read Before It Starts?](/articles/what-context-should-an-agent-read) - [How Corrections Become Part of the System](/articles/corrections-become-system-memory) ---- ```text .|########||. .|########||. .|########||. |##||. .||##|. |##||. .||##|. |##||. .||##|. |#|. .|#|. |#|. .|#|. |#|. .|#|. |#| |#| |#| |#| |#| |#| .#| |#. .#| |#. .#| |#. .#. .#| .#. .#| .#. .#| |#. .#| |#. .#| |#. .#| |# ...... #| |# ...... #| |# ...... #| .# ||#########| #| .# ||#########| #| .# ||#########| #| .#. |######||######|. .#. .#. |######||######|. .#. .#. |######||######|. .#. #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# #. .##|###|##|#|######| .# || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || || |##|#||||||||||||#||#| || # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. # |#||||||||||||||||||||#| #. || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # # .#||||||||||||||||||||||||#| # || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || || ....|||#||||||##|#|||#|#||##||||....|. || #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# #. . ....|# ....#||| ||| .#||| ....#. .# # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. # . ||||#| .#####|| . .#| .#||| ||||# #. .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || .| ||||| #. |#|||||. || #. |#||. .||||| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || || |.... #. ....|#. |. ...|. ....|| || #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# #. .||||||##||||||##||####|||||||#||||||#| .# # .||####|###########################|. # # .||####|###########################|. # # .||####|###########################|. # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#||||||.#.|| # |. ..# #| #|||||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||||| .. |# ## |#| ...#|#|||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#|||#|# .# .#| #| ##|.#..#|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # .#||||||############|######|||||#| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # # |...||#||||||#||||#|#||||||#|||||#||| |.| # #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# #. .. ||||# .|||##|. |#| .|| || .|||#. #| # .# || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || | ...|| ...##| | #| .|. |. ##### .. .| || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || || ||||| || ||||#| . |. |. |#. ||||| |#| || || .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. .# |.....#|....|#.||||.|||##.|#|....||.#||.#. #. #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# #.|||||||#############################| ||| .# || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || || ##||#||#||||||#||#||#||#||##. || # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # # .#|||||||||#||#|||||||||||#| # || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || || |#||||||||||||||||||||||#| || .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. .# |#||||||||||||||||||||#| #. || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || || |#||||||||||||||||||#| || #. .######|#|#########| .# #. .######|#|#########| .# #. .######|#|#########| .# .#. |#####||||#####| .#. .#. |#####||||#####| .#. .#. |#####||||#####| .#. |# ||########|| #. |# ||########|| #. |# ||########|| #. |# ...... #| |# ...... #| |# ...... #| |#. .#| |#. .#| |#. .#| |#. .#. |#. .#. |#. .#. .#| |#. .#| |#. .#| |#. |#| |#| |#| |#| |#| |#| .|#|. .|#|. .|#|. .|#|. .|#|. .|#|. |##||. .||##| |##||. .||##| |##||. .||##| .||########||. .||########||. .||########||. Scale With Search 2026 [scalewithsearch.com](https://scalewithsearch.com) ```