Bounded Loops & Graphs: loops and graphs
69 shipped loops and 7 reference graphs. Browse independent acceptance gates for engineering, finance, content and operations. Default examples and live execution have different requirements.
Choose a task, inspect its requirements and open the linked source before running it. Examples describe declared checks and inputs; they do not certify broader outcomes.
Catalog reviewed 1 October 2026. Source snapshot · Repository
Frontend
Loop
A11y
Check an HTML fixture for the shipped static accessibility rules.
bl loops install a11y --dest ./loops bl run loops/a11y --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Backend
Loop
Adk example
Wires an existing Google Agent Development Kit (ADK) LoopAgent/Runner into bounded-loops, per this repo's own frozen vocabulary (LoopAgent(max_iterations/escalate, deterministic)).
bl loops install adk-example --dest ./loops bl run loops/adk-example --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest. Framework adapter dependencies and provider configuration may be required.
- Status
- Framework example; review setup
- Scope
- A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Autogen example
Wires an existing Microsoft Agent Framework team (AutoGen's converged successor) into bounded-loops.
bl loops install autogen-example --dest ./loops bl run loops/autogen-example --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest. Framework adapter dependencies and provider configuration may be required.
- Status
- Framework example; review setup
- Scope
- A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Bug fix red green
Drive an agent until a failing pytest passes.
bl loops install bug-fix-red-green --dest ./loops bl run loops/bug-fix-red-green --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Crewai example
Wires an existing CrewAI crew into bounded-loops.
bl loops install crewai-example --dest ./loops bl run loops/crewai-example --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest. Framework adapter dependencies and provider configuration may be required.
- Status
- Framework example; review setup
- Scope
- A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Data contract
Drive an agent until a JSON API response conforms to its contract.
bl loops install data-contract --dest ./loops bl run loops/data-contract --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Dead import clean
Drive an agent until a Python module imports nothing it doesn't use.
bl loops install dead-import-clean --dest ./loops bl run loops/dead-import-clean --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Json config schema
Drive an agent until a service's JSON runtime config conforms to its contract.
bl loops install json-config-schema --dest ./loops bl run loops/json-config-schema --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Langgraph example
Wires an existing LangGraph graph into bounded-loops.
bl loops install langgraph-example --dest ./loops bl run loops/langgraph-example --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest. Framework adapter dependencies and provider configuration may be required.
- Status
- Framework example; review setup
- Scope
- A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Openapi schema valid
Drive an agent until an OpenAPI 3 document satisfies the structural minimum every downstream tool (codegen, docs, gateways) assumes: a top-level `openapi` version, `info.title`/`info.version`, and every operation under `paths` declaring a non-empty `responses` object.
bl loops install openapi-schema-valid --dest ./loops bl run loops/openapi-schema-valid --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Type annotations present
Drive an agent until every public top-level function in a Python module fully annotates its parameters and return type.
bl loops install type-annotations-present --dest ./loops bl run loops/type-annotations-present --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Agent Development
Operations
Loop
Alert runbook link
Drive an agent until every alert definition links to a non-empty runbook URL.
bl loops install alert-runbook-link --dest ./loops bl run loops/alert-runbook-link --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Conventional commits
Drive an agent until every commit subject in a list conforms to Conventional Commits (`type(scope): description`).
bl loops install conventional-commits --dest ./loops bl run loops/conventional-commits --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Oncall coverage
Drive an agent until an on-call schedule covers every hour of its rotation period exactly once.
bl loops install oncall-coverage --dest ./loops bl run loops/oncall-coverage --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Runbook completeness
Drive an agent until an operations runbook contains every section an on-call engineer needs during an incident.
bl loops install runbook-completeness --dest ./loops bl run loops/runbook-completeness --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Slo error budget
Drive an agent until a failing pytest passes.
bl loops install slo-error-budget --dest ./loops bl run loops/slo-error-budget --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Content
Loop
Alt text present
Drive an agent until every markdown image in a post has non-empty alt text.
bl loops install alt-text-present --dest ./loops bl run loops/alt-text-present --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Broken internal links
Drive an agent until every relative internal markdown link in a content directory resolves to a real file on disk.
bl loops install broken-internal-links --dest ./loops bl run loops/broken-internal-links --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Content fact gate
Drive an agent until markdown-link-check reports every citation link alive.
bl loops install content-fact-gate --dest ./loops bl run loops/content-fact-gate --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Frontmatter schema
Drive an agent until a blog post's parsed frontmatter conforms to its contract.
bl loops install frontmatter-schema --dest ./loops bl run loops/frontmatter-schema --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Reading level gate
Drive an agent until a post's average words-per-sentence drops to a readable level.
bl loops install reading-level-gate --dest ./loops bl run loops/reading-level-gate --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Seo meta limits
Check a metadata export against the example's configured title and description length limits.
bl loops install seo-meta-limits --dest ./loops bl run loops/seo-meta-limits --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Testing
Loop
Assertion density
Drive an agent until every `test_*` function in a Python test file contains at least one `assert` statement.
bl loops install assertion-density --dest ./loops bl run loops/assertion-density --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
No hardcoded sleep
Drive an agent until no test contains a hardcoded `time.sleep(...)` call.
bl loops install no-hardcoded-sleep --dest ./loops bl run loops/no-hardcoded-sleep --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Test naming contract
Drive an agent until every test-like function in a Python test file is named `test_*` so a pytest-style runner actually collects it.
bl loops install test-naming-contract --dest ./loops bl run loops/test-naming-contract --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Test presence per module
Drive an agent until every module in `seed/src/` has a matching `seed/tests/test_<mod>.py` file.
bl loops install test-presence-per-module --dest ./loops bl run loops/test-presence-per-module --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Enterprise
Loop
Bapi payload contract
Drive an agent until a BAPI_MATERIAL_SAVEDATA-style JSON payload conforms to its contract.
bl loops install bapi-payload-contract --dest ./loops bl run loops/bapi-payload-contract --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Cds view annotations
Drive an agent until a SAP Core Data Services (CDS) view DDL source declares its required annotations: @AccessControl.authorizationCheck, @EndUserText.label, and @Metadata.allowExtensions.
bl loops install cds-view-annotations --dest ./loops bl run loops/cds-view-annotations --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Idoc xml schema
Drive an agent until a simplified SAP IDoc (Intermediate Document) XML file is structurally valid: a control segment (EDI_DC40 with MANDT/DOCNUM/MESTYP), a header segment (E1EDK01), and at least one item segment (E1EDP01) carrying a non-empty quantity (MENGE).
bl loops install idoc-xml-schema --dest ./loops bl run loops/idoc-xml-schema --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Material master completeness
Drive an agent until a SAP material master record carries every field its material type requires, mirroring SAP's material-type-driven field selection (OMS9/MM01): a FERT (finished product) requires non-empty base_uom, description, and division; a ROH (raw material) requires base_uom and description.
bl loops install material-master-completeness --dest ./loops bl run loops/material-master-completeness --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Transport request manifest
Drive an agent until a SAP change-and-transport-system (CTS) transport request manifest has no dangling dependencies: every id in `dependencies` must also appear in `objects`.
bl loops install transport-request-manifest --dest ./loops bl run loops/transport-request-manifest --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Research
Loop
Bibliography completeness
Drive an agent until every pandoc-style inline citation key `[@key]` in a paper's body appears as a listed key under its own `## References` section.
bl loops install bibliography-completeness --dest ./loops bl run loops/bibliography-completeness --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Claim source mapping
Drive an agent until every inline citation marker `[S#]` in a research article resolves to a real source id in a trusted source list.
bl loops install claim-source-mapping --dest ./loops bl run loops/claim-source-mapping --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Dataset license allowed
Drive an agent until every dataset in a training manifest carries a license that is in a trusted allowlist.
bl loops install dataset-license-allowed --dest ./loops bl run loops/dataset-license-allowed --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Reproducibility manifest
Drive an agent until a research run's reproducibility manifest conforms to its contract.
bl loops install reproducibility-manifest --dest ./loops bl run loops/reproducibility-manifest --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Retail
Loop
Catalog required fields
Drive an agent until a retail catalog entry carries all of its required display fields.
bl loops install catalog-required-fields --dest ./loops bl run loops/catalog-required-fields --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Gtin checkdigit
Drive an agent until every product's GTIN-13 has a valid mod-10 check digit.
bl loops install gtin-checkdigit --dest ./loops bl run loops/gtin-checkdigit --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Inventory nonnegative
Drive an agent until a sequence of inventory movements never oversells a SKU below zero.
bl loops install inventory-nonnegative --dest ./loops bl run loops/inventory-nonnegative --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Price margin floor
Drive an agent until every SKU in a retail catalog is priced at or above its minimum-margin policy floor.
bl loops install price-margin-floor --dest ./loops bl run loops/price-margin-floor --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Product feed schema
Drive an agent until a Google-Merchant-style product feed item conforms to its contract.
bl loops install product-feed-schema --dest ./loops bl run loops/product-feed-schema --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Security
Loop
Checkov example
Drive an agent until checkov reports zero failed infrastructure-as-code checks.
bl loops install checkov-example --dest ./loops bl run loops/checkov-example --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Cors not wildcard
Drive an agent until `seed/security_config.json` never combines a wildcard CORS origin with credentialed requests.
bl loops install cors-not-wildcard --dest ./loops bl run loops/cors-not-wildcard --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Dependency pinning
Drive an agent until every dependency in `seed/requirements.txt` is pinned to an exact version.
bl loops install dependency-pinning --dest ./loops bl run loops/dependency-pinning --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Dockerfile no root
Drive an agent until `seed/Dockerfile` pins its base image and drops root before CMD.
bl loops install dockerfile-no-root --dest ./loops bl run loops/dockerfile-no-root --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Jwt alg not none
Drive an agent until `seed/auth_config.json` never sets the JWT signing algorithm to `none`.
bl loops install jwt-alg-not-none --dest ./loops bl run loops/jwt-alg-not-none --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Osv scanner example
Drive an agent until osv-scanner reports no known vulnerabilities.
bl loops install osv-scanner-example --dest ./loops bl run loops/osv-scanner-example --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Secret scan keyless
Drive an agent until `seed/app_config.py` contains no hardcoded secrets.
bl loops install secret-scan-keyless --dest ./loops bl run loops/secret-scan-keyless --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Legal
Loop
Citation existence check
Check example case citations against a fixed reporter fixture list.
bl loops install citation-existence-check --dest ./loops bl run loops/citation-existence-check --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The gate compares citations with a fixed local fixture; it does not query a live reporter or establish legal validity.
Loop
Contract clause extraction
Drive an agent until extracted contract clauses conform to a required-clauses schema.
bl loops install contract-clause-extraction --dest ./loops bl run loops/contract-clause-extraction --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Contract defined terms
Drive an agent until every bolded defined term used in a contract's body is actually defined in its "## Definitions" section.
bl loops install contract-defined-terms --dest ./loops bl run loops/contract-defined-terms --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
DPA keyword coverage
Check an example agreement for nine declared keyword categories.
bl loops install gdpr-dpa-terms --dest ./loops bl run loops/gdpr-dpa-terms --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- Keyword coverage does not establish GDPR Article 28(3) compliance or the legal adequacy of an agreement.
Loop
Nda required clauses
Drive an agent until a mutual Non-Disclosure Agreement contains every clause a working NDA needs: Confidentiality, Term/Duration, Governing Law, Return of Materials, and Permitted Disclosures.
bl loops install nda-required-clauses --dest ./loops bl run loops/nda-required-clauses --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Privacy policy completeness
Drive an agent until a privacy policy contains every section a working policy needs: Data We Collect, How We Use Your Data, Data Sharing, Data Retention, Your Rights, and Contact.
bl loops install privacy-policy-completeness --dest ./loops bl run loops/privacy-policy-completeness --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Healthcare
Loop
Clinical note completeness
Check the example note for declared SOAP fields and an ICD-10 code pattern.
bl loops install clinical-note-completeness --dest ./loops bl run loops/clinical-note-completeness --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- These are structural field and pattern checks on a synthetic example, not an assessment of clinical correctness or medical safety.
Engineering
Loop
Convergence demo
Demonstrate a bounded evaluator-optimizer loop across three genuine gate verdicts: a wrong fix, a partial fix, and a correct fix.
bl loops install convergence-demo --dest ./loops bl run loops/convergence-demo --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Finance
Loop
Expense policy compliance
Drive an agent until every line of an expense report complies with the company's expense policy: allowed category and within its per-line cap.
bl loops install expense-policy-compliance --dest ./loops bl run loops/expense-policy-compliance --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Fx rate sanity
Drive an agent until a table of FX rates is internally consistent: every rate is positive, the base-to-base rate is 1, and every inverse pair multiplies to ~1 (no-arbitrage).
bl loops install fx-rate-sanity --dest ./loops bl run loops/fx-rate-sanity --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Invoice 3way match
Drive an agent until an invoice's every line matches its purchase order (unit price) and its goods receipt (quantity) — the classic accounts-payable three-way match.
bl loops install invoice-3way-match --dest ./loops bl run loops/invoice-3way-match --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Iso20022 payment valid
Drive an agent until a simplified ISO 20022 pain.001 credit-transfer instruction conforms to its contract.
bl loops install iso20022-payment-valid --dest ./loops bl run loops/iso20022-payment-valid --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Journal entries balance
Drive an agent until every entry in a double-entry journal balances — sum(debit) == sum(credit) per entry.
bl loops install journal-entries-balance --dest ./loops bl run loops/journal-entries-balance --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Ledger reconciliation
Drive an agent until a double-entry ledger reconciles: debits equal credits AND every transaction is categorized.
bl loops install ledger-reconciliation --dest ./loops bl run loops/ledger-reconciliation --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools. Install pytest.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Business
Loop
Meeting action items
Drive an agent until every action item in meeting minutes names an owner and a due date.
bl loops install meeting-action-items --dest ./loops bl run loops/meeting-action-items --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Okr measurable
Drive an agent until every key result in an OKR document is genuinely measurable.
bl loops install okr-measurable --dest ./loops bl run loops/okr-measurable --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Prd acceptance criteria
Drive an agent until every user story in a PRD has real, testable acceptance criteria.
bl loops install prd-acceptance-criteria --dest ./loops bl run loops/prd-acceptance-criteria --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Rfc decision recorded
Drive an agent until an RFC document actually records a decision, not just a problem statement.
bl loops install rfc-decision-recorded --dest ./loops bl run loops/rfc-decision-recorded --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Loop
Roadmap field contract
Drive an agent until a product roadmap item conforms to its field contract.
bl loops install roadmap-field-contract --dest ./loops bl run loops/roadmap-field-contract --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Quality
Loop
Record completeness
Drive an agent until every record in a JSON dataset carries a checksum field.
bl loops install record-completeness --dest ./loops bl run loops/record-completeness --yes- Requirements
- Python 3.11+ and bounded-loops. Review the linked package for its gate tools.
- Status
- Keyless example
- Scope
- The default runner uses a recorded or scripted example; it is not a live model benchmark. A passing gate covers its declared checks; it does not establish general correctness, legal compliance or model accuracy.
Graph pipelines
Graph
Customer data request
A customer data-subject request clears processor terms, privacy-notice completeness and a secret scan in parallel before a privacy officer authorises the response. A failed terms check routes to clause extraction so the gap can be named.
bl graph lint graphs/customer-data-request/graph.yaml bl graph plan graphs/customer-data-request/graph.yaml- Requirements
- Clone the linked source snapshot and install bounded-loops; these commands validate and plan the manifest.
- Status
- Reference graph; lint and plan example
- Scope
- Planning does not execute a graph. Real execution requires admitted packages, runner configuration and OS-level isolation. Default examples use stubs and mechanical gates.
Graph
Engineering release gate
A release candidate clears dependency pinning, CORS posture and container-user checks in parallel before a release manager cuts the tag. A failed CORS check routes to a secret scan, on the reasoning that a permissive origin policy rarely travels alone.
bl graph lint graphs/engineering-release-gate/graph.yaml bl graph plan graphs/engineering-release-gate/graph.yaml- Requirements
- Clone the linked source snapshot and install bounded-loops; these commands validate and plan the manifest.
- Status
- Reference graph; lint and plan example
- Scope
- Planning does not execute a graph. Real execution requires admitted packages, runner configuration and OS-level isolation. Default examples use stubs and mechanical gates.
Graph
Finance payment assurance
Four independent accounting checks run in parallel — three-way match, balanced journal, FX sanity and ISO 20022 validity — then join and gate the payment instruction behind a controller's approval. A failed balance check routes to reconciliation instead of stopping the run.
bl graph lint graphs/finance-payment-assurance/graph.yaml bl graph plan graphs/finance-payment-assurance/graph.yaml- Requirements
- Clone the linked source snapshot and install bounded-loops; these commands validate and plan the manifest.
- Status
- Reference graph; lint and plan example
- Scope
- Planning does not execute a graph. Real execution requires admitted packages, runner configuration and OS-level isolation. Default examples use stubs and mechanical gates.
Graph
Marketing campaign release
Campaign copy clears SEO limits, reading level and image alt-text in parallel, then an editor approves publication. Missing alt text routes to claim-source mapping, since both are evidence gaps in the same asset, instead of killing the campaign.
bl graph lint graphs/marketing-campaign-release/graph.yaml bl graph plan graphs/marketing-campaign-release/graph.yaml- Requirements
- Clone the linked source snapshot and install bounded-loops; these commands validate and plan the manifest.
- Status
- Reference graph; lint and plan example
- Scope
- Planning does not execute a graph. Real execution requires admitted packages, runner configuration and OS-level isolation. Default examples use stubs and mechanical gates.
Graph
Operations incident gate
An SRE incident-response release clears SLO error-budget, on-call coverage and runbook completeness in parallel before a site-reliability lead approves the incident plan. A failed on-call check routes to alert-runbook-link verification; if that verification also fails, a single repair round re-runs the on-call check and its descendants so the graph retries rather than sealing a potentially-transient gap.
bl graph lint graphs/operations-incident-gate/graph.yaml bl graph plan graphs/operations-incident-gate/graph.yaml- Requirements
- Clone the linked source snapshot and install bounded-loops; these commands validate and plan the manifest.
- Status
- Reference graph; lint and plan example
- Scope
- Planning does not execute a graph. Real execution requires admitted packages, runner configuration and OS-level isolation. Default examples use stubs and mechanical gates.
Graph
Retail listing release
A product listing clears barcode, feed-schema and margin checks in parallel before a merchandiser releases it to the storefront. A failed margin check routes to a stock review rather than stopping the release pipeline.
bl graph lint graphs/retail-listing-release/graph.yaml bl graph plan graphs/retail-listing-release/graph.yaml- Requirements
- Clone the linked source snapshot and install bounded-loops; these commands validate and plan the manifest.
- Status
- Reference graph; lint and plan example
- Scope
- Planning does not execute a graph. Real execution requires admitted packages, runner configuration and OS-level isolation. Default examples use stubs and mechanical gates.
Graph
Solo builder ship
A solo builder's change clears test presence, type annotations and commit convention in parallel before they approve their own ship — the commit history is what the release notes are generated from. A missing test routes to a red-green fix loop.
bl graph lint graphs/solo-builder-ship/graph.yaml bl graph plan graphs/solo-builder-ship/graph.yaml- Requirements
- Clone the linked source snapshot and install bounded-loops; these commands validate and plan the manifest.
- Status
- Reference graph; lint and plan example
- Scope
- Planning does not execute a graph. Real execution requires admitted packages, runner configuration and OS-level isolation. Default examples use stubs and mechanical gates.
Explore the Qualixar catalog and ask questions in the community hub.