Starintel 4th generation Ideas
Tasks
Step one
Pick a single topic
Step 2
Preform Deep Research on said topic Provide code example
Sort Design Files
Ok can we sort the design by like project? I want you to optmize the research until it is very advanced. git track them after you move them to the right dir structure
Design File requirments
All Design files will Use a numbering system Ignore the , which is org-mode escaping
#+title: STAR-00 <short title> ,| Version | Date | Description of change | Did nsaspy aprove it | ,|---+---+---+---| ,| | | | | * Design File System * Foot Notes * Citations
Agentic Design File scripts
write a script that a agent can write and forbid agent from context hogging all the files Write a org-capture system for the agents to mantain this database of research it is going to be a org-roam system so use org roam rules. design/star-server/<numbered-org-files> is an example i want you to iteravly look up other projects (git cloning if you must) and do deep research on k
Research Topics
Use org trackers to matain state
1. other osint project features
2. Actors Needed for Collection
a. FEC Data Actors
Collect and normalize public FEC datasets, filings, committees, candidates, donors, expenditures, independent expenditures, and related records.
b. Melissa Data Actor
Integrate Melissa services for identity, address, telephone, email, property, demographic, and entity enrichment.
c. Telephone Number Actors
Collect, validate, normalize, and enrich telephone-number data through public, account, provider, organization, or acquired sources.
d. Bug-Bounty Tool Actors
Implement bug-bounty actors for targets in scope of the applicable program, customer, organization, or asset owner.
Evaluate extending the existing BBPD service.
e. Discord Self-Bot Collection Actors
- TODO Collection capabilities
- Account and session handling
- Guild traversal
- Channel traversal
- Thread traversal
- Message collection
- Reply and reaction collection
- Member and role collection
- Attachment and embed collection
- Audit-log ingestion
- Deleted-content ingestion when supplied by Discord
- Incremental checkpoints
- Historical backfill
- Content preservation
- Account and session handling
- TODO Coordination requirements
- Record Discord access reference
- Record approved accounts
- Record approved guilds and channels
- Enforce collection boundaries
- Respect Discord-supplied rate limits
- Support access expiration
- Preserve Discord-supplied data packages
- Produce complete audit logs
- Disable collection automatically when access expires
- Record Discord access reference
f. Telegram Collection Actors
Support collection through public Telegram sources, account sessions, organization-controlled accounts, Telegram access, or acquired data.
- TODO Collection capabilities
- Public channel collection
- Public group collection
- Private-group collection
- Account history collection
- Message and reply collection
- Topic and thread collection
- Member and administrator metadata
- Forwarded-message relationships
- Reactions and polls
- Media and attachment collection
- Edit and deletion tracking
- Invite-link metadata
- Incremental synchronization
- Public channel collection
- TODO Client support
- TODO Access controls
g. Signal Messenger Collection Actors
Support collection from account exports, organization-controlled devices, platform-coordinated data, or acquired sources.
- TODO Signal Mobile Actor
Ingest Signal artifacts extracted from Android and iOS devices.
- TODO Signal Desktop Actor
Parse Signal Desktop databases, configuration, attachments, account metadata, and locally retained history.
- TODO Signal Backup Actor
Detect, validate, decrypt when keys or passphrases are available, and ingest supported Signal backup formats.
- TODO Signal Image Actor
Locate Signal-related files and databases inside mounted images without modifying the source.
- TODO Signal Attachment Actor
Extract images, video, audio, documents, stickers, contact cards, and other attachments while preserving message relationships.
- TODO Signal Conversation Actor
Normalize recovered conversations, groups, calls, reactions, replies, edits, deletions, and disappearing-message remnants.
- TODO Signal Identity Actor
Correlate account identifiers, usernames, phone numbers, profile keys, device records, contacts, and existing StarIntel identities.
- TODO Signal Group Actor
Model group membership, administrators, membership changes, titles, avatars, invitations, and related data.
- TODO Signal Timeline Actor
Construct an ordered timeline across messages, calls, attachments, device events, and account changes.
h. News-Ingestion Actors
Collect news through public feeds, licensed services, publisher APIs, organization subscriptions, platform access, and supplied archives.
i. Scrapy Spider Actors
Use for public-source collection, organization sites, platform projects, or sites whose operators have approved collection.
Primary targets include public NGO membership lists, directories, registries, reports, and similar structured sources.
j. Playwright Actor
Complement Scrapy and other actors for JavaScript-heavy sites, authenticated workflows, rendered content, and browser interaction.
k. Proxy-Rotation Actor
Manage proxies supplied or approved by the operating organization, customer, platform, or collection provider.
l. Expert-System Actor Framework
Allow rule engines and expert systems to operate as first-class StarIntel actors.
m. Mastodon Actors
Support public Mastodon data, account access, instance access, organization-controlled instances, and supplied exports.
- TODO Capabilities
- Instance discovery
- Public timeline collection
- Account collection
- Status and reply collection
- Boost and favorite relationships
- Hashtag collection
- Media collection
- Profile collection
- Follower and following collection where available
- Moderation-event ingestion when supplied by an instance
- Instance-block and federation metadata
- Incremental synchronization
- Instance discovery
- TODO Access controls
n. Bluesky Actors
Support public Bluesky and AT Protocol data, account access, service access, organization-controlled infrastructure, and supplied exports.
o. 4chan and 8kun.top Actors
Support public-source collection and platform collection from 4chan, 8kun.top, and compatible imageboard software.
p. Reddit Collection Subsystem
Build a Reddit collection subsystem using star-router as the actor library.
Support public Reddit data, official API access, account access, Reddit access, organization-controlled communities, licensed datasets, and supplied exports.
q. LinkedIn Actors
Support LinkedIn access, account-owner exports, organization-controlled company data, licensed LinkedIn products, or supplied records.
- TODO Supported sources
- TODO Capabilities
- Profile ingestion
- Company-page ingestion
- Employment-history normalization
- Education-history normalization
- Job-posting ingestion
- Organization and person resolution
- Connection and relationship ingestion when available
- Post and article ingestion when available
- Attachment and media ingestion
- Incremental synchronization
- Profile ingestion
- TODO Access controls
r. Shared Collection-Actor Requirements
- TODO Common actor features
- Typed input and output documents
- Actor manifests
- Dataset manifests
- Capability discovery
- Dataflow integration
- Pagination
- Incremental checkpoints
- Rate limiting
- Backoff
- Bounded concurrency
- Target locks
- Provenance
- Source timestamps
- Raw-response preservation
- Deduplication
- Attachment handling
- Authentication isolation
- Credential revocation
- Health checks
- Mock transports
- Replay transports
- Persistent documents
- Transient documents
- Structured errors
- Dead-letter handling
- Audit logs
- Typed input and output documents
3. Spec extensions
expand all fields to cover as much metadata as possible
- TODO Stable document identifiers
- TODO Dataset identifiers
- TODO Document type identifiers
- TODO Schema identifiers
- TODO Schema version
- TODO Document version
- TODO Revision number
- TODO Creation timestamp
- TODO Ingestion timestamp
- TODO Observation timestamp
- TODO Publication timestamp
- TODO Last-seen timestamp
- TODO Last-updated timestamp
- TODO Expiration timestamp
- TODO Collector identity
- TODO Actor identity
- TODO Source identity
- TODO Source URL
- TODO Source mirror URLs
- TODO Archive URLs
- TODO Source access method
- TODO Source account
- TODO Source platform
- TODO Source dataset
- TODO Author
- TODO Publisher
- TODO Organization
- TODO Language
- TODO Locale
- TODO Country
- TODO Region
- TODO Timezone
- TODO Geographic coordinates
- TODO Confidence
- TODO Reliability
- TODO Relevance
- TODO Priority
- TODO Sensitivity
- TODO Classification
- TODO Access-control labels
- TODO Retention policy
- TODO Expiration policy
- TODO Deletion policy
- TODO Legal hold
- TODO Chain of custody
- TODO Provenance
- TODO Transformation history
- TODO Parent documents
- TODO Child documents
- TODO Related documents
- TODO Superseded documents
- TODO Duplicate documents
- TODO Near-duplicate documents
- TODO Content hashes
- TODO Fingerprints
- TODO Tags
- TODO Topics
- TODO Entities
- TODO Events
- TODO Claims
- TODO Relationships
- TODO Attachments
- TODO Citations
- TODO Processing state
- TODO Validation state
- TODO Review state
- TODO Error records
- TODO Warning records
- TODO Human edits
- TODO Human approvals
- TODO Actor configuration
- TODO Model metadata
- TODO Prompt metadata
- TODO Tool metadata
- TODO Pipeline metadata
- TODO Extension namespaces
- TODO Dataset-specific fields
a. unstructured docs
- TODO Plain text
- TODO Org
- TODO Markdown
- TODO HTML
- TODO XML
- TODO JSON
- TODO JSONL
- TODO CSV
- TODO Email
- TODO Chat logs
- TODO PDFs
- TODO Office documents
- TODO Reports
- TODO Transcripts
- TODO Scraped pages
- TODO OCR documents
- TODO Scanned documents
- TODO Preserve the original document
- TODO Preserve normalized text
- TODO Preserve extracted text
- TODO Preserve document structure
- TODO Headings
- TODO Sections
- TODO Paragraphs
- TODO Lists
- TODO Tables
- TODO Citations
- TODO Page numbers
- TODO Character offsets
- TODO Byte offsets
- TODO Bounding boxes
- TODO Source spans
- TODO Extracted entities
- TODO Extracted claims
- TODO Extracted events
- TODO Extracted quotations
- TODO Extracted relationships
- TODO Document summaries
- TODO Section summaries
- TODO Chunk summaries
- TODO Multiple translations
- TODO Redacted variants
- TODO Public-safe variants
- TODO Parsing errors
- TODO Partially processed documents
- TODO Chunking without losing hierarchy
- TODO Every derived chunk links to its original source span
- TODO Store embeddings separately from canonical document data
b. custom view code (with security mitigation)
- TODO Dataset-specific renderers
- TODO Document-specific renderers
- TODO Table views
- TODO Graph views
- TODO Timeline views
- TODO Map views
- TODO Gallery views
- TODO Evidence views
- TODO Declarative views
- TODO Server-side views
- TODO Client-side views
- TODO Common Lisp views
- TODO JavaScript views
- TODO WebAssembly views
- TODO Sandboxed execution
- TODO Capability-based permissions
- TODO Read-only data access by default
- TODO No network access by default
- TODO No filesystem access by default
- TODO No process execution by default
- TODO No secret access by default
- TODO CPU limits
- TODO Memory limits
- TODO Execution-time limits
- TODO Output-size limits
- TODO Dependency allowlists
- TODO Dependency lockfiles
- TODO Signed view packages
- TODO Versioned view packages
- TODO Static analysis before installation
- TODO Content Security Policy
- TODO HTML sanitization
- TODO Output escaping
- TODO Per-view field allowlists
- TODO Per-view dataset allowlists
- TODO Audit every execution
- TODO Administrative approval
- TODO Emergency disable controls
- TODO Safe fallback renderer
- TODO Deterministic rendering where possible
c. attachments and files
- TODO Attach files to documents
- TODO Attach files to entities
- TODO Attach files to events
- TODO Attach files to cases
- TODO Original filename
- TODO Normalized filename
- TODO MIME type
- TODO Detected file type
- TODO File size
- TODO Cryptographic hashes
- TODO Fuzzy hashes
- TODO Content-addressed storage
- TODO Deduplication
- TODO Versioning
- TODO Revision history
- TODO Archive support
- TODO Compressed-file support
- TODO Image previews
- TODO Audio previews
- TODO Video previews
- TODO Document previews
- TODO Source-code previews
- TODO Thumbnail generation
- TODO Derivative generation
- TODO Text extraction
- TODO Metadata extraction
- TODO Embedding extraction
- TODO Malware scanning
- TODO Quarantine
- TODO Encryption at rest
- TODO Per-file access control
- TODO Retention policy
- TODO Deletion policy
- TODO Legal holds
- TODO Evidence-preservation flags
- TODO Chain-of-custody records
- TODO External object-storage references
- TODO Multipart uploads
- TODO Resumable uploads
- TODO Download limits
- TODO Bandwidth limits
- TODO Inline attachment references
- TODO Attachment relationships
- TODO Verify integrity during storage
- TODO Verify integrity during retrieval
d. support for expert systems
- TODO Facts
- TODO Rules
- TODO Predicates
- TODO Frames
- TODO Ontologies
- TODO Prolog
- TODO LISA
- TODO CLIPS-compatible systems
- TODO Datalog
- TODO Rete-based systems
- TODO Custom Common Lisp rule systems
- TODO Dataset-to-fact transformation rules
- TODO Namespaced predicates
- TODO Namespaced vocabularies
- TODO Forward chaining
- TODO Backward chaining
- TODO Explanation traces
- TODO Proof traces
- TODO Confidence-weighted facts
- TODO Contradiction detection
- TODO Conflict detection
- TODO Temporal facts
- TODO Interval reasoning
- TODO Geospatial facts
- TODO Relationship reasoning
- TODO Rule versioning
- TODO Rule migration
- TODO Rule activation
- TODO Rule deactivation
- TODO Incremental fact updates
- TODO Materialized conclusions
- TODO Separate observed facts
- TODO Separate asserted facts
- TODO Separate inferred facts
- TODO Human confirmation
- TODO Fact correction
- TODO Query budgets
- TODO Recursion limits
- TODO Inference-time limits
- TODO Sandboxed custom predicates
- TODO Rule dependency graphs
- TODO Dataset dependency graphs
- TODO Export conclusions back into documents
- TODO Store exact rules used for each conclusion
- TODO Store exact evidence used for each conclusion
e. SIMHASH/MINHASH/JARM/custom finger print
- TODO SimHash
- TODO MinHash
- TODO TLSH
- TODO ssdeep
- TODO JARM
- TODO JA3
- TODO JA3S
- TODO JA4
- TODO HTTP fingerprints
- TODO Header-order fingerprints
- TODO DOM fingerprints
- TODO Favicon hashes
- TODO Certificate fingerprints
- TODO SSH host-key fingerprints
- TODO Service-banner fingerprints
- TODO File-structure fingerprints
- TODO Image perceptual hashes
- TODO Audio fingerprints
- TODO Video fingerprints
- TODO Source-code fingerprints
- TODO Dataset-defined fingerprint algorithms
- TODO Fingerprint algorithm version
- TODO Fingerprint parameters
- TODO Fingerprint confidence
- TODO Fingerprint provenance
- TODO Multiple fingerprints per document
- TODO Fingerprint comparison records
- TODO Threshold configuration
- TODO Approximate nearest-neighbor indexes
f. Uiniqly Fingerprint a document
- TODO Canonical document representation
- TODO Canonical field ordering
- TODO Canonical whitespace normalization
- TODO Canonical Unicode normalization
- TODO Canonical timestamp normalization
- TODO Canonical URL normalization
- TODO Canonical attachment ordering
- TODO Canonical relationship ordering
- TODO Exclude volatile metadata
- TODO Include schema version
- TODO Include document type
- TODO Include dataset namespace
- TODO Strong cryptographic document hash
- TODO Stable semantic fingerprint
- TODO Structural fingerprint
- TODO Content fingerprint
- TODO Source fingerprint
- TODO Attachment-set fingerprint
- TODO Revision fingerprint
- TODO Cross-format equivalence detection
- TODO Exact-duplicate detection
- TODO Near-duplicate detection
- TODO Collision handling
- TODO Fingerprint migration
- TODO Fingerprint verification
- TODO Fingerprint audit history
h. Speficic Stable id for people if entity too brad
- TODO Person-specific stable identifier
- TODO Never use a display name as the stable identifier
- TODO Preserve all aliases
- TODO Preserve former names
- TODO Preserve usernames
- TODO Preserve telephone numbers
- TODO Preserve email addresses
- TODO Preserve account identifiers
- TODO Preserve external provider identifiers
- TODO Preserve government identifiers only in restricted fields
- TODO Preserve source-specific identifiers
- TODO Identifier namespace
- TODO Identifier type
- TODO Identifier issuer
- TODO Identifier validity interval
- TODO Identifier confidence
- TODO Identifier provenance
- TODO Identity-resolution evidence
- TODO Identity-resolution confidence
- TODO Competing identity hypotheses
- TODO Reversible entity merges
- TODO Entity split history
- TODO Entity merge history
- TODO Never destroy source records during a merge
- TODO Person-to-account relationships
- TODO Person-to-organization relationships
- TODO Person-to-location relationships
- TODO Person-to-device relationships
- TODO Temporal identity changes
j. Confidence, reliability, relevance, and priority scores
- TODO Separate every score
- TODO Score range definitions
- TODO Confidence in extracted value
- TODO Confidence in entity resolution
- TODO Confidence in inferred relationship
- TODO Confidence in expert-system conclusion
- TODO Source reliability
- TODO Collector reliability
- TODO Actor reliability
- TODO Model reliability
- TODO Document relevance
- TODO Field relevance
- TODO Case relevance
- TODO Search-result relevance
- TODO Processing priority
- TODO Review priority
- TODO Alert priority
- TODO Severity
- TODO Urgency
- TODO Score provenance
- TODO Scoring algorithm
- TODO Scoring algorithm version
- TODO Scoring inputs
- TODO Human overrides
- TODO Override reason
- TODO Score history
- TODO Calibration data
- TODO Uncertainty intervals
- TODO Missing-data indicators
- TODO Do not collapse unrelated scores into one number
k. Schema version and migration history
- TODO Version every schema
- TODO Version every document type
- TODO Version every dataset manifest
- TODO Version every actor manifest
- TODO Forward compatibility
- TODO Backward compatibility
- TODO Schema registry
- TODO Migration registry
- TODO Automated migrations
- TODO Manual migrations
- TODO Dry-run migrations
- TODO Per-dataset migration state
- TODO Per-document migration state
- TODO Migration checkpoints
- TODO Migration rollback
- TODO Migration validation
- TODO Migration error records
- TODO Migration provenance
- TODO Migration code version
- TODO Deprecated-field lifecycle
- TODO Field rename history
- TODO Field split history
- TODO Field merge history
- TODO Type-change history
- TODO Preserve unknown extension fields
- TODO Preserve original pre-migration document
- TODO Generate migration reports
i. Facial Reconigition Support
- TODO Face detection
- TODO Face embeddings
- TODO Face clustering
- TODO Face comparison
- TODO Face-search actors
- TODO Image ingestion
- TODO Video-frame ingestion
- TODO Keyframe extraction
- TODO Multiple faces per frame
- TODO Bounding boxes
- TODO Landmarks
- TODO Pose
- TODO Occlusion
- TODO Image quality
- TODO Detection confidence
- TODO Match confidence
- TODO Model identity
- TODO Model version
- TODO Embedding version
- TODO Comparison threshold
- TODO Candidate rankings
- TODO Human review state
- TODO Confirmed match state
- TODO Rejected match state
- TODO Unknown match state
- TODO Reference-image provenance
- TODO Source-image provenance
- TODO Frame timestamp
- TODO Track identity across video frames
- TODO Link faces to person hypotheses
- TODO Preserve competing person hypotheses
- TODO Facial-recognition actor support
- TODO Vector indexes partitioned by dataset
- TODO Support replacing models without destroying old embeddings
l. Expansive Support For Bug bounty, recon, and cyber security document types
- TODO Program
- TODO Scope
- TODO Target
- TODO Domain
- TODO Subdomain
- TODO IP address
- TODO CIDR
- TODO ASN
- TODO URL
- TODO Endpoint
- TODO Service
- TODO Port
- TODO Protocol
- TODO DNS record
- TODO Certificate
- TODO TLS configuration
- TODO HTTP response
- TODO HTTP request
- TODO WebSocket traffic
- TODO Proxy traffic
- TODO Screenshot
- TODO Technology
- TODO Product
- TODO Product version
- TODO CPE
- TODO CVE
- TODO CWE
- TODO CAPEC
- TODO Vulnerability
- TODO Finding
- TODO Evidence
- TODO Reproduction steps
- TODO Exploitability
- TODO Impact
- TODO Severity
- TODO CVSS
- TODO EPSS
- TODO KEV status
- TODO Remediation
- TODO Retest result
- TODO Duplicate finding
- TODO False positive
- TODO Accepted risk
- TODO Disclosure status
- TODO Bounty submission
- TODO Program response
- TODO Tool execution
- TODO Command provenance
- TODO Scan
- TODO Scan configuration
- TODO Scan result
- TODO Wordlist
- TODO Payload
- TODO Request template
- TODO Response template
- TODO Authentication material reference
- TODO Secret finding
- TODO Credential finding
- TODO Cloud asset
- TODO Container image
- TODO Package
- TODO Dependency
- TODO Repository
- TODO Commit
- TODO Branch
- TODO Pull request
- TODO Issue
- TODO Build
- TODO CI run
- TODO Artifact
- TODO SBOM
- TODO Malware sample
- TODO IOC
- TODO TTP
- TODO ATT&CK technique
- TODO Campaign
- TODO Threat actor
- TODO Incident
- TODO Timeline event
- TODO Scope validation state
- TODO Target lock
- TODO Rate-limit state
- TODO Raw tool output
- TODO Normalized tool output
- TODO Structured parser errors
- TODO Tool version
- TODO Tool configuration hash
- TODO Complete audit trail
m. Entity and relationship model
- TODO People
- TODO Organizations
- TODO Accounts
- TODO Locations
- TODO Devices
- TODO Documents
- TODO Events
- TODO Cases
- TODO Typed relationships
- TODO Directional relationships
- TODO Relationship confidence
- TODO Relationship provenance
- TODO Relationship evidence
- TODO Alias history
- TODO Identity-resolution history
- TODO Merge history
- TODO Split history
- TODO Competing relationship hypotheses
- TODO Temporal relationship validity
- TODO Dataset-specific entity extensions
- TODO Graph traversal limits
n. Event model
- TODO Point-in-time events
- TODO Interval events
- TODO Estimated timestamps
- TODO Uncertain timestamps
- TODO Recurring events
- TODO Participants
- TODO Locations
- TODO Causes
- TODO Outcomes
- TODO Evidence
- TODO Event correlation
- TODO Event deduplication
- TODO Timelines
- TODO Source disagreement
- TODO Derived events
- TODO Event severity
- TODO Event priority
- TODO Document links
- TODO Entity links
- TODO Attachment links
o. Provenance and lineage
- TODO Record every collector
- TODO Record every transformation
- TODO Preserve original representations
- TODO Preserve derived representations
- TODO Actor version
- TODO Tool version
- TODO Model version
- TODO Prompt version
- TODO Code version
- TODO Configuration version
- TODO Input hashes
- TODO Output hashes
- TODO Reproducible processing runs
- TODO Manual edit records
- TODO Source deletion records
- TODO Source disappearance records
- TODO Broken-source state
- TODO Unverifiable-source state
- TODO Provenance graph queries
- TODO Export complete evidence bundles
p. Search and retrieval
- TODO Full-text search
- TODO Structured field search
- TODO Graph search
- TODO Semantic vector search
- TODO Hybrid ranking
- TODO Temporal search
- TODO Geospatial search
- TODO Exact source-span retrieval
- TODO Saved queries
- TODO Query explanations
- TODO Per-dataset indexes
- TODO Index versioning
- TODO Index rebuild state
- TODO Search-result confidence
- TODO Search-result provenance
- TODO Permission-aware indexing
q. Validation and data quality
- TODO Required fields
- TODO Optional fields
- TODO Computed fields
- TODO Forbidden fields
- TODO Type validation
- TODO Range validation
- TODO Format validation
- TODO Relationship validation
- TODO Dataset-specific validation
- TODO Duplicate detection
- TODO Near-duplicate detection
- TODO Conflicting-value detection
- TODO Missing-source detection
- TODO Staleness indicators
- TODO Human-review queues
- TODO Quality scoring
- TODO Repair suggestions
- TODO Quarantine invalid documents without deleting them
r. Interchange and export
- TODO JSON
- TODO JSON-LD
- TODO RDF
- TODO CSV
- TODO Org
- TODO Markdown
- TODO Archive bundles
- TODO Dataset snapshots
- TODO Incremental exports
- TODO Signed exports
- TODO Portable evidence packages
- TODO Import validation
- TODO Import conflict handling
- TODO Preserve identifiers across systems
- TODO Export redacted variants
- TODO Streaming import
- TODO Streaming export
- TODO External schema mappings
s. Case and investigation containers
- TODO Group documents
- TODO Group entities
- TODO Group events
- TODO Group files
- TODO Group conclusions
- TODO Case-specific access controls
- TODO Notes
- TODO Tasks
- TODO Hypotheses
- TODO Review states
- TODO Evidence numbering
- TODO Chain of custody
- TODO Exportable case bundles
- TODO Case templates
- TODO Cross-case linking
- TODO Prevent automatic cross-case data leakage
- TODO Preserve investigator annotations separately from source data
t. Annotation system
- TODO Human annotations
- TODO Machine annotations
- TODO Span annotations
- TODO Field annotations
- TODO Document annotations
- TODO Entity annotations
- TODO Relationship annotations
- TODO Annotation author
- TODO Annotation timestamp
- TODO Annotation confidence
- TODO Annotation review state
- TODO Competing annotations
- TODO Correction history
- TODO Supersession history
- TODO Annotation schemas
- TODO Convert approved annotations into facts
- TODO Convert approved annotations into training data
u. Reproducibility and snapshots
- TODO Immutable dataset snapshots
- TODO Snapshot manifests
- TODO Snapshot hashes
- TODO Pin schema versions
- TODO Pin actor versions
- TODO Pin model versions
- TODO Pin rule versions
- TODO Pin dependency versions
- TODO Re-run historical queries against historical state
- TODO Compare snapshots
- TODO Record non-reproducible external dependencies
- TODO Generate reproducibility reports
7. Actor Client Library
This will mirror a mini ingress server that routs to actors. handling manfests locally. They will handle the manual registration, by library api the manifest will be made for you, this simply submits it on actors behalf P2P health system
a. Lisp actor library
b. Python Actor Library
c. Nim Based lmdb database based entirly on tek-9 called star-actor cache. which will be used in python, lisp, other langs that have c calling functionality. it will be a full database, but for most use cases will be a cache
8. Improvments to The gserver
a. Review design of gserver actor system and how cl-gserver intends actor systems to be made
k. Review latest cl-gserver code and improvments that can be applied after all bug fixes are made
b. Dead code drop
c. document validation
d. Target Locks
e. normalizing text and inputs
f. Actor Manifest support
g. Dataset Manifest support (the flow based programming backend for directing docuemnts to actors) MUST use the orignal ibm guy idea of flow based programming.
k. database agnostic protocols (actors across languages will use it, via json when sexp not avail)
l. data wearhousing scheama
m. prolog Based graph Search
n. Full p2p system with libp2p
Maybe have to be implemented with nim
o. Fully consider going back to zmq <starRouter> but in common lisp
p. Full StarIntel-Centered DSL
Implement and compare complete prototypes in:
- TODO Common Lisp
Attempt first. Use ordinary Lisp syntax, macros, CLOS, generic functions, and condition handling. Do not use reader macros.
- TODO Racket
Implement as both:
- TODO Scheme
Implement the same DSL using portable Scheme syntax where possible. Test at least one practical implementation such as Chez Scheme, Guile, or Gambit.
- TODO Comparative Research Loop
The research agent must implement the same example workflow in every language.
Each prototype must:
- TODO Use the same document schemas
- TODO Use the same actor behaviors
- TODO Use the same agent and tool definitions
- TODO Use the same dataflow
- TODO Use transient and persistent documents
- TODO Produce inspectable macro expansions or compiled forms
- TODO Include runnable tests
- TODO Record syntax advantages and limitations
- TODO Compare implementation complexity
- TODO Compare debugging and editor support
- TODO Compare runtime performance
- TODO Compare interoperability with existing StarIntel components
- TODO Recommend a language only after all prototypes are complete
Do not select a language based only on appearance.
- TODO Use the same document schemas
- TODO Required Example Workflow
The comparison prototype must implement this complete workflow:
- TODO Receive a
Targetdocument - TODO Match
target.options.enumeration = true - TODO Require
target.datato contain aUser - TODO Extract the user's username
- TODO Append several common email domains
- TODO Emit one transient email-candidate document per domain
- TODO Send each candidate to a testing actor
- TODO Use a fake testing implementation during DSL research
- TODO Submit found candidates to an LLM-native agent
- TODO Allow the agent to call declared tools
- TODO Persist only the final review document
- TODO Express the entire process as a named dataflow
- TODO Receive a
- Minimum DSL Surface
The following names describe required language operations. The exact spelling may vary slightly between implementations, but every prototype must provide equivalent behavior.
- Definition Forms
define-document
Define a typed StarIntel document schema.
define-actor
Define an actor, its accepted documents, matching rules, and behavior.
define-agent
Define an LLM-native actor with a runtime-provided agent loop.
define-tool
Define a typed tool available to an agent.
define-dataflow
Define a named graph connecting actors, agents, sources, and sinks.
- Document Forms
document
Create and validate a document instance.
transient-document
Create a document that may move through a dataflow but must not be persisted.
document-ref
Read a field using a structured path.
Example:
(document-ref target 'data 'username)
document-type-p
Test whether a value is a specific document type.
validate-document
Validate a document against its registered schema.
- Actor Forms
receive
Declare or wait for accepted input.
send
Send a document to a specific actor.
ask
Send a request and receive a reply.
emit
Publish a document into the current dataflow.
find-actor
Find actors by name, capability, accepted document type, dataset, or availability.
actor-ref
Return a stable actor reference.
- Agent Forms
- Dataflow Forms
from
Declare the input source or accepted document type.
through
Route documents through an actor or agent stage.
branch
Route documents conditionally.
filter
Allow only matching documents to continue.
map
Transform one input document into one output document.
flat-map
Transform one input document into multiple output documents.
parallel
Run a stage with bounded concurrency.
merge
Combine multiple branches.
into
Declare a sink.
run-dataflow
Start a named dataflow with an input document.
- Persistence Forms
persist
Save a persistent document.
delete-document
Delete a stored document.
couch-get
Retrieve a CouchDB document by ID.
couch-put
Create or update a CouchDB document.
couch-find
Run a Mango query.
couch-view
Query a CouchDB view.
couch-changes
Consume the CouchDB changes feed.
couch-attach
Add an attachment to a document.
- Required Symbols and Values
persistent
Document may be stored.
transient
Document must not be stored.
actor
A normal actor stage.
agent
An LLM-native actor stage.
source
A dataflow input.
sink
A dataflow output.
capability
A declared actor or tool capability.
document-type
A registered StarIntel document type.
dataset
A registered StarIntel dataset.
found
Candidate was found.
not-found
Candidate was not found.
unknown
Candidate could not be conclusively tested.
- Definition Forms
- Minimum Type System
The DSL must support:
- Minimum Runtime Requirements
- TODO Actor registry
- TODO Capability-based actor discovery
- TODO Typed actor mailboxes
- TODO In-process transport
- TODO RabbitMQ transport
- TODO Dataflow scheduler
- TODO Bounded parallelism
- TODO Backpressure
- TODO Retry policies
- TODO Timeouts
- TODO Cancellation
- TODO Dead-letter handling
- TODO Document provenance
- TODO Dataflow trace IDs
- TODO Structured errors
- TODO CouchDB integration
- TODO LLM provider abstraction
- TODO Tool-call validation
- TODO Deterministic test runtime
- TODO Actor registry
- Required Research Artifacts
For each implementation, create:
- TODO Syntax examples
- TODO Runnable prototype
- TODO Macro-expansion examples
- TODO Runtime architecture notes
- TODO Test results
- TODO Benchmark results
- TODO Interoperability notes
- TODO Advantages
- TODO Disadvantages
- TODO Unresolved problems
The final comparison document must rank:
- TODO Common Lisp
- TODO Racket
- TODO Scheme
The ranking must separately score:
- Syntax
- Extensibility
- Macro system
- Type and schema support
- Actor implementation
- Agent implementation
- Dataflow implementation
- CouchDB integration
- LLM integration
- Debuggability
- Performance
- Deployment
- Existing StarIntel compatibility** Languages
- TODO Syntax examples
q. Possible port to Scheme + Scheme Actors
There seems to be libs for actors on scheam complements the DSL idea
Common Lisp
Nim
Couchdb
Design custom zmq protocol for couchdb
Research how to package FTS server
Use couchdb native fts
A wrapping http service with internal zmq eventing
Optional ZMQ interface
Modify couchdb to accept ZMQ protocol
Implement a couchdb internal graph database
Support Cypher Query
Support
Facial Reconigition
Support Facial Reconigition
- Video: video frames uploaded
- Picture
- Actors if possible
Forensic Liguistic analysis
using llms + prolog expert system fingerprint a style of typing Maybe use ADVANCE, top end maths to do this if thats the cheapest way