Area guide · 28 packages · Follow source qualification and revalidation
Intelligence, research and applications
Manage discovered sources, evidence-backed facts and candidate-authorized application work through explicit lifecycle boundaries.
What are you trying to do?
- How do discovered sources become eligible facts?
- What makes a research result traceable?
- Where does a candidate authorize an application effect?
Why start with these packages
Suggested entry point
@lssm-tech/lib.osint-specStart with the source, policy and evidence contract vocabulary for OSINT.
Suggested entry point
@lssm-tech/lib.osint-runtimeUse source lifecycle, ingestion, qualification and scoped evidence projection logic.
Suggested entry point
@lssm-tech/lib.job-application-runtimeInspect candidate-private lifecycle, headless orchestration and effect-fencing responsibilities.
Suggested entry point
@lssm-tech/lib.opportunity-specInspect the opportunity domain’s declared models and references before composing an opportunity workflow.
Practical steps
Step 01
Inspect a source-backed fact
Follow discovery to qualification and ingestion, then inspect the fact’s source version, evidence locator and permitted purpose. Check how suppression, revalidation and source changes affect later retrieval.
Step 02
Trace an application proposal
Locate the candidate scope, proposed field values and evidence, confirmation step and exact reserved effect. Identify the selected browser/email provider boundary before interpreting a proposal as a submitted application.
Step 03
Review ambiguous completion
Inspect how a timed-out external action is reconciled using its existing receipt or fence. Verify the owning source’s deletion and retention flow separately from ordinary retry behavior; a repeated click should not be assumed to mean a safe repeat effect.
Treat research as a source lifecycle
OSINT runtime separates source discovery, qualification, ingestion and revalidation. Its documented behavior applies scoped policy, idempotent receipts, suppression filtering and evidence-backed assertions before projecting eligible facts into knowledge. Start with the source manifest and the policy that makes the source usable for the selected purpose. A recently discovered URL and an eligible, versioned evidence source are different states; preserve that difference in both the interface and downstream agent context.
Keep evidence and replay attached
Lifecycle mutations are revision-fenced and produce append-only receipts tied to the resulting source manifest. Lease and request bindings let durable workers resume interrupted work and recognize a committed transition. Knowledge projections retain source versions, evidence locators, grant references, purpose and index generation. Query receipts bind returned facts back to evidence and policy decisions. This provides a concrete review path: inspect the assertion, its selected source revision and the authority under which it entered the result set.
Separate application planning from effects
JobApplicationRuntime owns candidate-authorized lifecycle, field proposals, effect fences and replay. Its headless entry provides orchestration without requiring a UI module. The documented default is confirmation for each application; unsupported interactive challenges have no automatic bypass. Browser sessions, email, credentials, persistence and runtime execution remain injected boundaries. The candidate-private contracts and deletion-worker semantics make consent and recovery part of the same lifecycle instead of an afterthought in a connector.
Follow a practical path.
Inspect a source-backed fact
Follow discovery to qualification and ingestion, then inspect the fact’s source version, evidence locator and permitted purpose. Check how suppression, revalidation and source changes affect later retrieval.
Trace an application proposal
Locate the candidate scope, proposed field values and evidence, confirmation step and exact reserved effect. Identify the selected browser/email provider boundary before interpreting a proposal as a submitted application.
Review ambiguous completion
Inspect how a timed-out external action is reconciled using its existing receipt or fence. Verify the owning source’s deletion and retention flow separately from ordinary retry behavior; a repeated click should not be assumed to mean a safe repeat effect.
Packages in this area (28)
28 of 28 shown
- Open manual →
Headless candidate and operator job-application controllers.
- Open manual →
Composable OSINT provider routing and operator view models
- Open manual →
Candidate-bound browser form execution, takeover, and reconciliation.
- Open manual →
Railway S3 and candidate-vault dependencies for managed job application effects.
- Open manual →
Candidate-scoped managed browser effects composition for job applications.
- Open manual →
Opaque Resend inbox aliases, signed receiving webhooks, and verification effects.
- Open manual →
Managed PostgreSQL and AuthOS composition for candidate-private job applications.
- Open manual →
FORCE-RLS Postgres persistence for candidate-private job applications.
- Open manual →
Generated candidate-private job-application transport projections.
- Open manual →
Managed signed Resend receiving composition for candidate-private applications.
- Open manual →
Named France, ATS, and job-board connectors for governed OSINT ingestion.
- Open manual →
Managed PostgreSQL and AuthOS adapters for OSINT services.
- Open manual →
Postgres, object, and vector storage adapters for OSINT v0.
- Open manual →
Generated REST, SDK, MCP, webhook, SSE, WebSocket, and CLI adapters for OSINT
- Open manual →
Pure deterministic projections for founder intelligence, selective sharing, and clarity reviews.
- Open manual →
Provider-neutral contracts for founder-private intelligence, executive projections, and operating archetypes.
- Open manual →
Provider-free candidate-authorized application lifecycle and exact-effect runtime.
Inspect candidate-private lifecycle, headless orchestration and effect-fencing responsibilities.
- Open manual →
Provider-free job-source discovery, acquisition, normalization, closure, and replay runtime.
- Open manual →
Pure deterministic scoring, gate, simulation, and portfolio projections for autonomous-company opportunities.
- Open manual →
Provider-neutral opportunity scoring and selection contracts for autonomous-company candidates.
Inspect the opportunity domain’s declared models and references before composing an opportunity workflow.
- Open manual →
Provider-neutral OSINT lifecycle, ingestion, governance, and knowledge runtime.
Use source lifecycle, ingestion, qualification and scoped evidence projection logic.
- Open manual →
Evidence-first, provider-neutral OSINT v0 schemas and contracts.
Start with the source, policy and evidence contract vocabulary for OSINT.
- Open manual →
Provider-neutral cofounder, hiring, and cobuild contracts.
- Open manual →
Governed, provider-neutral community analytics, coaching views, and Kaizen digests.
- Open manual →
Reusable founder cockpit application services and safe presentation models.
- Open manual →
Headless candidate-authorized job application orchestration.
- Open manual →
Headless orchestration for governed company, job, and people OSINT
- Open manual →
Cofounder, hiring, and cobuild application services over canonical references.
Curated but no longer in the workspace: app.job-application-api, app.job-application-cli, app.job-application-console, app.job-application-mcp, app.job-application-worker, app.osint-api, app.osint-cli, app.osint-console, app.osint-mcp, app.osint-worker
What connects to this area
Identity and access
Actor, candidate, purpose and grant scope are inputs to each effect and read.
Knowledge and retrieval
Searchable knowledge is a derived aid with source/evidence bindings.
Integrations and providers
Web access, browser, email, storage and transport adapters execute selected external work.
Company and work
Company and talent workflows consume eligible research through explicit references.