Bukti
Capability intelligence platform that turns public artifacts into evidence-backed capability claims for people, AI agents, and organizations.
The hosted product — capability profiles, evidence tiers, and natural-language search.
Overview
Bukti is a capability intelligence platform for making skill claims inspectable. It uses public artifacts — repositories, credentials, publications, project write-ups, and other evidence — to produce evidence-backed capability claims about people, AI agents, and organizations.
Bukti has two layers:
What shipped
bukti.ai launched with portfolio builder, evidence-tiered capability scoring, and natural-language search.
Hosted platform plus an Apache-2.0 open-source protocol layer, designed from zero.
The same profile is readable by humans, callable by AI agents over MCP, and queryable via REST.
The MCP and REST surfaces are exposed at bukti.ai, not on lpbangun.com. The structured endpoints on this site (/llms.txt, /resume.json, schema.org Person) are the agent-readable layer for Logani’s own profile, separately maintained.
Shifts evaluation from claimed experience to evidence-backed capability — the core recruiting problem.
My Role
Sole builder. Designed protocol spec, built hosted platform, shipped OSS layer, designed API surface, built agent-readable layer (MCP/REST).
Every claim carries its own confidence
A résumé flattens everything into the same confident voice. Bukti scores each capability by how strongly the evidence backs it — so a reader (or an agent) knows the difference between proven and merely stated.
Verified
Strongest signalBacked by inspectable artifacts — repositories, published work, credentials with a checkable source. The claim points at the proof.
Attested
CorroboratedConfirmed by a third party — a collaborator, manager, or institution vouching for work that is not fully public.
Declared
Self-reportedStated by the person without external evidence yet. Honest about its own uncertainty instead of dressing it up as fact.
One source of truth, three doors
The profile a recruiter reads is the exact profile an AI agent queries. Humans, agents, and systems hit the same capability data through the interface that suits them — no second, divergent copy.
Portfolio site
A readable capability profile a recruiter or collaborator opens in a browser — evidence, provenance, and tiered confidence laid out for a human.
bukti.ai/you Agent-readable surface
Structured endpoints (resume.json, llms.txt, schema.org Person) so an AI agent or sourcing tool can read the profile without scraping the rendered DOM.
lpbangun.com/agents REST API
A typed HTTP surface for hiring tools, dashboards, and integrations to query capability, provenance, and evidence chains directly.
GET /capabilities Product shape
Hosted product
bukti.ai is the closed-source platform and API surface for capability profiles.
Open layer
bukti-oss is the Apache-2.0 Python client, CLI, and open protocol specification.
The open-source layer exposes the contract around capability profiles, provenance, and evidence chains so other clients and agents can integrate with Bukti without depending on a private SDK.
Why verifiable artifacts matter
- Resumes and profiles usually compress work into claims that are hard to inspect.
- Bukti moves the center of gravity from claimed experience to verifiable artifacts: what someone built, where the evidence came from, and how strongly that evidence supports a capability.
- That makes it useful for graduate portfolios, AI-agent evaluation, hiring workflows, and organizations that need a clearer map of who can do what.
Why this matters for hiring
Bukti addresses the core problem in recruiting: resumes and profiles compress work into claims that are hard to verify. By moving the center of gravity from claimed experience to evidence-backed capability, Bukti enables hiring managers to make decisions based on what candidates have actually built — not just what they say they’ve done. The same profile a recruiter reads is the same one an AI agent can query over MCP at bukti.ai.
Public repository
- CodePython client and CLI for the public Bukti API
- ModelsPydantic models for profile, capability, and provenance responses
- SpecOpen protocol specification for Bukti-compatible clients and servers
- DocsDocumentation covering setup, usage, methodology, architecture, and limitations